Datasets:

---

mano-wii

/

blender_duplicates

like 0

[ "Support FFMPEG Autosplit filepaths for animation playback\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 442.19\n\nBroken: version: 2.83.0\n\nIf you enable \"Autosplit Output\" under ffmpeg you cannot view the animation within blender cause \"//\" will be added to the file path.\n\n\n - load up new blender project\n - Set frame range to like 1-20\n - Set output format to ffmpeg, doesn't matter what settings\n - Enable \"Autosplit Output\" just below the container setting\n - In the viewport select \"View/ Viewport Render Animation\"\n - Wait for render to complete then select \"Render/view animation\" (ctrl+F11) \n # note incorrect file path, in my case it reads \"I:\\\\BlenderRenderCache\\\\0001-0020.mp4\"\n\n\n\n", "Render error not shown in VSE\nOperating system: Windows 10\n\nBroken: 2.90 8956e9e5f6c0\nWorked: Unknown\n\n**Reproduction**\n1. In a scene's Compositing workspace, enable Use Nodes and delete all the nodes.\n2. In another scene's VSE, add the above scene as a strip.\n3. Render\n\n**Description**\n\nF12 on the outputless scene throws an error.\n\nThe VSE shows the checkerboard pattern even if there's another strip (e.g. solid black) below the outputless scene strip and does not throw any error. The outputless scene strip should either not interfere with the layers below it or show an error indicating it doesn't have an output.", "VSE: Sound strips without valid source shrink to 0 length on project frame rate selection\nOperating system: Linux-5.4.0-81-generic-x86_64-with-glibc2.31 64 Bits\nGraphics card: GeForce 9800 GT/PCIe/SSE2 NVIDIA Corporation 3.3.0 NVIDIA 340.108\n\nBroken: version: 3.0.0 Alpha\nWorked: Unsure\n\nA sound strip without a valid data source will shrink to 0 length as soon as the project frame rate selection is touched (i.e. changed, or even current frame rate re-selected).\n\n[sound-strip-without-valid-source-shrinks-to-zero-length-on-project-frame-rate-selection.blend](sound-strip-without-valid-source-shrinks-to-zero-length-on-project-frame-rate-selection.blend)\n- Add sound strip\n- Remove source file on disk or change source file name in strip settings to an invalid file name\n- Select any value in project frame rate drop-down\n- Sound strip will \"disappear\" to 0 length", "Geometry Nodes: 3D Viewport Overlay Statistics don't respect Viewer Node\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 471.11\n\nBroken: version: 3.4.0 Alpha\n\n\nThe geometry output by the preview node is not taken into account by the overlay scene statistics:\n[2022-09-30 11-35-43.mp4](2022-09-30_11-35-43.mp4)\n", "Dyntopo detail size drawing with Shift + D seems broken\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce RTX 2080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 460.79\n\nBroken: version: 2.92.0 Alpha\nWorked: At least a few alpha versions ago\n\nDyntopo size control with Shift + D seems broken.\nThe first time I tried, the gizmo simply froze and the Resolution value didn't change after mouse movement.\nAfter installing the latest GeForce Studio Driver — version 457.30 — I tried again. The gizmo casts infinite lines, but I noticed that the Resolution value does change after mouse movement.\n\n1. Enter Sculpt Mode.\n2. Activate Dyntopo.\n3. Switch to Constant Detail mode.\n4. Press Shift + D.", "High Resolution Render support\nSome fields needs to be able to output really high resolution renders out of EEVEE.\n\nFor now, we just try to allocate the asked buffer size and live with the consequence.\n\nMost GPU have a limit of 16K² texture resolution and some really high renders could not fit GPU memory in some cases.\n\nTo workaround this issue, we need to implement a multipass rendering.\n\nThe issue with this approach is that Screen Space effects (GTAO, SSR, SSRefract, SSSubSurfaceScattering) cannot take the whole picture into account.\n\nWe can split the rendering into tiles, scanlines, checkerboard pattern but the artifacts of SS effects will likely to be different for each", "Assigning shortcut to Sequencer data is wrongly 'hooked' the Scene instead of Scene's VSE data\nThis error shows all the time when I right click on the viewscreen, and I have to right click and press ESC, so the message disappears and I keep working this way, right click, a message appears, ESC.\nIt suddenly started to happen and even deleting and downloading blender again, even with the other 2.8 version it happens.\n\nThank you, I love your work guys\nJust trying to help\n\nOperating system: Windows 10 Home, Core i7 - 7500U @ 2.70GHz, 8.00 GB RAM\nGraphics card: AMD Radeon R7 M440 32GB, 4.00 Gbps\n\nBroken: (example: 2.80, edbf15d3c044, master, 2018-11-28, as found on the splash screen)\nWorked: (optional) ", "2022.1.21 30x24k crash after finisx1.6\n[2022.1.21_HK_tree.crash.txt](2022.1.21_HK_tree.crash.txt)\n\n{[F12814049](2022.1.21_blender_debug_output.txt)}Operating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: NVIDIA RTX A6000/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 472.47\n\nBroken: version: 3.1.0 Alpha\n\n超过20K高分辨 仍旧在GPU渲染完成后崩溃\nMore than 20K high resolution，Still crashes after GPU rendering is complete\n但当分辨率降至 20k 以下，比如 19 x 16 k 则不会发生崩溃\nBut when the resolution drops below 20k, such as 1.9 x 1.6k, the crash will not occur.\n\n\n", "VSE: Separate images does not handle resolution correctly\nOperating system: Windows 11 Home 64 bit\nGraphics card: NVidia GeForce 2060 RTX SUPER\n\nBroken: 3.0.1\n\n- Add image sequence to VSE where each image has different resolution\n- Separate images to discrete strips\n\nCheck resolution of separated strips - they are all same, but they should not be.\n\n\n\n\n\n\n\n", "Make switching between custom and generated preview more clear\nOperating system: Linux-5.11.0-7620-generic-x86_64-with-glibc2.33 64 Bits\nGraphics card: AMD NAVY_FLOUNDER (DRM 3.40.0, 5.11.0-7620-generic, LLVM 11.0.1) AMD 4.6 (Core Profile) Mesa 21.0.1\n\nBroken: version: 3.0.0 Release Candidate\nWorked: -\n\nWhen using a thumbnail image from disk, refreshing the thumbnail will not only reload the image from disk, but also downscale its resolution to 128 pixels (or maybe less ?).\n\nIn this example I am using a 256px resolution thumbnail.\n[asset_browser_thumbnail_reload_128px-2021-12-01_10.52.47.mp4](asset_browser_thumbnail_reload_128px-2021-12-01_10.52.47.mp4)\n\nMake asset with 256px thumbnail on disk and press the refresh button.\n\n**Thumbnail used**\n", "VSE editor is 1 pixel off when aligning non-frame images\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce RTX 2070 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.71\n\nBroken: version: 2.93.0\n\nVSE editor is 1px off when compositing with non-full frame images\n\n1) open a VSE empty project\n2) set the scene dimension resolution to 1080x1080\n3) add a 1080x900 picture and check the left side pixels\n\n", "EEVEE-Next Volume Rendering\nUnified volume rendering will first be ported before adding any other feature. Fixing the flickering and object bounds are still major targets for the 4.0 release but are considered nice to have and could be made during Bcon2\n\nHere is a breakdown of the tasks:\n- [x] Needs compute shader based pipeline for world and geometry.\n- [x] #107176 Port current unified volume rendering as-is using compute shaders.\n#### Milestone: Volume Rendering minimum-viable-product for 4.0 release.\n\n- [ ] Correct bounds.\n- [ ] Fix flickering by changing the sampling scheme.\n#### Milestone: Extended feature for 4.0\n\n#### Extended targets\n- [ ] Support shadow casting onto opaque surfaces.\n- [ ] Support irradiance.\n- [ ] Support in bake.\n- [ ] Support per pixel volumes.\n- [x] Optimize by skipping unseen tiles.\n", "VSE: Frame Rate Base Rounding Leads To Vid / Audio Strips Out Of Sync by 1 Frame\nOperating system: Windows 10\nGraphics card: RTX 3090\n\nBroken: since 3.0 until now in 3.6.1 hash 8bda729ef4dc\nWorked: 2.93.18\n\nIt looks like the same bug #21138 is reappearing in the 3.x series and since the bug report is 14 years old, I created a new one.\nOld bug report: 21138\n\nMy description:\nI have a lot of projects with 29.97fps (30000/1001) and those work correctly with Blender 2.9x but since Blender 3.0 all video clips are 1 frame shorter and my projects are a mess, because the frames in all clips are also shifted 1 frame !!\nIf I set the base to 1.0009 instead of 1.001 its all fine in the Timeline, but if I set 29.97 from the drop-down menu, all clips are 1 frame shorter and shifted.\nI have this also tested with videos without any audio and that makes no difference.\nThe audio length btw is alway correct.\nI also noticed that the trick with 1.0009 is not good, because this value is then transferred to the exported video and thats wrong!\n\nI think the problems in the following posts also have the same cause?\n[audio-is-1-frame-off](https://blender.stackexchange.com/questions/51650/audio-is-1-frame-off)\n[blender-3-0-vse-audio-and-video-out-of-sync-bug](https://blender.stackexchange.com/questions/250718/blender-3-0-vse-audio-and-video-out-of-sync-bug)\n\n", "Cutting meta strip including keyframed speed effect produces right part with incorrect length of internal strips\nWindows 8.1 64 bit\n\nBroken: 2.75rc2 (c85a58a)\nWorked: 2.74 I think was fine\n\nWhen I cut a meta strip that includes a speed effect strip that has keyframes right part of the result of cut has all strips cut to width of left part (effectively leaving it with no visible output).\n\n\n- Open blender.\n- Select \"Video editing\" layout.\n- Add -> Movie -> select any movie (I tested with mp4, h264 codec, with audio)\n- Select video strip\n- Add -> Effect -> Speed\n- In \"Effect Strip\" setting uncheck \"Stretch to input strip length\" and add keyframe to \"Speed factor\"\n- Select all strips \n- Make meta strip \n- Set active frame somewhere over this new meta strip.\n- Strip -> Cut (soft) at frame \n\n\nTo make it more interesting if after creating meta strip and then doing unmeta strip and meta strip again it seems to work as I would expect it (just cuts it nicely).", "Python: bpy_prop_array does not support full slicing syntax\nOperating system: Windows 11\nGraphics card: NVIDIA RTX 3060 Ti\n\nBroken: 3.6.0\nWorked: Never\n\n\nI am unable to use the full slicing syntax on a `bpy_prop_array` type.\n\nFor example, I am trying to read the pixels of an image, like so:\n\n```\nimage = bpy.data.images['my_image']\npixels = image.pixels[3::4]\nprint(pixels)\n```\n\nHere the `start` is `3`, the `stop` is `None` and the `step` is 4. This works for regular lists etc., but this code produces the following error:\n\n```\nTraceback (most recent call last):\n File \"<blender_console>\", line 1, in <module>\nTypeError: slice indices must be integers or None or have an __index__ method\n```\n\nSee the attached .blend file.\n\n" ]

[ "VSE proxy fails to build with newer FFmpeg version?\nOperating system: Arch Linux\nGraphics card: GTX 970\n\nBroken: 2.92 (Arch Linux package), but also manual 3.0 alpha (c154b078b5297b53cdc2aa5fbbc3f9d51dcac527) build\nWorked: 2.92 and 2.83.13 official binaries\n\n\nI have a short video clip in Matroska format (see below for details). With the 2.92 and 2.83.13 official binaries creating a VSE proxy for this video works as expected. But with the Arch Linux 2.92 package and a locally built 3.0 alpha version VSE proxy generation fails:\n\n* The proxy_50.avi file is only a few kilobytes and subsequently the preview doesn't use the proxy\n* With 3.0 alpha the errors printed are `Can't send video frame: Invalid argument`, apparently coming from \n\n```\nstatic void add_to_proxy_output_ffmpeg(struct proxy_output_ctx *ctx, AVFrame *frame)\n{\n ....\n\n int ret = avcodec_send_frame(ctx->c, frame);\n if (ret < 0) {\n /* Can't send frame to encoder. This shouldn't happen. */\n fprintf(stderr, \"Can't send video frame: %s\\n\", av_err2str(ret));\n return;\n }\n```\n\nI suspect this is an issue related to the used FFmpeg version. As reported in system-info.txt these differ:\n\n```\nOfficial 2.92 binaries\n\nFFmpeg:\n=====================================\n\navcodec: '58, 54, 100'\navdevice: '58, 8, 100'\navformat: '58, 29, 100'\navutil: '56, 31, 100'\nswscale: ' 5, 5, 100'\n\n\nArch Linux 2.92 binaries, same for manually built 3.0 alpha:\n\nFFmpeg:\n=====================================\n\navcodec: '58, 134, 100'\navdevice: '58, 13, 100'\navformat: '58, 76, 100'\navutil: '56, 70, 100'\nswscale: ' 5, 9, 100'\n```\n\nAre there restrictions on which FFmpeg versions are used when building Blender? My current globally installed ffmpeg is 4.4.\n\n\n* Add movie to VSE as strip, select it\n* Enable strip proxy & timecode\n* Set size to 50%\n* Set timecode index to Record Run\n* Rebuild proxy and Timecode indices\n* Console shows errors\n\nVideo contents:\n```\nffmpeg version n4.4 Copyright (c) 2000-2021 the FFmpeg developers\n built with gcc 10.2.0 (GCC)\n configuration: --prefix=/usr --disable-debug --disable-static --disable-stripping --enable-amf --enable-avisynth --enable-cuda-llvm --enable-lto --enable-fontconfig --enable-gmp --enable-gnutls --enable-gpl --enable-ladspa --enable-libaom --enable-libass --enable-libbluray --enable-libdav1d --enable-libdrm --enable-libfreetype --enable-libfribidi --enable-libgsm --enable-libiec61883 --enable-libjack --enable-libmfx --enable-libmodplug --enable-libmp3lame --enable-libopencore_amrnb --enable-libopencore_amrwb --enable-libopenjpeg --enable-libopus --enable-libpulse --enable-librav1e --enable-librsvg --enable-libsoxr --enable-libspeex --enable-libsrt --enable-libssh --enable-libtheora --enable-libv4l2 --enable-libvidstab --enable-libvmaf --enable-libvorbis --enable-libvpx --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxcb --enable-libxml2 --enable-libxvid --enable-libzimg --enable-nvdec --enable-nvenc --enable-shared --enable-version3\n libavutil 56. 70.100 / 56. 70.100\n libavcodec 58.134.100 / 58.134.100\n libavformat 58. 76.100 / 58. 76.100\n libavdevice 58. 13.100 / 58. 13.100\n libavfilter 7.110.100 / 7.110.100\n libswscale 5. 9.100 / 5. 9.100\n libswresample 3. 9.100 / 3. 9.100\n libpostproc 55. 9.100 / 55. 9.100\nInput #0, matroska,webm, from '2021-05-21_09-20-55.mkv':\n Metadata:\n ENCODER : Lavf58.76.100\n Duration: 00:05:33.85, start: 0.000000, bitrate: 876 kb/s\n Stream #0:0: Video: h264 (High), yuv420p(tv, bt709, progressive), 1920x1080 [SAR 1:1 DAR 16:9], 60 fps, 60 tbr, 1k tbn, 120 tbc (default)\n Metadata:\n DURATION : 00:05:33.850000000\n Stream #0:1: Audio: aac (LC), 48000 Hz, stereo, fltp (default)\n Metadata:\n title : simple_aac_recording\n DURATION : 00:05:33.781000000\n```", "Failure to build movie strip proxy / ffmpeg incompatibility\nOperating system: Linux-5.11.16-arch1-1-x86_64-with-glibc2.33 64 Bits\nGraphics card: NVIDIA GeForce 940M/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 465.27\n\nBroken: version: 2.92.0\nWorked: version 2.92.0\n\nIn VSE, when I want to build the proxy, it pretends that the proxy is being built but when it finishes, the proxy is not actually used. When I run blender from terminal, I can see error messages `Error encoding proxy frame X for '<path>'` where X are frame numbers from 0 up and `<path>` is the path to the proxy file.\n\n**Other important info**\nThe version of blender where this behaviour happens is the version from the Arch Linux community repository. The behaviour reported in this bug does not happen with blender downloaded from the official blender download page. However, there is a different bug reported for the Arch blender package (see 70630?project=5&string=blender) where one of the maintainers of the package says\n\n> Pretty sure this is due to our usage of a very recent ffmpeg as compared to the official blender builds.\n\n\nand further suggests that that issue could be resolved if blender was made (more) compatible with newer ffmpeg. That issue is not related to mine but since it is revolves around ffmpeg, I think it might be very relevant to my issue as well.\n\nI've reported this bug in the Arch package bug tracker as well (see 70657?project=5&string=blender) but I suspect I will get similar answer as for the other bug mentioned above.\n\nSee the attached screen recording: [screenrec-2021-04-30_10.44.59.mkv](screenrec-2021-04-30_10.44.59.mkv)\n\n\nFrom #88459:\n\n\nI have a short video clip in Matroska format (see below for details). With the 2.92 and 2.83.13 official binaries creating a VSE proxy for this video works as expected. But with the Arch Linux 2.92 package and a locally built 3.0 alpha version VSE proxy generation fails:\n\n* The proxy_50.avi file is only a few kilobytes and subsequently the preview doesn't use the proxy\n* With 3.0 alpha the errors printed are `Can't send video frame: Invalid argument`, apparently coming from \n\n```\nstatic void add_to_proxy_output_ffmpeg(struct proxy_output_ctx *ctx, AVFrame *frame)\n{\n ....\n\n int ret = avcodec_send_frame(ctx->c, frame);\n if (ret < 0) {\n /* Can't send frame to encoder. This shouldn't happen. */\n fprintf(stderr, \"Can't send video frame: %s\\n\", av_err2str(ret));\n return;\n }\n```\n\nI suspect this is an issue related to the used FFmpeg version. As reported in system-info.txt these differ:\n\n```\nOfficial 2.92 binaries\n\nFFmpeg:\n=====================================\n\navcodec: '58, 54, 100'\navdevice: '58, 8, 100'\navformat: '58, 29, 100'\navutil: '56, 31, 100'\nswscale: ' 5, 5, 100'\n\n\nArch Linux 2.92 binaries, same for manually built 3.0 alpha:\n\nFFmpeg:\n=====================================\n\navcodec: '58, 134, 100'\navdevice: '58, 13, 100'\navformat: '58, 76, 100'\navutil: '56, 70, 100'\nswscale: ' 5, 9, 100'\n```\n\nAre there restrictions on which FFmpeg versions are used when building Blender? My current globally installed ffmpeg is 4.4.\n\n\n* Add movie to VSE as strip, select it\n* Enable strip proxy & timecode\n* Set size to 50%\n* Set timecode index to Record Run\n* Rebuild proxy and Timecode indices\n* Console shows errors\n\nVideo contents:\n```\nffmpeg version n4.4 Copyright (c) 2000-2021 the FFmpeg developers\n built with gcc 10.2.0 (GCC)\n configuration: --prefix=/usr --disable-debug --disable-static --disable-stripping --enable-amf --enable-avisynth --enable-cuda-llvm --enable-lto --enable-fontconfig --enable-gmp --enable-gnutls --enable-gpl --enable-ladspa --enable-libaom --enable-libass --enable-libbluray --enable-libdav1d --enable-libdrm --enable-libfreetype --enable-libfribidi --enable-libgsm --enable-libiec61883 --enable-libjack --enable-libmfx --enable-libmodplug --enable-libmp3lame --enable-libopencore_amrnb --enable-libopencore_amrwb --enable-libopenjpeg --enable-libopus --enable-libpulse --enable-librav1e --enable-librsvg --enable-libsoxr --enable-libspeex --enable-libsrt --enable-libssh --enable-libtheora --enable-libv4l2 --enable-libvidstab --enable-libvmaf --enable-libvorbis --enable-libvpx --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxcb --enable-libxml2 --enable-libxvid --enable-libzimg --enable-nvdec --enable-nvenc --enable-shared --enable-version3\n libavutil 56. 70.100 / 56. 70.100\n libavcodec 58.134.100 / 58.134.100\n libavformat 58. 76.100 / 58. 76.100\n libavdevice 58. 13.100 / 58. 13.100\n libavfilter 7.110.100 / 7.110.100\n libswscale 5. 9.100 / 5. 9.100\n libswresample 3. 9.100 / 3. 9.100\n libpostproc 55. 9.100 / 55. 9.100\nInput #0, matroska,webm, from '2021-05-21_09-20-55.mkv':\n Metadata:\n ENCODER : Lavf58.76.100\n Duration: 00:05:33.85, start: 0.000000, bitrate: 876 kb/s\n Stream #0:0: Video: h264 (High), yuv420p(tv, bt709, progressive), 1920x1080 [SAR 1:1 DAR 16:9], 60 fps, 60 tbr, 1k tbn, 120 tbc (default)\n Metadata:\n DURATION : 00:05:33.850000000\n Stream #0:1: Audio: aac (LC), 48000 Hz, stereo, fltp (default)\n Metadata:\n title : simple_aac_recording\n DURATION : 00:05:33.781000000\n```" ]

[ "Frame Selected does not respect GN output if it has different type from input\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: Radeon RX 570 Series ATI Technologies Inc. 4.5.0 Core Profile Context 23.3.1.230305\n\nBroken: version: 3.6.0 Alpha\nWorked: n/a\n\nI think this is an old known issue but I can't find this documented here.\n\nFrame Selected does not respect the output geometry by Geometry Nodes if its type is different from the input type.\n\nFor example, if the input is mesh and the output is curve, Frame Selected works on the original input mesh. (The same if the output has multiple types of geometry including the original type.) If the output is mesh it respects the output geometry even if it has noting to do with the input and has a wildly different bounding box.\n\nOn the other hand, a curve object converted to a mesh by Solidify modifier is still respected as the final output mesh.\n\n(Maybe related? #104738)\n\n[FrameSelectedWithGN.mp4](attachment)\n\n[FrameSelectedWithGN.blend](attachment)\n\n", "Applying geometry nodes modifier create an empty material slot for no reason\nOperating system: W10\n\n3.4.1\n\nApplying geometry nodes modifier create an extra empty material slot for no reason.\n\n| Geometry node tree | Applyed mesh |\n| -- | -- |\n|  |  |\n\n1. Create an object\n2. Add a GN modifier\n3. Create some geometry inside GN and assign a material to it (or don't)\n4. Apply the modifier, you'll get an empty material slot on the object.\n\n", "Realtime compositor: Texture node output textures with artifacts\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2060 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 531.61\n\nBroken: version: 3.6.0 Alpha\nWorked: n/a\n\nTexture node support was introduced in 151a53110c\n\nIn realtime compositor, Texture node outputs textures with artifacts.\n\n\n\nIt happens with most procedural textures I tested. This does not happen when a custom image is used, or when Blend texture is connected from the Value output (instead of the Color output) of the node. (With other textures it happens with both outputs.)\n\nThis does not happen when running Blender with a single thread (`blender -t 1`)\n\nFile for quick test: [TextureNodeTest.blend](attachment)\n\n", "Output from Geonodes Edge Domain to Shader behaves strangely\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: Quadro K3100M/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 426.78\n\nBroken: version: 3.1.0 Beta\nWorked: Never, as far as I know\n\nWhen retrieving data stored on the edge domain in a shader, the values are inconsistent between adjacent faces, and generally do not interpolate across faces as expected.\n\n\n\n\nOpen attached file and switch to rendered mode.\n\n[Edge Domain Issue Demo.blend](Edge_Domain_Issue_Demo.blend)\n", "Nodes: Changing Input position with a Driver\nBroken: 3.1 and 3.2 Alpha\n\nIn Geometry Nodes editor, shift an input socket above a socket having a Driver and it will switch the Driver.\n[2022-03-12 15-58-02.mp4](2022-03-12_15-58-02.mp4)\n\n- Enter Group Node\n- Shift \"Vertices Y\" input above\n[Test.blend](Test.blend)", "Outliner Floating window Drag & Drop to Geometry Nodes editor not working\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 536.23\n\nBroken: version: 3.6.0\nWorked: I don't know if it ever worked\n\nDragging and dropping objects from the outliner to the geometry node editor is not working\n\nFloat outliner to be its own window\ndrag and drop an object from the node editor to the geometry nodes\ndoesn't work.\n\n", "Breaking nodes API changes for 4.0\n- [x] Remove `Node.width_hidden`\n- [ ] Complete \"Spline\" to \"Curve\"/\"Curves\" rename in geometry nodes\n- [ ] Stop storing node locations relative to their parents\n- [ ] Use a separate type for internal links (in RNA too) (see [D16960](D16960))\n- [ ] Use a separate type for node group interface inputs and outputs and node sockets\n- [ ] Use dynamically allocated `char *` array to store string socket default values", "Annonymous Attributes and Simulation Nodes - propagation issue\nBroken: version: 4.0.0 Alpha\nWorked: Never.\n\nSome annonymous attributes are not working with the output of the Simulation Nodes if the attribute is not propagate via the simulation.\n\n\n1. [simulation_attribute_bugs.blend](attachment).\n2. Toggle the \"Show Bug\" modifier property to see when the bug happens.\n\n", "Links persist invisibly after their source sockets are removed\nBroken: Probably forever (e.g. 3.0, 3.1)\nWorked: N/A\n\nIf the output socket of a node is removed, the input socket it was connected to does not show a button to input a value. It acts like it's still connected.\n\n**Detailed replication**\n1. Go into the Shader Editor\n2. Create the *Voronoi Texture* node and a *Math* node\n3. Set the *Voronoi Texture* node to 4D, and plug its new W output into the *Math* node\n4. Set the *Voronoi Texture* node to 3D. Observe that the input on the *Math* node still acts like it's connected\n5. Click and drag from that input on the *Math* node. A bogus wire will appear from the *Voronoi Texture* node, coming from where the W output used to be.\n\nYou can then plug this wire into another socket. It will disappear again once you let go. If you set the *Voronoi Texture* node back to 4D, the wires will re-appear.\n\nVideo example:\n\n[XwBMq60aMI.mp4](XwBMq60aMI.mp4)\n\n**Another example:**\n\n{[F13079601](free-socket.blend), size=full}\n\nIn this case you can even save the file in this state where you can't see or edit the value of the node until you disconnect it.\n\n", "Frame Selected does not respect Geometry Nodes viewer geometry\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: Radeon RX 570 Series ATI Technologies Inc. 4.5.0 Core Profile Context 22.11.2.221130\n\nBroken: version: 3.5.0 Alpha\nWorked: n/a\n\nNot sure if this is a bug to report or feature request but a minor inconvenience with the Viewer node.\n\nWhen Geometry Nodes Viewer node is active, Frame Selected works on the (invisible) output geometry instead of the Viewer geometry. Since the Viewer geometry is actually selectable, I think it makes sense that Frame Selected to respect it.\n\n[FrameSelectedViewerNode.mp4](attachment)\n\n", "A combination of shader nodes causes lack of material emission at any brightness (Cycles)\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 528.49\n\nBroken: version: 3.4.0(tested as well on 3.3.0 and 3.5.0 alpha)\nWorked: (none)\n\nA difference of bevel node and normal coordinates produces no light at any emission level. Clipping the result due to suspicion about negative brightness has no result.\nNote: adding any value to the result adds corresponding emission, but despite visuals, brightness from combination above does not.\n\n\n- Open attached file\n[cycles_where_is_emission__bug.blend](attachment)\n\n\n", "Crazy-space support for meshes in geometry nodes\nIn [D15407](D15407) crazy-space support for curves is added to geometry nodes. Currently, meshes use a different system for crazy-space that is not really compatible with nodes as is.\n\nIt should be possible to use the approach implemented for curves for meshes as well, but that needs to be investigated further.", "Simulation zone pauses randomly during render\nThis bug has an easy workaround, mostly posting this so anyone running into the same issue knows the solution.\n\nOperating system: Linux-5.15.0-79-generic-x86_64-with-glibc2.35 64 Bits, X11 UI\nGraphics card: NVIDIA GeForce RTX 3080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 530.30.02\n\nBroken: version: 3.6.2\nWorked: N/A, 3.6 is the first version with simulation nodes.\n\nIn Eevee, Cycles, and Workbench, rendering an object with a simple simulation zone applied in geometry nodes results in some rendered frames being duplicates of each other, effectively making the simulation pause at random in the final render.\n\nThe workaround is to bake the simulation zone result before rendering (see screenshot.)\n\nCheck out the attached .blend file for reproduction.\n\n", "Displacement Animation works only with keyframed node present\nOperating system: Linux-5.4.0-107-generic-x86_64-with-glibc2.27 64 Bits\nGraphics card: NVIDIA GeForce RTX 2070 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 510.60.02\n\nBroken: version: 3.1.2\nBroken: version: 3.2.0 Alpha\nWorked: unkown\n\nDisplacement Animation works only with a keyframed node present, even if it is not connected!\nI was trying to animate displacement using the object color, finding that it doesn't work.\nWhen trying a keyframed mapping node instead I found that it works but even if the node is not connected!\n\n - Open the attached blender file, switch to render mode and play the animation\n - The displacement will be animated.\n - Now delete the not connected mapping node or clear its keyframes.\n - The displacement will not be animated anymore. Even though I would expect it to be animated by the object color.\n\n[displace-test-1-1.blend](displace-test-1-1.blend) \n\n[displace-test-1-1.mp4](displace-test-1-1.mp4)\n", "Geometry Nodes: Mismatching data types connections work differently from other node editors\nOperating system: win10\nGraphics card: 1050ti\n\nBroken: 3.5.0\n\nFloat math node overrides color data, even though it is muted. Same setup works in shader editor without problems.\n\nOpen file and activate *Viewer* in GN. Expected result: colored output (judging by shader area behavior)\n\n" ]

[ "Geometry Nodes: Simulation nodes not work with sampling nodes if value is single\nBroken: Last main.\nWorked: Never.\n\nIf pass result of any sampling nodes (RayCast, Sample Index, Sample Nearest, ...) via simulation output node, result will be none.\n\n| Constant value input | Sampled value from geometry |\n| -- | -- |\n|  |  |\n\nFrom scratch:\n1. In Geometry Nodes tree add simulation nodes pair.\n2. Delete geometry socket from simulation node settings.\n3. Add Mesh Primitive → Capture Attribute → Sample Index → Simulation output.\n4. See different values in tooltips.\n\nOr open attached example file." ]

[ "VSE: issue moving strips with changed speed factor\nOperating system: Linux-5.18.10-76051810-generic-x86_64-with-glibc2.34 64 Bits\nGraphics card: AMD Radeon Vega Frontier Edition (VEGA10, DRM 3.46.0, 5.18.10-76051810-generic, LLVM 12.0.1) AMD 4.6 (Core Profile) Mesa 21.3.5\n\nBroken: version: 3.5.0 Alpha\n\nIn the Video Sequence Editor, when moving strips that have the speed factor different than default 1.0 the strips seem to glitch so the start and end of each strip fluctuates.\n\nNew file\nGo to VSE\nAdd a movie strip\nUse \"K\" to cut the movie strip a few times (so you have maybe 10-15 strips)\nSelect a strip, go to the Time panel and change the speed factor to 1.5\nOne by one, do the same with all the strips (perhaps giving them various values between 1.1 and 1.5)\nMove each strip up against each other, so there's no dead space between\nSelect all of them and move them around...\n{[F14115112](vse_speedup_movement_glitch.mp4)}you will see the length of each strip fluctuates\n\n", "Missing Freestyle Lines\nOperating system: Linux-5.4.0-91-generic-x86_64-with-glibc2.31 64 Bits\nGraphics card: AMD Radeon(TM) Vega 10 Graphics (RAVEN, DRM 3.35.0, 5.4.0-91-generic, LLVM 12.0.0) AMD 4.6 (Core Profile) Mesa 21.0.3\n\nBroken: 3.0.0, 3.2.0\nWorked: 2.83\n\nThere are missing Freestyle Lines when Rendering with Cycles. The results are better in eevee but the Problem is not gone. \nI use a similar file to create floorplans at work. \n\n- Open attached file or:\n - Create an orthographic camera,\n - create a huge amount of planes and text objects,\n - enable Freestyle, \n- render via Eevee or Cycles.\n[fs.blend](fs.blend)\n", "VSE placement of some elements is off by 1 px\nOperating system: Linux-5.4.0-81-generic-x86_64-with-glibc2.31 64 Bits\nGraphics card: GeForce GTX 1070 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 460.91.03\n\nBroken: version: 3.0.0 Alpha\n\n[blender vse placement bug.blend](blender_vse_placement_bug.blend)\n\nplace some elements smaller than the video canvas in vse by pixel and there will be errors. probably because the elements are smaller than full size.\n\nattached file:\nvse canvas size: 1000x1000px\nadded 500x500px image and placed it on -250;250\nduplicated the image to have 4 of them and tried to fill the whole canvas with the images by pixel perfect placement.\non my machine there are gaps appearing between the elements on the bottom right one, maybe there are different gaps on different machines, I had different gaps than these before.\neven setting it to 249 on one axis does not solve this but introduces unwanted gaps at the bottom. and it seems there are gaps between the bottom left image and the canvas bottom too.\n\nmy idea for a solution would be to start placing elements from to top left or bottom left and not from the center as the 0 in the center makes it weird and unintuitive, a 0 on x and y in a corner makes this a bit easier. That would lead to the need of anchor points on elements I think, elements should have an anchor on a corner (top left or the same as the canvas zero point as default) to make placing them also easier, switching corner for the anchor point could be an option as well as setting a value for the anchor point could be an option. rotation could also have the option to switch between center and anchor point rotation, same for scaling.\nAND anchor points would make it probably easier to implement object snapping, that would be really useful but has nothing to do with this bug report :)\n\n", "Weird VSE crash\nOperating system: Win11\nGraphics card: RX580\n\nBroken: (example: 2.80, edbf15d3c044, master, 2018-11-28, as found on the splash screen) blender 4.0.0 alpha\nWorked: (newest version of Blender that worked as expected) no clue\n\nBlender crashes on opening file. Was trying to make an alternative caption that gets around the Multiply color issue #112267 but came across a new bug.\n\nBased on the default startup or an attached .blend file (as simple as possible).\nOpen the file I've attached\n\n", "Camera background image is hidden by GPencil fill with alpha\nOperating system: Linux-4.14.78-gentoo-x86_64-AMD_FX-tm-8350_Eight-Core_Processor-with-gentoo-2.6 64 Bits\nGraphics card: GeForce GT 730/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 410.73\n\n\nBroken: version: 2.80 (sub 75)\nWorked: (optional)\n\n\nWhen I change the fill alpha while a background image is loaded, it behaves weirdly. i.e, when the alpha value is 0, the background image is visible, when alpha crosses 0.001, the image mysteriously disappears.\n\nIn Grease Pencil, I added a background image in the camera object data context menu,\nthen I drew a box and changed the alpha of fill, then the above said bug appears.\n\nI have included the blend file and a video to produce the bug.\n\n[blenderGreasePencilBugVid.mp4](blenderGreasePencilBugVid.mp4)\n\n[greasePencilBug.blend](greasePencilBug.blend)\n\n", "Empty Images react with Viewport World Opacity\nOperating system: Windows-7-6.1.7601-SP1 64 Bits\nGraphics card: GeForce GTX 980/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 461.40\n\nBroken: version: 2.83.8 through 2.93 master.\nWorked: version: 2.82a\n\nThere is a problem in EEVEE where Empty Images Depth set to \"Back\" will also draw them behind the world HDRI background. So the Image will fade into the background when changing the Viewport Shadings World Opacity.\n\n(From the image you can see that it only reacts with values below 1.0. Opacity maxed out will make Empty images non transparent again. That's where I started to think this isn't fully intended to work like that, because consequential the value 1.0 would go full transparent.)\n\nMy issue here is that working with images shouldn't be relevant to this setting at all (as its shown for value 1.0).\nThe current behavior discourages to use world opacity in order to keep a clear image or if I want to use the color picker in the image for example.\n\n[EmptyImageOpacityBug.blend](EmptyImageOpacityBug.blend)\n\n1. Create an Empty, set to Image display.\n2. Set its Depth Setting to \"Back\"\n3. Change the Viewport Shading > World Opacity\n\n", "Black Artefacts in EEVEE on Transparent BSDF when using Alpha Hashed for Blend mode and Camera Depth Of Field\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce RTX 2080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.71\n\nBroken: version: 2.92.0\n\nBlack Artefacts in EEVEE on Transparent BSDF when using Alpha Hashed for Blend mode and Camera Depth Of Field, if Shader Color is not pure white (similar problem with Layer Weight)\n\n**Exact steps for others to reproduce the error**\n- Open attached file (Just a simple box (can be a plane) with Transparent BSDF. Camera' distance or depth of focus influences the result. Attached pictures in Image Editor from a real case).\n- Press F12 to render\n[transprarent_shader_artifacts_with_camera_DOF1.blend](transprarent_shader_artifacts_with_camera_DOF1.blend)\n\n\n", "Black artifacts on smooth surface in 2.79 64bit daily build\nOperating system: Windows 10\nGraphics card: RTX 2080Ti\n\nBroken: blender-2.79.0-git.d7cecc2ba3f6-windows64\nWorked: 2.79b\n\n\nThe software renders black artifacts on a smooth surface. This happens both in the rendered preview and in the final render. Artifacts disappear when smooth is not applied.\n\n\n\n\n\n\n\nJust go in rendered preview mode in the editor, or just F12 to see the artifacts in the render.\n\n[bug report.7z](bug_report.7z)", "Modifier sub-panel drawn over selection outline\nOperating system: Windows-10-10.0.18363-SP0 64 Bits\nGraphics card: GeForce GTX 1080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.71\n\nBroken: version: 3.1.0 Alpha\n\nThis is just a visual bug and nothing serious, but I wanted to point out that the sub-panel background is drawn on top of the selected modifier outline, which causes it to look broken up when expanded. This is visible but not obvious in the default theme, but very obvious in custom themes where the sub-background alpha is set to 1. \n\n\n\n1. Add a Bevel or Subdiv modifier\n2. Expand a sub-panel \n3. Set the properties editor Sub Background theme color to an alpha of 1 \n\n", "Graph Editor: Curve drawing improvements\nProblem: keyframe dots can be somewhat hard to see, especially behind the current frame marker line.\n\n\n\n**Proposal**:\n- Change the default theme to draw the keyframe dots larger. The current theme draws at 6px, for modern displays this should probably be increased to 7-9px. It's probably better to defer a concrete choice for this until #68982 (Graph Editor: UI Scale effect on Curves drawing) is done.\n- Draw keyframe dots of selected keyframes on top of current frame line. Keep drawing unselected keyframe dots behind the line.\n", "VSE editor is 1 pixel off when aligning non-frame images\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce RTX 2070 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.71\n\nBroken: version: 2.93.0\n\nVSE editor is 1px off when compositing with non-full frame images\n\n1) open a VSE empty project\n2) set the scene dimension resolution to 1080x1080\n3) add a 1080x900 picture and check the left side pixels\n\n", "Texture paint: Bleedind around thin cuts do not fill texture\nOperating system: Windows-10-10.0.17763-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 512.95\n\nBroken: version: 3.6.0 Alpha\nWorked: never?\n\n\n1. Wider distance allow to overlap color\n2. closer distance between faces do not\n3. closer distance between islands allow to overflow to neighbour island…\n\n\n[bleeding test.blend](attachment)\nopen file, try to paint as seen in the [video 2023-03-24_05-09-32.mp4](attachment)\n\n", "Color space seems wrong while painting in image editor.\nIn the image editor add a new image and with default settings just do a red line paint over a green line paint.\nDark zone appear in the falloff between red and green. it should not be there and should stay clear, not dark. \nI don't know why, but it looks like a color space issue. \n\ndark falloff not ok in blender:\n\n\nclear falloff ok in gimp:\n\n\nblender 2.8\nWindows 7 \n\n\n", "Line Art isn't updating active camera\nOperating system: GNU/Linux mint 19\nGraphics card: gforce RTX 2060\n\nBroken: 2.93.4 \n\nLine Art isn't updating active camera when it's changed in a 'Scene' strip (VSE)\n\n\n- Create a scene, a camera and a LineArt object.\n- Now, create a 2nd camera (Cam.001) (not the active one)\n- Create an empty scene in order to be used for video creation.\n- In `Video Sequencer`, add a Scene strip and choose the Cam.001 camera ->it changes ok, but lineart goes dependent on the first one.\n\nTest File:\n[#91971.blend](T91971.blend)", "Diffuse Direct, Indirect Bake to Vertex Color Artefacts\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.94\n\nBroken: version: 3.5.0\nWorked: This has never worked properly (At least in my experience)\n\nWhen baking Diffuse light to vertex color, black triangles and random dark spots appear in vertex color. This effect is more apparent with:\n- Area lights.\n- Emissive surfaces.\n- Lights with very high power.\n\n1. Create geometry and give each a byte_color vertex color channel.\n2. Place area lights in scene with high power.\n3. Bake diffuse direct and indirect to color attribute.\n\nSee .blend of my project attached. In the file i've already baked the emissive light to vertex color where you can see the artefacts. Switch to cycles live render view to see the correct output. I think the tiled texture may have something to do with it. I've added a tiled texture and packed it in the example.\n\n" ]

[ "Video Editing (outline on the border)\nOperating system: Windows-10 64 Bits\nGraphics card: GeForce GTX 750 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 460.89\n\nBroken: version: 2.93.0 Alpha\n\nSome images with an alpha channel have grey outline when scaled up and different bicubic or bilinear interpolation is used.\n\nIssue is visible immediately after opening .blend file\n[#86861.zip](T86861.zip)" ]

[ "UV: (investigate) Render UVs of selected object in UV Editor while in Object Mode\nCurrently, the UV Editor only works while in \"Edit\" mode.\n\nHow firm is that restriction?\n\nIt would be nice if we could view UVs in the UV editor, even in \"Object\" mode.\n\nIf would be even nicer if we could use the basic edit tools, such as move, scale and rotate on those UVS, again, even in Object mode.\n\nUVs can be generated and adjusted procedurally, but there's no good way to see them. It would be nice to allow displaying (not editing) evaluated UVs. Combined with features like visualizing overlapping faces, this could save a lot of time.\n\nThis is an *investigate* task, to see if such a thing is even possible. If we discover it's possible, we'll then need to decide if it's desirable.\n\nThanks @BlenderBob for the inspiration!\n\n", "Keymap: use number keys for mode switching (2nd Iteration)\n## Motivation\n\n* Improve ease of mode switching with direct shortcuts (`Tab` and `Ctrl+Tab` functionality remains the same)\n* Consistent accross all object types.\n* Allows the use of pie menus for quickly going into sub-modes (and later adding more options like Island Select mode for meshes)\n* Having the `~` key (or equivalent left to 1) be the Switch Object (`Transfer Mode`) fits better with mode switching than D.\n* The current functionality of toggling Collections visibility often leads to confusion for new users:\n * Objects seemingly disappear\n * Risk of losing visibility settings or undo steps.\n\n## Proposed Solution\nThis task is to evaluate the possibility of using number keys for mode switching, in addition to the already existing `Tab` and `Ctrl+Tab` shortcuts to toggle Edit, Pose, and modes pie menu.\n\nThe current proposal:\n{[F10130043](image.png), size=full}\n\n\n* The numbers row will get a dedicated option in the Keymap Preferences:\n * Collection Visibility (default at least during the trial phase)\n * Mode Switching\n * Potentially `Emulate Numpad` could be part of this option as well.\n\n* Behavior for **Tab** and **Ctrl-Tab** for mode edit-mode switching will be kept as this is convenient and used in many other areas where numbers make less sense (graph-editor, NLA, sequencer... etc, where using numbers doesn't make so much sense).\n\n## Details\n\n- Modes would be mapped to a key consistently between object types, so the pose-mode number-key would do nothing for a curve (for example).\n- Equivalent modes would share a key between object types (grease-pencil Draw to use the same key as mesh Texture Paint, the same goes for Grease Pencil Sculpt, Vertex Paint... etc).\n- Modes to switch include (8): `OBJECT` `EDIT`, `SCULPT`, `TEXTURE_PAINT`, `VERTEX_PAINT`, `WEIGHT_PAINT`, `PARTICLE_EDIT`, `POSE`.\n\n\n## Open Topics\n\n**Should sub-modes have direct access?**\nWith the introduction of pie menus by default in 2.80, users are now used to these and pie menus could be used to access sub-modes. \n\nIt has been proposed in #84380 that vertex/edge/face be expanded into the number buttons too. This only works well for mesh and grease pencil edit-mode, however it has some down-sides.\n\n- We would be using all number keys by default, any additional modes or sub-modes wont fit.\n- Some sub-modes wont fit:\n - Particle edit (currently uses keys 1-3 to switch sub-modes).\n - Weight paint.\n - Texture/Vertex paint.\n\nThe ideal solution in this case isn't clear, either we accept loosing shortcuts for switching sub-modes or accept some inconsistency (having to add a different shortcut for sub-mode switching for modes besides edit-mode).\n\nThere are plans to merge painting modes into a single attribute painting mode, even if this is done, it might make sense to expose weight/texture/sculpt as sub-modes... either way, I don't think we should rely on this change unless it's committed to master before the keymap changes. The same applies to the future of Particle Edit once the new Hair object type is implemented.\n\n**What to do when pressing a modes key a second time?**\n\nCurrently nothing is planned. Options are to toggle the mode or cycle through sub-modes. Both have their pro's and con's so there are no plans to use second-time presses.\n\n**Is overwriting edit-mesh modes acceptable?**\n\nBy switching directly into vert/face/edge modes, this will overwrite the mixed modes which the user will need to manually set every time they enter edit-mode. Holding Shift while changing selection modes will extend the selection as it does currently.\nEven though this isn't such a common functionality, it would be good to avoid breaking peoples workflows.\n\n----\n\n*Updated by @pablovazquez based on notes from the UI meeting on 20 May.*", "Custom Property list not updating in UI when editing properties from Python\nOperating system: Arch Linux\nGraphics card: GTX 970\n\nBroken: 2.80.60 (1c5860407068)\n\n\nSetting a custom property from Python does not update the corresponding entry under Custom Properties until mouse over forces a refresh.\n\n\nUsing default blend file.\n\n1. Add a Python Console panel somewhere\n2. Select the cube object\n3. Switch to the object properties tab, unfold the Custom Properties panel\n4. In the python console:\n```\no = C.active_object\no['value'] = 456\n```\n5. No new entry is shown in the Custom Properties panel\n6. Move the mouse to the Custom Properties panel -> new value prop appears\n7. The same thing happens when updating a property value, a mouse over is also needed to refresh the property list being displayed\n", "Outliner: Sync Selection off not working when clicking LMB on data object name\nBroken: version: 3.0.0\nBroken: version: 2.91.2 (works correctly on meshes for some reason till worked version)\nWorked: version: 2.82\n\nWhen `Sync Selection` is off it's still possible to sync selection by clicking on object data name (+ small margin). Additionaly clicking on object data name will sync it with relevant slot in Properties editor. This bugged area becomes slightly longer to the left when active, see video.\n\n1. Open default scene.\n2. Turn off `Sync Selection` on Outliner.\n3. Deselect onjects in your scene by clicking on empty space in 3D Viewport.\n4. Select objects or object data in outliner without clicking on their name - see that `Sync Selection` is off\n5. Select objects or object data in outliner by clicking in their name - see that now parent onject is selected in Outliner aswell as in 3D Viewport.\n[Outliner Sync Selection off bug 2021-12-15 20-45-35-213.mp4](Outliner_Sync_Selection_off_bug_2021-12-15_20-45-35-213.mp4)", "Unable to get object to stay hidden while new mesh tool is active.\nWindows 10, 1060\n\nBroken: 2.80 master (also reproduced in a month-old daily build, so not new behavior).\n\nIf you have an active tool like Add Cylinder going, and you hide another object using the eye icon in the outliner, then as soon as you touch the cylinder parameters, the hidden object becomes unhidden.\n\n1) From the default scene, Add a Cylinder to the scene and expand the tool properties in the lower left of the 3dview.\n2) Because the cylinder is hidden inside the cube, we can't easily see the effect of changing the tool parameters, so go to the outliner and click the eyeball icon associated with the cube to hide it in the viewport.\n3) Change any parameter in the cylinder tool properties pop-up in the 3dview. The cube reappears and the eyeball indicates it is no longer hidden.\n\nBasically any changes made to the Outliner visibility icons gets reverted as soon as you change any active tool property, (at least for the Cylinder tool). This at least unexpected behavior.", "Needs two clicks with Ctrl or Shift key to deselect an item from a list of selected items\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 457.30\n\nBroken: version: 2.92.0\n\nNeeds two clicks with Ctrl or Shift key to deselect an item from a list of selected items.\n\n\n* Open the project file “OutlinerSelectItemWithCtrl.blend” in attachment as a starting point.\n\nThe same problem happens in different views with different details.\n\nOutliner:\n* Select the entire collection of objects from the collection “Item List”. You can click the “Cone” object, hold the Shift key and then click the “Torus” object to select the entire list.\n* Then hold the Ctrl key and click the Cylinder object to deselect.\n* The item’s color get a lighter color, but it is still not excluded from the whole selection.\n* Clicking a second time on the object really deselects the object.\n* If clicking the first item of the selection (which is the Cone is this example), it is deselected properly.\n\n3D View:\nThe Shift key is used to select/deselect objects individually. It is pretty much the same issue. It requires two clicks on the same object to really deselect the object, whatever the order of the object in the selection. However, if one object has already been selected before deselecting everything, then the same object gets selected with an additional highlight from the new selection (dragging the mouse cursor in the viewport).\n\nShading view:\nPretty similar to the 3D view.\nIf I deselect all the nodes (Alt + A for example) and I drag the mouse cursor to select all nodes, the last selected node is highlighted in white. If I click the node highlighted in white while holding the Shift key, the node is deselected. However, if I do the same thing for the other nodes, it selects the node with a white highlight and I have to click a second time on the same node to really deselect it. The behavior of the last selected item is also present in the 3D View, but not for the Outliner.\n\nThe Compositing view has the same problem and I suspect that the Geometry Node view has the same too (I haven’t tested the last one).\n\nIn general, it requires two clicks to deselect an object with the normal behavior.\n\nThe first click behaves as an “add to selection” click unless the object already had a selection before completely creating a new selection.\n\nIn all cases, there is an issue with deselecting (or excluding) and single object from a selection of objects in different views. As a reference how it should behave correctly, the Windows Explorer in Windows is able to deselect a single item on the first time it is clicked.\n\n[OutlinerSelectItemWithCtrl.blend](OutlinerSelectItemWithCtrl.blend)\n\n[Deselect_Issue.mp4](Deselect_Issue.mp4)", "Object properties In-front doesn't work in viewport shading\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1060 with Max-Q Design/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 445.87\n\nBroken: version: 2.82 (sub 7)\nWorked: Never\n\nObject properties In-front doesn't work in viewport shading\nAll the other options works in Object properties/viewport display, except for In-front \n\n\n- Open attached file\n- Go to Viewport Shading\n[lookdev_in_front_bug.blend](lookdev_in_front_bug.blend)\n", "Library overrides of custom properties not preserved when override made in Python\nOperating system: Windows 10 Pro (10.0.19044)\nGraphics card: NVIDIA GeForce RTX 3080\n\nBroken: 3.5 through to current 3.6 daily (`a9f02cd8d8cf`)\n\n\nCustom properties added to objects via add-ons are not preserved when the Python API is used to make the override libraries and set the property values.\n\n\n1. Install `testprop.py` as an add-on.\n2. Load `src.blend` and observe that the cube `my_source_cube` has a location (0, 0, 0) and the custom property added by `testprops.py` has a value of \"Foo\".\n3. Load `dest.blend` and observe that the object `my_source_cube` has been linked from `src.blend`, and that it has no library overrides.\n4. Execute the following code, either in a text script or the console:\n\n```\nimport bpy\n\nobj = bpy.data.objects['my_source_cube']\nbpy.context.view_layer.objects.active = obj\nbpy.ops.object.make_override_library()\n\nbpy.context.view_layer.objects.active.location.z = 2.0\nbpy.context.view_layer.objects.active.TestData.my_enum = 'bar'\n```\n\nObserve that the location in the panel is color-coded as overridden, but the `EnumProperty` is not, despite having changed in value.\n\n5. Save the file. Observe that the EnumProperty has been reset to \"Foo\".\n\nIt's possible that properties added with add-ons are not meant to be overriden in the same way, but considering that right-clicking the property, selecting _Define Override_, and modifying the property works as expected, I would expect this to work in the same manner.\n\nIt's also possible that the Python calls I'm making are not the correct ones (I've seen reference to `override_create`, `override_hierarchy_create`, etc) but `make_override_library` works for the location and there's no indication it wouldn't work for add-on properties.\n\n", "Particle System switching between Emitter/Hair \"Viewport Display\" sticks to Hair-Viewport-Display-Settings\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce RTX 2080 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 452.06\n\nBroken: version: 2.92.0 Alpha\n\nIn the particle system, when you change the particle type form Emitter to Hair and then back to Emitter , the \"Viewport Display\" panel does not change back accordingly - it remains as if the Hair-type emitter is still present.\n\nit would be just as easy to create a new particle system with the right settings, to remedy this situation\n\n- create a particle system on a object\n- Observe the properties of the \"Viewport Display\" panel\n- change the particle type form Emitter to Hair (properties have changed accordingly)\n- change it back to emitter\nViewport Display won't show Options Particles have, instead it will show Hair Options like \"Strand Steps\"\n\n", "Pinning Bones/Constraint Tabs in Properties Editor doesn't work\nBroken: 2.76 - master (1283c66)\nWorked: Unknown\n\nThe context pinning functionality in the Properties Editor doesn't work for Bones and Constraints. There may be a few others too that I haven't noticed yet, but these are the biggest ones.\n", "Outliner selection not working correctly after undoing object deletion\nOperating system: Linux-5.17.15-1-MANJARO-x86_64-with-glibc2.35 64 Bits\nGraphics card: AMD VERDE (LLVM 13.0.1, DRM 2.50, 5.17.15-1-MANJARO) AMD 4.5 (Core Profile) Mesa 22.1.2\n\nBroken: 2.91.2, 3.2.0, version: 3.3.0 Alpha\nWorked: prior to b077de086e\n\nAn object that is linked in another collection can usually be removed from just that collection by using unlink, while using delete removes it from both collections. But if you delete the object and undo, unlink does delete the object in both collections. \nRenaming the cube or creating a new object \"breaks\" the cubes apart so they can be selected/unlinked individually again.\n\nIn the default scene,\n- create a new collection\n- in the outliner, ctrl-drag the default cube object to that collection to link it\n- delete the cube\n- undo\n- in the outliner, it is now impossible to select the cube in the new collection separately and using unlink will delete both cubes", "Cloth physics cache gets reset by changing object materials\nOperating system: Darwin-19.4.0-x86_64-i386-64bit 64 Bits\nGraphics card: AMD Radeon Pro 5500M OpenGL Engine ATI Technologies Inc. 4.1 ATI-3.8.24\n\nBroken: version: 2.83 (sub 11)\nWorked: unknown\n\nWhen using a cloth physics simulation, if you add or remove a material from an object, the cloth system cache is reset (\"cache is outdated\"). I don't think the material affects the cloth physics in any way (as it seems you can change any parameter within the material without the cache resetting), and therefore this should probably not happen. Please correct me if I am wrong and the material does affect the physics somehow.\n\nIn the default scene, enable cloth physics for the cube. Let the playback run for a few frames, building up the cache. Add a material to the cube; now the cache gets reset.\n\n", "Show Emitter toggle still hides object after Particle System has been deleted.\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 1050/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 451.67\n\nBroken: version: \n- 2.92.0 Alpha\n- 2.92.0 Alpha\nWorked: version: \n- 2.92.0 Alpha\n- 2.91.0 Release\n\nWhen you delete a particle system with show emitter unchecked under Viewport Display the object stays hidden. You have to re-add the particle system and check Show Emitter.\n\n- Add Particle System to default cube.\n- Uncheck Show Emitter under Viewport Display\n- Delete Particle System. \n\n", "Properties don't update focus when copy/paste an object\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 511.79\n\nBroken: version: 3.0.1, 3.1.2\nWorked: I don't know\n\nIf you **Copy/Paste** object with **Ctrl+C**, **Ctrl+V** the Property window will still have focus on the old object (although a new one is selected).\nIn this case, the **Duplicate** operation works correctly, the focus is on the new object (the one that is selected).\nReproduce on **Scene objects** and on **Nodes** in node editor. I think there are many more places.\n\n[filename700.mp4](filename700.mp4)\n[filename704.mp4](filename704.mp4)\n\nIn node editor\n1) Select any node.\n2) Copy and paste node with Ctrl+C/Ctrl+V.\n3) Change any properties, you can see that you are changing the previous object, although a new one is selected.\n\n", "Wrong property changing `matrix_world` and destructive undo via RMB.\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1650/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 536.23\n\nBroken:\nversion: 3.6.1\nversion: 4.0.0 Alpha\nWorked: never\n\nReturning to the original state during editing (or canceling) does not give the result that was when editing was started.\n<video src=\"attachment\" title=\"2023-08-16 21-40-53.mp4\" controls></video>\n\n1. Copy script and run\n```python\nimport bpy\n\nclass Panel(bpy.types.Panel):\n bl_idname = 'Panel_PT_Panel'\n bl_label = \"Test Panel\"\n bl_space_type = 'TEXT_EDITOR'\n bl_region_type = 'UI'\n bl_category = 'Text'\n def draw(self, context):\n col = self.layout.column()\n ob = context.object\n col.prop(ob,'name')\n col.prop(ob,'matrix_world')\n\nif __name__==\"__main__\":\n bpy.utils.register_class(Panel)\n```\n2. Quickly change the scale value for the object matrix.\n3. Press Right mouse button.\n\n" ]

[ "Outliner/Properties syncing\nFrom inside the Outliner, it would be useful to be able to sync with the Properties editor, so that it would reflect whatever data you have selected there:\n\n\n*In this example clicking on the Material data would open the Material Properties.*\n\nHowever, this has other implications:\n\n - Currently, selecting obData (mesh, curve etc) switches the mode to Edit mode. But, this would conflict with Outliner-Properties syncing.\n - If we use selecting for switching Properties contexts, mode switching by clicking on data must be removed. In my estimation, making selection change the mode is not what you would expect anyway. Instead, switching modes could be done by holding a modifier key or via the Context Menu.\n\nThis can work as follows:\n\n - Selecting any data in the Outliner will make the parent object active, then switch to the correct tab depending on the data type\n - The mode doesn't change when you select odData (mesh, curve etc)" ]

[ "Blender crashes when merging mesh.vertices with a python script\nOperating system: Pop_OS 20.04\nGraphics card: Intel 4th Generation Core Proc \n\n\nBroken: 2.82a\n\n\nLink to stackoverflow question:\n\nblender-crashes-when-merging-mesh-vertices-with-a-python-script\n\nI have reuploaded the question here because I think It could be an error with blender code rather than mine. The steps to reproduce the crash would be creating an icosphere mesh, setting the max_distance variable to 0.3 and just moving nodes nearby to other ones and clicking run_script then. At some point it will close and save an ico.crash.txt. I have attached mine.\n\n[ico.crash.txt](ico.crash.txt)\n\n[#88022.blend](T88022.blend)\n\n[#88022.webm](T88022.webm)", "Laucnh error since version 3.4\nOperating system: High Sierra 10.13.6\nGraphics card: ATI 6970M 2 GB RAM\n\nBroken: (example: 3.4 and 3.4.1 2022-12-20, master)\nWorked: (Blender 3.3.1)\n\nWhen I launch blender a get this error that I have found others on the web reporting also on similar setups:\nError opening window.\nFailed to create metal device for offscreen GHOST Context.\n \n\nPrevious versions worked a charm and I really hope 3.3.1 wont be the last version for my humble setup.\n\nThanks and as always appreciate your amazing work!", "Crash after linked object is edited then switched to another scene (object origin scene)\nOperating system: Window10\nGraphics card:\n\n\nBroken: 2.90.1 ( hash: 3e85bb34d0d7 )\nWorked: unknown\n\n\nBlender simply crash and close the window directly after editing the \"linked\" object from 2nd screen and switched back to 1st screen.\n\n\n\nThis error can be reproduced with 2 scenes setup.\n\n- First scene contains a simple mesh object. Let's say a \"Cube\" object\n- Create a second scene.\n- In the first scene, select the \"Cube\" object then \"Make Links\" {key Ctrl L} to 2nd scene.\n- Now, we switch to 2nd scene. The 2nd scene now should contains the same \"linked\" Cube object from 1st scene.\n- Okay, this is important step. In this 2nd scene, we select this \"linked\" Cube object and set the Relations to: *Object > Relations > Make Single User > Object*\n- After that, we go to \"Edit Mode\" for this \"linked\" Cube object.\n- In \"Edit mode\", make a simple change to this \"linked\" Cube object.\n- Once simple change is done, don't do anything else. You just directly switch back to \"First scene\". CRASH here.\n\n", "Simulation Nodes: bake UI in simulation node\n_No response_\n\n", "Eevee: Unknown bug when just create first shader \nStep to reproduce an issue (or bug?):\n1. Run latest Blender v3.4alpha (Experimental build) with Eevee Next integrated.\n2. Create scene with just one Cube.\n3. Go to Properties > Material panel > Create New material.\n4. Boom! Blender shut down imidetily.", "Missing overlays in the viewport when viewing through transparent shader\nOperating system: Ubuntu 20.04\nGraphics card: Geforce MX230\n\nBroken: 3.6.1\n\nwhen viewing an LGT-shadow_caster.004 object, all overlays disappear from a certain side of it\n\n\nSee LGT-shadow_caster.004 in blender 3.6 splash screen\n\n", "Rendering animation using \"Generate\" type modifiers after animated displacement texture crashes Blender\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 1650/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 460.79\n\nBroken: version: 2.92.0 Alpha\nWorked: 2.83.10\n\nIf 2 high poly (~100k faces per obj) objects in modifiers properties have after Displacement modifier (with animated texture) \"Generate\" type modifiers (I've tried Boolean and Decimate), Blender cant run Animation render (but still can run Image render). \nCrash happens on 2.9 and higher. I've tested this on 2.83.10 LTS (from steam) and everything worked fine there.\nCrashes with Decimate a little bit hard to recreate. So I will go with steps for \"Boolean crash\"\n\n- create Cude, Empty and Icosphere\n- Add Subdivision mod. to Cube: set it to \"Simple\" and 7 levels\n- Add Displacement mod. to Cube and set texture (Clouds) coordinates to object - Empty\n- Add Boolean mod. (difference) and select object Icosphere\n- Select Cude with Icosphere and duplicate them\n- Save and press Render Animation\n[bug.blend](bug.blend)\n[bug.crash.txt](bug.crash.txt)\n\n```lines\n## 3883==ERROR: AddressSanitizer: heap-use-after-free on address 0x61500034ec08 at pc 0x000016fc795d bp 0x7fffc19dd6c0 sp 0x7fffc19dd6b0\nREAD of size 8 at 0x61500034ec08 thread #15\n - 0 0x16fc795c in blender::gpu::GLBatch::verts_(int) const /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/opengl/gl_batch.hh:112\n - 1 0x17013501 in blender::gpu::GLVertArray::update_bindings(unsigned int, GPUBatch const*, blender::gpu::ShaderInterface const*, int) /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/opengl/gl_vertex_array.cc:124\n - 2 0x16fa900b in blender::gpu::GLVaoCache::vao_get(GPUBatch*) /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/opengl/gl_batch.cc:273\n - 3 0x16fa84d8 in blender::gpu::GLVaoCache::base_instance_vao_get(GPUBatch*, int) /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/opengl/gl_batch.cc:234\n - 4 0x16fa9d3e in blender::gpu::GLBatch::bind(int) /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/opengl/gl_batch.cc:317\n - 5 0x16fa9f3f in blender::gpu::GLBatch::draw(int, int, int, int) /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/opengl/gl_batch.cc:328\n - 6 0x16eca2be in GPU_batch_draw_advanced /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/intern/gpu_batch.cc:282\n - 7 0x16fc1207 in blender::gpu::GLDrawList::append(GPUBatch*, int, int) /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/opengl/gl_drawlist.cc:122\n - 8 0x16efa9f5 in GPU_draw_list_append /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/intern/gpu_drawlist.cc:52\n - 9 0x5978acc in draw_indirect_call /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/intern/draw_manager_exec.c:598\n - 10 0x5978acc in draw_call_batching_flush /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/intern/draw_manager_exec.c:795\n - 11 0x597c5f0 in draw_call_batching_finish /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/intern/draw_manager_exec.c:891\n - 12 0x597eae7 in draw_shgroup /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/intern/draw_manager_exec.c:1056\n - 13 0x597f89f in drw_draw_pass_ex /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/intern/draw_manager_exec.c:1113\n - 14 0x597fe50 in DRW_draw_pass /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/intern/draw_manager_exec.c:1153\n - 15 0x5c51bc0 in EEVEE_velocity_resolve /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/engines/eevee/eevee_effects.c:466\n - 16 0x5c5202b in EEVEE_draw_effects /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/engines/eevee/eevee_effects.c:486\n - 17 0x5a21f42 in EEVEE_render_draw /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/engines/eevee/eevee_render.c:671\n - 18 0x59b1dbc in eevee_render_to_image /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/engines/eevee/eevee_engine.c:559\n - 19 0x5948820 in DRW_render_to_image /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/intern/draw_manager.c:1909\n - 20 0x14ff1d77 in engine_render_view_layer /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/engine.c:786\n - 21 0x14ff40a3 in RE_engine_render /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/engine.c:914\n - 22 0x1501aafb in do_render_3d /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/pipeline.c:1132\n - 23 0x1501c24a in do_render /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/pipeline.c:1209\n - 24 0x1501e036 in do_render_composite /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/pipeline.c:1349\n - 25 0x15022789 in do_render_all_options /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/pipeline.c:1615\n - 26 0x15027839 in RE_RenderFrame /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/pipeline.c:2015\n - 27 0xd7b42cf in render_startjob /home/dev/01-data/01-git/blender-git/blender/source/blender/editors/render/render_internal.c:630\n - 28 0x4fed4f1 in do_job_thread /home/dev/01-data/01-git/blender-git/blender/source/blender/windowmanager/intern/wm_jobs.c:389\n - 29 0x18de408d in tslot_thread_start /home/dev/01-data/01-git/blender-git/blender/source/blender/blenlib/intern/threads.cc:231\n - 30 0x7ffff7566608 in start_thread /build/glibc-ZN95T4/glibc-2.31/nptl/pthread_create.c:477\n - 31 0x7ffff6f46292 in __clone (/lib/x86_64-linux-gnu/libc.so.6+0x122292)\n\n0x61500034ec08 is located 8 bytes inside of 512-byte region [0x61500034ec00,0x61500034ee00)\nfreed by thread #15 here:\n - 0 0x7ffff76927cf in __interceptor_free (/lib/x86_64-linux-gnu/libasan.so.5+0x10d7cf)\n - 1 0x18e0eab0 in MEM_lockfree_freeN /home/dev/01-data/01-git/blender-git/blender/intern/guardedalloc/intern/mallocn_lockfree_impl.c:129\n - 2 0x16ef56c6 in blender::gpu::GLVertBuf::operator delete(void*) /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/opengl/gl_vertex_buffer.hh:57\n - 3 0x17018021 in blender::gpu::GLVertBuf::~GLVertBuf() /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/opengl/gl_vertex_buffer.hh:35\n - 4 0x16f8bb19 in blender::gpu::VertBuf::reference_remove() /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/intern/gpu_vertex_buffer_private.hh:94\n - 5 0x16f877f5 in GPU_vertbuf_discard /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/intern/gpu_vertex_buffer.cc:161\n - 6 0x5a0eab8 in EEVEE_motion_blur_cache_finish /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/engines/eevee/eevee_motion_blur.c:449\n - 7 0x59b1bef in eevee_render_to_image /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/engines/eevee/eevee_engine.c:542\n - 8 0x5948820 in DRW_render_to_image /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/intern/draw_manager.c:1909\n - 9 0x14ff1d77 in engine_render_view_layer /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/engine.c:786\n - 10 0x14ff40a3 in RE_engine_render /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/engine.c:914\n - 11 0x1501aafb in do_render_3d /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/pipeline.c:1132\n - 12 0x1501c24a in do_render /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/pipeline.c:1209\n - 13 0x1501e036 in do_render_composite /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/pipeline.c:1349\n - 14 0x15022789 in do_render_all_options /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/pipeline.c:1615\n - 15 0x15027839 in RE_RenderFrame /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/pipeline.c:2015\n - 16 0xd7b42cf in render_startjob /home/dev/01-data/01-git/blender-git/blender/source/blender/editors/render/render_internal.c:630\n - 17 0x4fed4f1 in do_job_thread /home/dev/01-data/01-git/blender-git/blender/source/blender/windowmanager/intern/wm_jobs.c:389\n - 18 0x18de408d in tslot_thread_start /home/dev/01-data/01-git/blender-git/blender/source/blender/blenlib/intern/threads.cc:231\n - 19 0x7ffff7566608 in start_thread /build/glibc-ZN95T4/glibc-2.31/nptl/pthread_create.c:477\n\npreviously allocated by thread #15 here:\n - 0 0x7ffff7692bc8 in malloc (/lib/x86_64-linux-gnu/libasan.so.5+0x10dbc8)\n - 1 0x18e0f510 in MEM_lockfree_mallocN /home/dev/01-data/01-git/blender-git/blender/intern/guardedalloc/intern/mallocn_lockfree_impl.c:276\n - 2 0x16ef5676 in blender::gpu::GLVertBuf::operator new(unsigned long) /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/opengl/gl_vertex_buffer.hh:57\n - 3 0x16ef70a2 in blender::gpu::GLBackend::vertbuf_alloc() /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/opengl/gl_backend.hh:121\n - 4 0x16f85d10 in blender::gpu::VertBuf::duplicate() /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/intern/gpu_vertex_buffer.cc:77\n - 5 0x16f8750d in GPU_vertbuf_duplicate /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/intern/gpu_vertex_buffer.cc:149\n - 6 0x5a0ee06 in EEVEE_motion_blur_cache_finish /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/engines/eevee/eevee_motion_blur.c:467\n - 7 0x59b1953 in eevee_render_to_image /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/engines/eevee/eevee_engine.c:507\n - 8 0x5948820 in DRW_render_to_image /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/intern/draw_manager.c:1909\n - 9 0x14ff1d77 in engine_render_view_layer /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/engine.c:786\n - 10 0x14ff40a3 in RE_engine_render /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/engine.c:914\n - 11 0x1501aafb in do_render_3d /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/pipeline.c:1132\n - 12 0x1501c24a in do_render /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/pipeline.c:1209\n - 13 0x1501e036 in do_render_composite /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/pipeline.c:1349\n - 14 0x15022789 in do_render_all_options /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/pipeline.c:1615\n - 15 0x15027839 in RE_RenderFrame /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/pipeline.c:2015\n - 16 0xd7b42cf in render_startjob /home/dev/01-data/01-git/blender-git/blender/source/blender/editors/render/render_internal.c:630\n - 17 0x4fed4f1 in do_job_thread /home/dev/01-data/01-git/blender-git/blender/source/blender/windowmanager/intern/wm_jobs.c:389\n - 18 0x18de408d in tslot_thread_start /home/dev/01-data/01-git/blender-git/blender/source/blender/blenlib/intern/threads.cc:231\n - 19 0x7ffff7566608 in start_thread /build/glibc-ZN95T4/glibc-2.31/nptl/pthread_create.c:477\n\nThread #15 created by T0 here:\n - 0 0x7ffff75bf805 in pthread_create (/lib/x86_64-linux-gnu/libasan.so.5+0x3a805)\n - 1 0x18de42b9 in BLI_threadpool_insert /home/dev/01-data/01-git/blender-git/blender/source/blender/blenlib/intern/threads.cc:245\n - 2 0x4fee516 in WM_jobs_start /home/dev/01-data/01-git/blender-git/blender/source/blender/windowmanager/intern/wm_jobs.c:484\n - 3 0xd7b88bd in screen_render_invoke /home/dev/01-data/01-git/blender-git/blender/source/blender/editors/render/render_internal.c:1041\n - 4 0x4f8f3ba in wm_operator_invoke /home/dev/01-data/01-git/blender-git/blender/source/blender/windowmanager/intern/wm_event_system.c:1293\n - 5 0x4f912f2 in wm_operator_call_internal /home/dev/01-data/01-git/blender-git/blender/source/blender/windowmanager/intern/wm_event_system.c:1500\n - 6 0x4f91551 in WM_operator_name_call_ptr /home/dev/01-data/01-git/blender-git/blender/source/blender/windowmanager/intern/wm_event_system.c:1548\n - 7 0x90f1c95 in ui_apply_but_funcs_after /home/dev/01-data/01-git/blender-git/blender/source/blender/editors/interface/interface_handlers.c:932\n - 8 0x917b8d0 in ui_handler_region_menu /home/dev/01-data/01-git/blender-git/blender/source/blender/editors/interface/interface_handlers.c:10809\n - 9 0x4f88a9b in wm_handler_ui_call /home/dev/01-data/01-git/blender-git/blender/source/blender/windowmanager/intern/wm_event_system.c:636\n - 10 0x4f9ec36 in wm_handlers_do_intern /home/dev/01-data/01-git/blender-git/blender/source/blender/windowmanager/intern/wm_event_system.c:2757\n - 11 0x4f9fc78 in wm_handlers_do /home/dev/01-data/01-git/blender-git/blender/source/blender/windowmanager/intern/wm_event_system.c:2865\n - 12 0x4fa537f in wm_event_do_handlers /home/dev/01-data/01-git/blender-git/blender/source/blender/windowmanager/intern/wm_event_system.c:3288\n - 13 0x4f732cf in WM_main /home/dev/01-data/01-git/blender-git/blender/source/blender/windowmanager/intern/wm.c:635\n - 14 0x34f58fa in main /home/dev/01-data/01-git/blender-git/blender/source/creator/creator.c:522\n - 15 0x7ffff6e4b0b2 in __libc_start_main (/lib/x86_64-linux-gnu/libc.so.6+0x270b2)\n\nSUMMARY: AddressSanitizer: heap-use-after-free /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/opengl/gl_batch.hh:112 in blender::gpu::GLBatch::verts_(int) const\nShadow bytes around the buggy address:\n 0x0c2a80061d30: fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd\n 0x0c2a80061d40: fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd\n 0x0c2a80061d50: fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd\n 0x0c2a80061d60: fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd fa\n 0x0c2a80061d70: fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa\n# >0x0c2a80061d80: fd[fd]fd fd fd fd fd fd fd fd fd fd fd fd fd fd\n 0x0c2a80061d90: fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd\n 0x0c2a80061da0: fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd\n 0x0c2a80061db0: fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd\n 0x0c2a80061dc0: fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa\n 0x0c2a80061dd0: fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd\nShadow byte legend (one shadow byte represents 8 application bytes):\n```\nAddressable: 00\nPartially addressable: 01 02 03 04 05 06 07 \nHeap left redzone: fa\nFreed heap region: fd\nStack left redzone: f1\nStack mid redzone: f2\nStack right redzone: f3\nStack after return: f5\nStack use after scope: f8\nGlobal redzone: f9\nGlobal init order: f6\nPoisoned by user: f7\nContainer overflow: fc\nArray cookie: ac\nIntra object redzone: bb\nASan internal: fe\nLeft alloca redzone: ca\nRight alloca redzone: cb\nShadow gap: cc\n```\n", "Subdivision Surface Performance Degrades Markedly With Non-Quad Meshes\nOperating system: Windows 7 Pro X64\nGraphics card: Nvidia Geforce GTX 980 Ti\n\n\nBroken: Date: 2019-01-21\n```\n Hash: dd3f5186260e\n Branch: Blender 2.7 (although it's supposed to be a 2.80 Beta)\n```\n\n[OpenSubDiv_PerformanceBug.zip](OpenSubDiv_PerformanceBug.zip)\n\nSubdivision surfaces are fast and responsive in Edit mode *with all-quad meshes*, but very slow (~ 10x slower) with meshes comprised of triangles and quads.\n\n\nOpen the attached blend file and select **Cube.001**\n\nGo into Edit mode. Turn on Proportional Editing. Grab a point and move it - observe how responsive the viewport is with the all-quad mesh.\n\nGo back to Object Mode. Select **Cube**. Go back into Edit Mode. Grab a point and move it (with proportional editing still turned on) Observe how slow it is.\n\n", "use-after-free crash after setting scene's render.engine inside custom render engine context\nOperating system: Ubuntu 20.04.2LTS\n\nBroken: 2.93.0 Beta, branch: master, commit date: 2021-05-19 10:14, hash: a294670bacb1, type: release\nAlso Broken: 0efb627bbd3ac3e53afb95591ae4e244bd1f8af4\n\nBlender crashes when setting some scene's render.engine property inside `__init__` or the `view_update` context of a custom render engine to a different one than the custom engine.\n\n[T88404_asan_report.txt](T88404_asan_report.txt)\n**Code to reproduce the error**\n```\nimport bpy\nimport bgl\n\nclass CustomRenderEngine(bpy.types.RenderEngine):\n bl_idname = \"CUSTOM\"\n bl_label = \"Custom\"\n bl_use_preview = True\n \n def __init__(self):\n return\n scene = bpy.data.scenes.new(\"foo\")\n scene.render.engine = \"CYCLES\"\n bpy.data.scenes.remove(scene)\n\n def __del__(self):\n pass\n\n def view_update(self, context, depsgraph):\n scene = bpy.data.scenes.new(\"foo\")\n scene.render.engine = \"CYCLES\"\n bpy.data.scenes.remove(scene)\n\n def view_draw(self, context, depsgraph):\n pass\n\ndef get_panels():\n exclude_panels = {\n 'VIEWLAYER_PT_filter',\n 'VIEWLAYER_PT_layer_passes',\n }\n\n panels = []\n for panel in bpy.types.Panel.__subclasses__():\n if hasattr(panel, 'COMPAT_ENGINES') and 'BLENDER_RENDER' in panel.COMPAT_ENGINES:\n if panel.__name__ not in exclude_panels:\n panels.append(panel)\n\n return panels\n\n\ndef register():\n bpy.utils.register_class(CustomRenderEngine)\n\n for panel in get_panels():\n panel.COMPAT_ENGINES.add('CUSTOM')\n\n\ndef unregister():\n bpy.utils.unregister_class(CustomRenderEngine)\n\n for panel in get_panels():\n if 'CUSTOM' in panel.COMPAT_ENGINES:\n panel.COMPAT_ENGINES.remove('CUSTOM')\n\nif __name__ == \"__main__\":\n register()\n```\n\n1) Copy code from above into a new script file inside of Blender\n2) Run the script\n3) Select \"Custom\" as Render Engine inside Render Properties\n4) Select \"Material Preview\" as display method in the viewport.\n\n[crash.blend](crash.blend)\n\nCrash log:\n```\n# Blender 2.93.0, Commit date: 2021-05-19 10:14, Hash a294670bacb1\nEngine 'CUSTOM' not available for scene 'Scene' (an add-on may need to be installed or enabled) # Error\nbpy.ops.text.run_script() # Operator\n\n# backtrace\n./blender(BLI_system_backtrace+0x20) [0xa63c950]\n./blender() [0xf6082a]\n/lib/x86_64-linux-gnu/libc.so.6(+0x46210) [0x7f7ce17be210]\n./blender(GPU_framebuffer_blit+0x2f) [0x8664faf]\n./blender() [0x1485fc3]\n./blender() [0x14597fe]\n./blender(DRW_draw_render_loop_ex+0x242) [0x1459a72]\n./blender(view3d_main_region_draw+0x8f) [0x1d3570f]\n./blender(ED_region_do_draw+0x851) [0x189c771]\n./blender(wm_draw_update+0x52e) [0x130cc9e]\n./blender(WM_main+0x30) [0x130aa20]\n./blender(main+0x319) [0xe881c9]\n/lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xf3) [0x7f7ce179f0b3]\n./blender() [0xf5d1bc]\n\n# Python backtrace\n```", "Crash on calling bpy.ops.mesh.loopcut in python script\nOperation system: macOS BigSur 11.6 (20G165) \nApple M1\n\nVersion 3.4.1 (3.4.1 2022-12-20)\n\nBlender crashes in background mode when you are calling bpy.ops.mesh.loopcut with overridden context\n\nRun Blender from cli with args:\n/path/to/bin --background --python bug.py\n[bug.py](bug.py)\n\n", "Selection Next\nBecause of design decision, the Overlay-Next engine will not be able to rely on the draw manager to handle selection queries. We need to handle selection of objects in a separate engine.\n\nThe selection engine must not re-implement all of the logic of the overlay engine. Instead, we need to find a way to cleanly mix both. For instance, we can swap the `ShaderCache` type to use shaders for selection.\n\nIn order to not require too much refactor, it is better to interface this new selection engine as a GPU_select backend.\n\nAll object selection mode should be supported.", "Crash when rendering a specific 4k scene with Cycles Metal\nOperating system: macOS-13.3.1-arm64-arm-64bit 64 Bits\nGraphics card: Apple M2 Pro Apple 4.1 Metal - 83.1\n\nBroken: version: 3.5.1\n\nWhen I render with Blender 3.5.1 on a MacBook 16\" with an M2 Pro chip at a 4K quality, Blender closes before finishing the rendering process.\"\n\n- Open attached .blend file\n- Make sure you have enabled Metal GPU+CPU in the Preferences\n- Render the scene\n\nIt should crash about halfway through the rendering process.\n\n\n", "Scene update bug\nSystem Information\nwindows7 64\nASUS ATI Radeon HD6870\n\nBlender Version\nBroken: (All)\nWorked: (None)\n\n\n\n\n\n\n\n\nopen this file, select plane and try to select for this plane any mesh - all this meshes don't have users, but was saved, try to select material - materials was saved, but don't have users, \nwe can revert, close and open again many times - materails and meshes will be saved.\ntry to add gruop instance - we will have all this groups is empty.", "GPU Compute is not available\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: GeForce GTX 660/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.71\n\nBroken: version: 3.6.2\n\nGPU Compute is not available in render device untill recheck\n\n\nAn interesting circumstance was discovered:\nIf open `preferences/System/CUDA` and just uncheck and check the checkbox of GPU Device, the problem is solved, then GPU Compute is accessible and fully functional.\n\nhad a similar problem: 94699\nThe solution is supposedly installing a newer version of nvidia driver. However, in fact, Blender works fine with the existing version.\n\n", "Blender will not use new GPU for rendering or recognize it for OpenCL or\nOperating system: Windows 10, all up to date\nGraphics card: msi 4070ti (brand new)\n\nBroken: 2.93.18 as well as the most recent 3.5 on the splash screen, tried most of the in-between builds on clean reinstalls but same issue on all of them.\nWorked: no version\n\n**Error**\nAfter getting a new 4070ti, blender started crashing every 5-10mins or so while moving around a simple scene in the viewport set to material shaders mode, not render mode. The scene consisted of a couple of primitives with very simplistic wood, glass, and metal shaders. When I tried rendering as is it also crashes immediately upon rendering no matter what I use. GPU wasn't greyed out in render settings initially (though it now is), but only showed in preferences under CUDA, and wasn't recognized in OptiX or OpenCL (in newer blender models it also wasn't recognized by anything but CUDA). \n\n1. tried to render a simple scene with render settings set to new gpu\n2. render crashes immediately\n3. move around in viewport on material mode, barely move and blender crashes\n4. check Blender preferences <system <cycles rendering devices: gpu is visible in CUDA but not in any other options (originally was working on 2.93 where I wanted to use OpenCL, I know this isn't in the newer versions with OptiX/all non-CUDA options). Try unchecking CPU hoping to force blender to actually use the GPU and have tax manager up to see if gpu is even being used\n5. blender crashes without rendering, no sign of gpu being used on task manager but CPU usage spikes to 100%\n\n**What I've tried so far that might be helpful**\n\n- tried clean reinstalling different versions of blender, both newer/experimental and older/more stable builds\n- made sure I reset BIOS in case this is a new component/migration error\n- made sure the settings for blender in NVIDIA control panel were all compatible or high performance \n- made sure NVIDIA drivers are all studio version and up to date\n- made sure blender cache folder didn't have any kernel errors\n- installed any updates to windows, this did make the gpu visible in OptiX but unfortuantely the same issue persists even with OptiX on any version where blender just won't use it and uses only cpu (despite being unchecked)\n- double checking how much memory I have on my new gpu vs the old one - yup it does\n- benchmarked my gpu using port royal and making sure I could render a project in Premiere. GPU did fine on the benchmark test and was able to render a 20min video fine. Only blender seems to be the problem\n- noticed today in stackexchange there are a lot more reports with others having the same issue, even with equivalent AMD gpus, and all comments said this seems to be a bug\n\n" ]

[ "Blender crashes when I enable 'use opensubdiv' in subsurf modifier.\nWindows XP/SP3 32bits\n\nBroken: version 2.75 (sub 4), branch b'master', commit date b'2015-08-24' b'22:51', hash b'a7dd209', b'Release'\nWorked: Unknown\n\nBlender crashes when I enable use opensubdiv in subsurf modifier.\nRun Blender.\nOpen attached file.\nTurn on 'use opensubdiv' option.\n[CrashOpenSubdiv.blend](CrashOpenSubdiv.blend)", "OSD Crash\nWindows 8.0\nIntel HD 4600\n\n[System Info](system-info.txt)\n\nBroken: Latest master\nWorked: (optional)\n\nOSD Crash With this error\n\n\n```\nError linking GLSL program : Vertex shader does not write to gl_Position.\n\nDefines: #define FLAT_SHADING\n\n```\n\nIn default scene add subdiv modifier and enable osd" ]

[ "Mesh tool to 'unsplit' while preserving custom normals\nSome exporter generate split geometry to preserve 'sharp edges' (like Blender used to do not so long ago). However, now that we support custom normals, it would be nice to have e.g. an object-mode 'unsplit' modifier to remove doubles while preserving custom normals.\n\nSee #47540 for a usecase.", "Spin Duplicate Tool gizmo uses the wrong operation\nOperating system: Linux-5.15.0-27-generic-x86_64-with-glibc2.35 64 Bits\nGraphics card: Quadro GP100/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 510.60.02\n\nBroken: version: 3.3.0 Alpha\n\nThe gizmo of the Spin Duplicate tool in the toolbar uses executes the wrong operation. Instead of using the Spin Duplicate operation it will use the Spin operation.\n\nTo get the correct behaviour, try changing the \"Drag\"setting in the header tool settings to \"Active Tool\".\nThen click and drag anywhere.\n\n----\n\nI also want to include the proposal to just merge the Spin and Spin Duplicate tools and instead add an option in the tool settings called \"Use Duplicates\".\nBoth tools use the same operator anyway with the only difference that this one option is toggled.", "Custom Normals: some operations that alter topology (bisect, loopcut, subdivide, knife, triangulate modifier) destroy them while other operations (e.g. Connect Vertex Path, triangulate operator) preserve them\nOperating system: Windows-10-10.0.19043-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 511.65\n\nBroken: version: 3.1.2\nWorked: unknown\n\nMesh bisect generates incorrect normals when applied to a face generated using bevel with \"harden normals\" applied\n\n1. Start with the default cube\n2. In the cube properties, enable auto smooth normals (Object Data -> Normals -> Auto Smooth)\n3. Enter edit mode\n4. Select one edge and bevel it\n5. In the operator panel, ensure that \"harden normals\" is enabled\n6. Select the whole model\n7. Bisect the cube so the bevelled edge is split in half\n\nThe bad normals should now be visible. You can enable normal indicators in Overlays -> Normals -> Split face normals (the second option of the three icons). You can also go to object mode to remove the wireframe and make the bad shading more obvious.\n\n[bevel and bisect.mp4](bevel_and_bisect.mp4)\n\nNote that traditionally, topology altering operations were not expected to preserve custom normals, see for example \n- #69290 (Loop cut and knife tools makes wrong shading on the mesh)\n- #53763 (Operations that alter topology (e.g. Mirror modifier AND mirror operator, loopcut, subdivide, knife) destroy custom normals)\n\nThen there has been 93c8955a72 which was a nice improvement for some operators (but was missing others afaict).\n(check usages of `BM_custom_loop_normals_to_vector_layer`)\n\nSo this now leaves us in the situation which feels puzzling/inconsistent:\nOn the reported geometry for example `Connect Vertex Path` works as expected, whereas `Knife` doesnt:\n[#97698.webm](T97698.webm)", "Mirror Modifier On Cage Selection Papercut \nBroken: 2.93.4\nWorked: Maybe never.\n\nWhen in edit mode, using the \"On Cage\" option for the mirror modifier results in some unintuitive behavior when selecting vertices created by the bisect feature. The vertices created by the by the bisection are mapped to the original mesh data in a way that appears to be based on blenders internal representation of the mesh and not based on something observable to the user.\nThe behavior when selecting should feel consistent to the user. To fix this, I suggest that the vertices created by the bisection should always map to the vertices on the original mesh that are on the removed side of the bisection plane. This way users would be able to tweak the centerline of the mirrored mesh in while using \"On Cage.\" As it is currently implemented, there are vertices that affect the mesh that cannot be accessed through normal click selection unless \"On Cage\" is off.\n\n**Example**\n\n\nSelect the +X or -X face of the default cube and scale it by 0.5.\nAdd the mirror modifier and select bisect for the X axis so that the object is mirrored and bisected about the X axis.\nTurn on \"On cage\" for the mirror modifier then enter edit mode for the object.\nWhen selecting the vertices on plane X=0, you should notice that two of the vertices correspond to visible mirrored vertices and two correspond to vertices removed by the modifier. \n\nSimple scene showing the behavior. \n[MirrorSelectOnCagePapercut.blend](MirrorSelectOnCagePapercut.blend)\n", "The Tool Selector Window close one time every two times, when opened too closed to the bottom border of the screen\nOperating system: Windows 10\nGraphics card: GTX 1080\n\nBroken: 2.80\nWorked: (optional)\n\nThe Tool Window, open and close immediately, choosing the tool under the mouse cursor, if the tool window is opened too close from the bottom of the screen.\n\n1) In Edit > Preferences > Keymap, set the \"Space Bar action\" to \"Tools\"\n2) On the 3D viewport, place your mouse cursor in the top area of the viewport\n3) press the spacebar, the tool menu open without any problem, and you can choose a tool with a shortcut key (for example press B to choose the Box selection)\n4) You can repeat that several times in a row without any problems.\n5) Now move your mouse cursor much more closer to the bottom border of the screen, so that you know that the tool window will have to be shifted upward to be displayed correctly.\n6) Press the spacebar a first time. This is ok, you can select your tool\n7) Now press the spacebar a second time. This time, the tool window open and intermediately closed, and the tool selected is the one that is under your mouse cursor.\n8) Press the spacebar again: the tool window open normally\n9) Press the spacebar again: the bug reappear\n10) Now on, every one time over two, the bug appear.\n\n**Expected behavior**\nThe Tool Window should open and stay opened consistently, no matter where the mouse cursor is placed on the screen, when the spacebar is pressed.", "Measuring tool reports incorrect diameter of circle\nOperating system: Linux-5.19.0-42-generic-x86_64-with-glibc2.35 64 Bits\nGraphics card: Quadro K2200/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 470.182.03\n\nBroken: version: 3.5.1\nWorked: unknown, but this bug has been present since at least 3.0.1\n\nMeasuring tool reports incorrect diameter of circle.\n\n1. Create new General file\n2. Change length unit from meters to millimeters\n3. Add a Mesh->Circle with the default radius of 1 ( mm )\n4. Activate the Measuring Tool, and measure the diameter of your circle\n\nExpected result: 2 mm\nObserved result: 2000 mm ( +/- the accuracy of the user's operation of the tool )\n\nSee attached screenshot for the status at this point.\n\nThe same behavior can be observed with Imperial units, as shown in the second screenshot. The default length unit here was changed to inches, and Blender reports that a 1\" radius circle has a diameter of ~24\".\n\n\nConclusion: Non-default unit settings are not applied to the measuring tool.\n\n", "Bevel when part of mesh is hidden results sometimes in wrong selection of vertices\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: GeForce GTX 1050/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 451.67\n\nBroken: version: 3.2.0 Alpha\nalso broken in the latest official release of version 3.1.2, also broken in 2.93.\n\nBevel when part of the mesh is hidden and the selected points are not all connected results sometimes in the wrong selection of vertices and lines. I have noticed that the selection is wrong if the number of segments in the bevel attrributes is odd.\n\nOn the default startup file do the following:\n1. Edit the default cube, inset the top and bottom face \n2. Inset again, but only 0.001 \n3. Scale the selection in the Z by 0, \n4. Expand the selection (CTRL++)\n\n5. Shift+H to hide everything currently not selected\n6. Only select the top and bottom \n7. Try beveling, play a little with the segments and switch between vertex and line selection, you will see that not all the lines and not all the vertices that should be selected are selected. It seems to be wrong when you try odd number of segments\n\n\n\n\n\n\n\n", "Inconsistency of some tools/operators (e.g. Subdivide) in regards to hidden geometry\nBroken: all or none\n\nI'm not really sure this is even an issue, though it feels like it. As far as I understand it, mesh editing tools are not supposed to affect hidden geometry. But the Subdivide tool does. Consider this mesh:\n\n\nIf you hide the middle face...\n\n\n...and then Subdivide the two remaining faces...\n\n\n...Subdivide will create new edges cutting through the hidden face, also creating inconsistent selection (i.e. selected edges in face select mode). Shouldn't the operator check that that face is hidden and not create edges through it?", "Regression: Damaged normals in sculpting with Multires modifier\nBroken: version: 3.4.0 Alpha\nWorked: version: 3.4.0 Alpha\n\nWhen sculpting, sometimes breaks normals. Switching between modes fixes this.\n| Edited sculpture | Object mode |\n| -- | -- |\n|  |  |\n\n1. Open attached file.\n2. Go to Sculpt mode.\n3. Do something.\nLook at the fact that in some places the normals are damaged.", "Poly Build Tool: Editing poly build keymap causes poly build tool to not work properly\nWindows 10\nGeForce GTX 1080\n\nv2.81 Alpha\nDate: 2019-09-10 15:07\nHash: 8f55794c0e80\n\nWith a default blender open, edit the poly build keymap. The tool doesn't work properly with the new key assignment. \n\nFor example, switch 'face at cursor move' to shift + left mouse, and 'poly build delete at cursor' to ctrl + left mouse.", "non-manifold edge generated after remesh modifier\nOperating system: Windows 10\nGraphics card: NVIDIA Quadro K600\n\nBroken: 2.80 official \nWorked: (optional)\n\nafter applying Remesh modifier with mode as Block, the result object has non-manifold edges.\nthe blocks of issue are sharing only edges instead of faces and make non-manifold edges.\nits rough sketch looks as below:\n\n```\n |-----|\n | |\n |-----x-----|\n | |\n |-----|\n```\n\nwhere non-manifold edge is marked as x\n\nI am not sure if this behavior is a bug or not but I guess this behavior could be avoided by inspecting shared edge and face or something.\n\nBased on the default startup or an attached .blend file (as simple as possible).\n\n[sample.blend](sample.blend) ", "Normal orientation for complex selection work unpredictable\nOperating system: Windows-7-6.1.7601-SP1 64 Bits\nGraphics card: GeForce GTX 660 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.71\n\nBroken: version: 2.92.0 Alpha\n\n\nHere is simple example with cylinder. Same surface with same normals changes it's normal orientation and scale-cage orientation after rotation.\nIt is unexpected and unpredictable.\n\n[2020-11-21_14-27-41.mp4](2020-11-21_14-27-41.mp4)\n\nIt work as expected only if I deselect one of opposite faces\n[2020-11-21_14-31-14.mp4](2020-11-21_14-31-14.mp4)\n", "Workspace duplicate resets my tweak tool back to default select box\nOperating system: Windows-7-6.1.7601-SP1 64 Bits\nGraphics card: GeForce GTX 660 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 442.59\n\nBroken: version: 2.83 (sub 11)\nWorked: ???\n\nAnd EACH time I create new workspace I forced to change it back to tweak, because i prefer it.\nIt is a pain in the … code.\n\nWhy do I think it is a bug? Because scale cage or annotation are saved.\n[2020-04-02_16-44-41.mp4](2020-04-02_16-44-41.mp4)\n\n", "Edge Angle Node -> Signed Angle returns wrong sign\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: GeForce RTX 2070 with Max-Q Design/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 462.59\n\nBroken: version: 3.3.1\nWorked: Never (to my knowledge)\n\nThe Edge Angle Node computes the angle between the two faces connected by an edge. If one of the faces stretch *behind* the edge, then the node may report the edge as being convex even though the edge is concave. \n\nThe .blend-file: [debug_edge_angle_sign.blend](debug_edge_angle_sign.blend)\n\nIllustration of the problem:\n\n\nIn the example, I have two meshes that share a geometry-node setup that highlights convex edges in blue and concave edges in red. A single vertex is different in the two input meshes, but that vertex is not part of any of the highlighted edges. It still affects the signed angle of the concave edge though, making it appear convex.\n", "Remesh Resolution strange behaviour (Shift-R)\nOperating system: MacOS 10.14.6\nGraphics card: AMD Radeon R9\n\nBroken: Daily Build\nWorked: Don't know\n\n**Description**\nSetting the remesh resolution with Shift-R during sculpt when object is unproportional in size makes the grid fade (too quickly) and when zoomed in the numeric display of scale is upside down\n\nIn this small video clip, first part is correct behaviour.. second with disproportional object\n<video src=\"remesh_behaviour.mov\" controls></video>\n\n**Steps**\n- Use default cube\n- In edit mode scale the object 10x in Z only\n- Enter Sculpt mode\n- zoom in\n- Shift-R\n- observe grid fading and display upside down..\n\n[remesh_behaviour.mov](remesh_behaviour.mov)\n\n\n" ]

[ "Bridge tool works a bit odd for co-planar loops\nWin7 64bit | FX2700M\n\nBroken: 2.70 RC\n\nI noticed this when trying to make handle for a mug, It seems not working as expected.\n\nPlease see the attached image. Thanks for checking.\n\n\n" ]

[ "Crash when selecting line art\nOperating system: Linux mint 21. Blender installed using flatpak.\nGraphics card: See attached system_information.txt\n\nBroken: version: 3.5.1\n\nBlender Crashes.\n```\nbacktrace\n\nPython backtrace\nFile \"/app/blender/3.5/scripts/startup/bl_ui/space_view3d.py\", line 704 in draw\nFile \"/app/blender/3.5/scripts/modules/bpy_types.py\", line 903 in draw_ls\n```\n\n```Py\nbpy.ops.outliner.item_activate(deselect_all=True) # Operator\nbpy.ops.outliner.item_activate(deselect_all=True) # Operator\n```\nWith attached blend file, in the outliner, select cameras, line-art then try to change the color of the background.\n\n", "Meshes are invisible if any objects are set to display 'In Front'\nOperating system: Darwin-18.7.0-x86_64-i386-64bit 64 Bits\nGraphics card: Intel(R) HD Graphics 6000 Intel Inc. 4.1 INTEL-12.10.14\n\nBroken: version: 2.83.0\nWorked: 2.82\n\nIf any objects are set to display as 'in front' in a project, all meshes will display only as faint outlines. In edit mode, meshes display similar to how they would if they were set to display as wireframe.\n\nAdd a new mesh, curve, or armature to the scene.\nObject properties -> Viewport display -> Enable 'In Front'\n\n", "Add Image/Sequence to video editing Squencer: number of images are truncated\nOperating system: Win10\nGraphics card: GeForce GTX 1050Ti\n\n**Blender Version** \nbroken: 2.8 RC3\nworked: 2.8 build from ca 1 month ago\n\nWhen you add images by drag 'n drop in de video sequencer, the number of images in the resulting strip is truncated aaround 27. \n\n(1) Open Blender\n(2) from the splash-screen choose: video editing\n(3) Open a windows explorer and select ca 100 jpg files;\n(4) use drag n drop to drop the files in the sequencer;\n\nYou then have a strip of just 27 frames where the selected number of files is expected. This used to work in the beta's", "Application handler not executed when video sequencer contains a strip\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce RTX 3080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 462.59\n\nBroken: version: 2.93.4\n\nA registered handler for frame_change_post does not get executed while rendering an animation when there is a strip present in the video sequencer.\n\n1. Open the example file and turn on the system console\n2. Execute the example script present in the file\n3. Render an animation, note how you can see the handler getting called before each rendered frame in the console\n4. Add a strip to the video sequencer (for example a solid color)\n5. Render another animation, note how this time the handler doesn't get called per frame anymore\n6. Remove the strip again\n7. Render once more, note how it works again as expected\n\n**Example file**\n[bug_example.blend](bug_example.blend)\n", "Eevee camera clipping inconsistent when using child of constraint on an armature.\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 471.41\n\nBroken: version: 2.83.15\nversion: 2.83.17\nversion: 2.83.18 Release Candidate\nversion: 2.93.2\nversion: 2.93.3 Release Candidate\nversion: 3.0.0 Alpha\n\nViewport's material preview and viewport's render has different camera cull than F12's render image, despite same clipping settings\n[2021-08-15_12-16-50.mp4](2021-08-15_12-16-50.mp4)\n\n[CullBugSimplified.blend](CullBugSimplified.blend)\nVerify viewport is in ACamera's view\nSwitch to viewport rendered and material preview to verify cubes on foreground are visible\nRender to verify cubes in foreground have disappeared\nSwitch to ACamera.001 and render to verify cubes in foreground are consistent with viewport\n\n", "Cycles render artifacts relating scene scale\nOperating system: Windows 10\nGraphics card: tested multiple graphics cards with same issue\n\nBroken: 2.83.6 (file originally made in 2.79 and had the same issue)\n\nThere exists in the blend file data only two objects - the camera and a simple plane. Yet when rendering, the result does not match the viewport and there seems to be another ghost object appearing in the scene.\n\nBased on the following startup file: [strange-error.blend](strange-error.blend)\n\n1. Switch to rendered view in the viewport. This is how things should look.\n2. Now render the current frame at frame 97. Notice the strange ghosting effect in the center of the frame.", "Bug with environment texture\nOperating system: Windows 10\nGraphics card: Nvidia GeForce RTX 3050 Laptop GPU\n\n3.6.2\n\nAfter I checked and unchecked \"Open Shading Language\" checkbox Blender crashed, after that I opened my project and turned rendered viewport the environment texture become a one solid color, and as I was rotating my camera it was changing the color.\n\nBased on an attached .blend file you can go to the Render viewport and take a look at this error.\n\n", "When turning VR mode on, the controllers show, even when there are no controllers in the system\n**System**\n* ConceptD 7 (laptop with holographic display using OpenXR backend on Windows)\n\nBroken: version: 4.0.0 Alpha\nWorked: - (technically in old Blenders before we had XR input support)\n\nWhen turning VR mode on, the controllers show, even when there are no controllers in the system.\n\n\nI don't know if the issue is unique to this system, but this is a laptop that uses OpenXR backend to handle head tracking and has an holographic display but no \"VR controllers\".\n\n* Turn on VR\n* You will see a big gizmo in the screen (black circle, with a XYZ gizmo on top of it).\n\nThe gizmo is following Blender's camera (making the experience unusable).\n\n**Workaround**\n\nI can work around that by disabling the visualization of VR controllers. However if the system has no controllers, they shouldn't be drawn in the first place:\n\n", "Regression: Failure to render stereoscopy correctly in background (rendering of curve objects with cycles and motion blur)\nOperating system: Linux-5.19.0-41-generic-x86_64-with-glibc2.35 64 Bits\nGraphics card: AMD Radeon Vega 8 Graphics (raven, LLVM 15.0.6, DRM 3.47, 5.19.0-41-generic) AMD 4.6 (Core Profile) Mesa 22.2.5\n\nBroken: version: 3.5.1\n\nCaused by b32f5a922f05989ff129c3d266ebbed14b751e2a\n\nIn stereoscopic rendering with cycles curve objects are missing (mesh objects are fine) for the right eye if motion blur is activated. This only occurs in rendering without GUI via command line.\n\nCreate new \"General\" file\nDelete Cube\nAdd Bezier-Curve\nSet Curve Data-Properties: Geometry-Bevel Depth to 0.1 m\n\nRender Properties: Cycles; Motion blur: ON\nOutput Properties: Stereoscopy ON\nSet Framerange from 1 to 1\nsave as \"test.blend\"\n\nRender animation from GUI -> works properly\nCommand line: \n$ blender --background test.blend -a\n\nproduces a frame without the curve for the right eye.\n\nThe bug has similarities to the one reported by @SimeonConzendorf in /blender/blender/issues/70114#issue-38464 several years ago (Version 2.81). The bug seems to have irregularities, I had a similar behaviour in version 3.0.0 where it occured not in every frame in a quite complex scene. But I was not able to reproduce it in a simple way. The uploaded file doesn't reproduce the bug in that version and 3.0.1, but after these in 3.1.0, 3.3.1 and the actual daily build blender-3.6.0-alpha+main.d5fc1b9ba4b8-linux.x86_64-release \n\nI uploaded smaller anaglyph images instead of the separated images for better illustration.\n\n", "Camera background image is hidden by GPencil fill with alpha\nOperating system: Linux-4.14.78-gentoo-x86_64-AMD_FX-tm-8350_Eight-Core_Processor-with-gentoo-2.6 64 Bits\nGraphics card: GeForce GT 730/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 410.73\n\n\nBroken: version: 2.80 (sub 75)\nWorked: (optional)\n\n\nWhen I change the fill alpha while a background image is loaded, it behaves weirdly. i.e, when the alpha value is 0, the background image is visible, when alpha crosses 0.001, the image mysteriously disappears.\n\nIn Grease Pencil, I added a background image in the camera object data context menu,\nthen I drew a box and changed the alpha of fill, then the above said bug appears.\n\nI have included the blend file and a video to produce the bug.\n\n[blenderGreasePencilBugVid.mp4](blenderGreasePencilBugVid.mp4)\n\n[greasePencilBug.blend](greasePencilBug.blend)\n\n", "Black Artefacts in EEVEE on Transparent BSDF when using Alpha Hashed for Blend mode and Camera Depth Of Field\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce RTX 2080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.71\n\nBroken: version: 2.92.0\n\nBlack Artefacts in EEVEE on Transparent BSDF when using Alpha Hashed for Blend mode and Camera Depth Of Field, if Shader Color is not pure white (similar problem with Layer Weight)\n\n**Exact steps for others to reproduce the error**\n- Open attached file (Just a simple box (can be a plane) with Transparent BSDF. Camera' distance or depth of focus influences the result. Attached pictures in Image Editor from a real case).\n- Press F12 to render\n[transprarent_shader_artifacts_with_camera_DOF1.blend](transprarent_shader_artifacts_with_camera_DOF1.blend)\n\n\n", "Blender 4.0.0 - Armature Rig with IK broken\nOperating system: MacOs Ventura 13.4.1\nGraphics card: Radeon Pro\n\nBroken: version: 4.0.0 Alpha, branch: main, commit date: 2023-08-23 19:52, hash: 7c2dc5183d03, type: release\nCaused by 6b872079fec9336865638ad9cfa76e37e8e627c5\n\nI have rigged mesh, commercially available at ArtStation. It worked well up until now. In 4.0.0 it is broken:\n- shapes not rendered until i switch to wireframe mode and back\n- IK ignored\n\nLook at video - i open the file in 3.6.3 and in 4.0.0. In 3.6.3 everything fine, in 4.0.0 the problem mentioned occur.\nBlend file also attached\n\n", "Faces culled with \"backface culling\" still receive shadows in the viewport\nSorry if this is a repeat report. I was unable to find another report for this.\n\nOperating system: NA\nGraphics card: NA\n\nBroken: All versions of Blender 2.8X and 2.90 that I am able to test.\nWorked: Never\n\n**Short description of error:**\nWhile the \"Backface culling\" option is enabled in the viewport, the culled backface will still receive a shadow when the \"shadow\" option is also enabled.\n\n\n\nFor example, here's a box with all it's normals facing inwards. With backface culling, we are able to see inside the box. But with shadows also enabled (as seen in the picture), shadows are cast onto the \"culled\" faces.\n\n\n\n**Exact steps for others to reproduce the error:**\n1. Create scene with an object that has it's normal facing away from the camera (you're seeing the backface).\n2. Enable \"Backface culling\" in the viewport shading options.\n3. Enabled \"Shadows\" and adjust the position of the \"light\" such that the \"back face is in shadow.\n4. The backface should now have a shadow placed on it.\n5. If you reproduce the steps above for the workbench render engine, the same issue applies.\n\n[Backface Shadowing.blend](Backface_Shadowing.blend)", "shadow catcher data missing from diffuse passes when transparency is disabled, and shadow catcher object is masking environment when transparency is enabled (can't recombine moving scene envronment)\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 526.98\n\nBroken: version: 3.5.1 Release Candidate\n\nshadow catcher data missing from diffuse passes when transparency is disabled, and shadow catcher object is masking environment when transparency is enabled (can't recombine moving scene envronment). This makes it impossible to rebuild the image unless the shadow catcher pass is used, which is not suitable when using vector motion blur.\n\nOpen below blend file (choose to load ui)\ngo to scene 'no transparency'\nin the compositor press f12 to render a frame\ncompare the combined image and the rebuilt image\n\ngo to the second scene 'transparency'\nin the compositor hit f12 to render again\nthis time the environment is masked out by the shadow catcher, making it impossible to use the rendered environment (which is necessary if the camera is animated)\n\nrequired behaviour - \n\nshadows appear in diffuse passes when transparency is disabled, so the rebuilt image matches the viewport.\n\nshadow catcher object doesn't mask out the environment pass when transparency is enabled (so the environment can be added back to the image when the camera is moving (backplate cant be used here))\n\n\n\nThe only possible solution currently is to always render with transparency enabled, and then render your scene a second time with only the environment visible, which causes issues when rendering via scripts.\n\n", "Shadow catcher issue with Cycles Standalone\nOperating system: Win10\nGraphics card: 2070 super\n\nN/A Cycles standalone\n\nShadow catcher is not working properly\n\nI've modified the XML parser of Cycles standalone to read shadow_catcher tag as well as create a new sample, derived from scene_monkey.xml, that uses a plane with shadow catcher enabled.\n\nNotice that the shadow is not visible on the output (you can use something like in the commandline \"scene_monkey_shadow.xml --samples 128 -output Temp.png --background\" \n\n[shadow_catcher_issue.zip](shadow_catcher_issue.zip)" ]

[ "Camera Background Image disappears when using XRay\nOperating system: Windows-10-10.0.17134 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 419.67\n\n\nBroken: version: 2.80 (sub 53)\nWorked: (optional)\n\nIt worked a week ago.\n\nWhen reducing the XRay slider to a value less then 1, the background image of the camera disappears.\n\nAdd background image to camera.\nIn Workbench mode, reduce XRay slider in the shading popover to less then 1.\nBackground image disappears.\n\n", "Problem with camera background image and wireframe mode (for camera mapping)\nOperating system: Darwin-17.3.0-x86_64-i386-64bit 64 Bits And Linux Mint.\nGraphics card: NVIDIA GeForce GTX 780M OpenGL Engine NVIDIA Corporation 4.1 NVIDIA-10.28.10 355.11.10.10.20.111\n\n\nBroken: version: 2.80 (sub 51)\nWorked: (optional)\n\n\nWhen we are inside a camera with a \"background image\" when I have to do a \"camera mapping\", the moment I activate the wireframe mode to model the objects that are going to be projected, the background image disappears, and I can't continue modeling the objects since I don't see the background image. If I put the object in its options that appears as wire does not help me because it does not allow me in solid mode to select the faces behind the object through the front faces. \n\n1 Open blender\n2 Go to inside camera\n3 Select the camera, and active background image\n4 Choose any image\n5 In viewport press z and select wireframe mode in pie menu.\n\nNow background image dissapear. ", "fspy image disappears in wireframe mode\nLinux Mint 19\nIntel Graphics\n\nblender-2.80-fc336f973d52-linux-glibc224-x86_64 (dated 30 May 2019)\n\nFollowing this video: watch?v=H3ZTweNN9WY it seemed to work ok around 4 February 2019 (date of video).\n\nUsing the Z pie menu and selecting wireframe, makes the background image disappear. Solid mode shows it again.", "Workbench Engine - Camera Background Image (depth back) disapears if viewport display color of one material has alpha value <1\nOperating system: Windows-10-10.0.17134 64 Bits\nGraphics card: GeForce GTX 1080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 419.35\n\n\nBroken: version: 2.80 (sub 55)\nWorked: (optional)\n\n\nWorkbench Engine - Camera Background Image disapears if the viewport display color of one material has alpha value <1.\n\nOpen the attached *.blend.\nSet the alpha value of the viewport display color <1.\n\nThe camera background image disapears.\nThe front option is working but you have to adjust the alpha value to see geometry.\nThe problem is only in workbench. \n[Camera_Background_Back_not_visible.blend](Camera_Background_Back_not_visible.blend)\n", "Camera's background image invisible in wireframe mode\nOperating system: Ubuntu 18.04 Linux-4.15.0-46-generic-x86_64-with-debian-buster-sid 64 Bits \nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 390.116\n\n\nBroken: version: 2.80 (sub 51)\nWorked: 2.79b\n\n\nCamera's background image invisible in wireframe mode\n\n1. Open new file.\n2. Go to camera view\n3. Put background image in the camera (drag&drop image while in camera view)\n4. Toggle wireframe and any outer shading mode. You will not see the image in wireframe mode.\n\n\n\n\n\n", "Background plate is not visible in viewport when x-ray mode is enabled.\nOperating system: Windows-10-10.0.17763 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 419.72\n\n\nBroken: version: 2.80 (sub 72)\nWorked: (optional)\n\n\nThe backGround plate/imagePlane goes invisible when xray is enabled .\n\n1) Open default blender 2.8 daily build.\n2) Look through the camera.\n3) setup backGround plate in the camera's properties window.\n4) The backGround Image is visible through the camera in solid mode .\n5) If you press \"z\" for wireframe mode, the backGround disappears. This is a bug in my opinion.\n6) The backGround image-plane also disappears when the x-ray toggle is \"on\"\n6) The bgPlate should be visible always in whichever mode. It is very essential when doing photomodels and VFX stuff in usual.\n7) This worked in blender 2.79.\n", "Camera Background Images not visibile in Wireframe nor Rendered Display\nOperating system: elementary 0.4.1\nGraphics card: GTX 1080\n\nBroken: 2.80, 6f9819e5fd56\nWorked: 2.80, 69b2f5268114 \n\nCamera Background Images are not visible in Wireframe mode nor in Cycles Rendered display mode, when the Depth option is set to Back. When the Depth option of background images is set to Front it works in both cases. \nIt it did work in Wireframe mode in February in 69b2f5268114, but not in Rendered view. \n\nHere's a file demonstratin the issue. [background_image_wireframe.blend](background_image_wireframe.blend)\nThere is no background image visible, unless you set Depth option to Front." ]

[ "Non animatable properties\nBroken: version: 3.3.0 Alpha\n\nWorking on refreshing the in-built AnimAll addon by adding the latest properties supported in Blender.\n@ideasman42 asked me to file a task\n\nCertain properties are missing animation capabilities:\n\n - Lattice Vertex Groups\n - Mesh Vertex Creases", "Curve \"solidify\" modifier glitch\nOperating system: Windows 7\nGraphics card: GT750M\n\nBroken: 2.80, 2019-02-28\nWorked:\n\nWhen applying a \"Solidify\" modifier to a bezier curve shape (or imported SVG), it creates random height vector points, creating strange triangles. This can be fixed by simply selecting the errored vector point then selecting one of the transform value (in my case it was Y) in the panel and either pushing enter (keeping the same value) or unselecting the vector point.\n\n\nAttached blend file with imported SVG.\n\n", "crash when i try to make instances real for nodes\nhi , in attached files all info about the crash , its when i try to make instances real for nodes modifier addon scifi pro + flex\nso can i fix it ?\n\n", "modifier display/render driver not responding.\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1050/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 527.56\n\nBroken: version: 3.4.0\nWorked: unknown\n\nIf I have my Mask display/render driver from another objects display/render setting, changing controlling obj disp/render doesn't take effect without coaxing.\n\n1. Open the attached file. \n2. In outliner, disable the viewport display of the Cube object (drives Mask visibility of part of Sphere object)\n3. For the Mask modifier, slide the Threshold a little. Display effect now takes place.\n\nThis all works without a problem if I animate the Cubes visibility. This appears to be a problem for any/all modifier display drivers (e.g., Subdiviison). Sometimes I just need to hover over a value like Threshold, or just over time it responds to the change.\n\n", "Blender FBX won't export Mesh Geometry converted from Curve with geometry nodes\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 535.98\n\nBroken: version: 3.5.1\n\nWhen creating meshes procedurally using geometry nodes modifier on a curve object, this object won't get exported by FBX exporter even if the result is a mesh, not a curve.\n\n1. Create a curve object\n2. Add geometry nodes modifier, create new GN node network \n3. Convert the curve to a mesh and add some topology (for example using curve fill)\n4. Export this object using FBX\nResult: The mesh is not exported\nExpected: The mesh is exported\n\n", "Polishing Correct Face Attributes\n- [ ] Initialize operators options based on the \"correct face attribute\" value (e.g., edge slide).\n- [ ] Unify naming of the option and the operator options.\n\nThe way we do Proportional Editing seems to be a good example to follow. These issues have been raised on 4387aff", "Curves edit mode Bezier handle drawing\n- [ ] Draw points at handle positions for Bezier curves\n\t- [ ] Use handle selection attributes for point highlight\n- [ ] Draw Bezier handle segments\n\t- [ ] Use different colors depending on handle type\n- [ ] Support `View3DOverlay.display_handle` property\n\nRelated: #105038, #96455", "Rendering animation using \"Generate\" type modifiers after animated displacement texture crashes Blender\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 1650/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 460.79\n\nBroken: version: 2.92.0 Alpha\nWorked: 2.83.10\n\nIf 2 high poly (~100k faces per obj) objects in modifiers properties have after Displacement modifier (with animated texture) \"Generate\" type modifiers (I've tried Boolean and Decimate), Blender cant run Animation render (but still can run Image render). \nCrash happens on 2.9 and higher. I've tested this on 2.83.10 LTS (from steam) and everything worked fine there.\nCrashes with Decimate a little bit hard to recreate. So I will go with steps for \"Boolean crash\"\n\n- create Cude, Empty and Icosphere\n- Add Subdivision mod. to Cube: set it to \"Simple\" and 7 levels\n- Add Displacement mod. to Cube and set texture (Clouds) coordinates to object - Empty\n- Add Boolean mod. (difference) and select object Icosphere\n- Select Cude with Icosphere and duplicate them\n- Save and press Render Animation\n[bug.blend](bug.blend)\n[bug.crash.txt](bug.crash.txt)\n\n```lines\n## 3883==ERROR: AddressSanitizer: heap-use-after-free on address 0x61500034ec08 at pc 0x000016fc795d bp 0x7fffc19dd6c0 sp 0x7fffc19dd6b0\nREAD of size 8 at 0x61500034ec08 thread #15\n - 0 0x16fc795c in blender::gpu::GLBatch::verts_(int) const /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/opengl/gl_batch.hh:112\n - 1 0x17013501 in blender::gpu::GLVertArray::update_bindings(unsigned int, GPUBatch const*, blender::gpu::ShaderInterface const*, int) /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/opengl/gl_vertex_array.cc:124\n - 2 0x16fa900b in blender::gpu::GLVaoCache::vao_get(GPUBatch*) /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/opengl/gl_batch.cc:273\n - 3 0x16fa84d8 in blender::gpu::GLVaoCache::base_instance_vao_get(GPUBatch*, int) /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/opengl/gl_batch.cc:234\n - 4 0x16fa9d3e in blender::gpu::GLBatch::bind(int) /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/opengl/gl_batch.cc:317\n - 5 0x16fa9f3f in blender::gpu::GLBatch::draw(int, int, int, int) /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/opengl/gl_batch.cc:328\n - 6 0x16eca2be in GPU_batch_draw_advanced /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/intern/gpu_batch.cc:282\n - 7 0x16fc1207 in blender::gpu::GLDrawList::append(GPUBatch*, int, int) /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/opengl/gl_drawlist.cc:122\n - 8 0x16efa9f5 in GPU_draw_list_append /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/intern/gpu_drawlist.cc:52\n - 9 0x5978acc in draw_indirect_call /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/intern/draw_manager_exec.c:598\n - 10 0x5978acc in draw_call_batching_flush /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/intern/draw_manager_exec.c:795\n - 11 0x597c5f0 in draw_call_batching_finish /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/intern/draw_manager_exec.c:891\n - 12 0x597eae7 in draw_shgroup /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/intern/draw_manager_exec.c:1056\n - 13 0x597f89f in drw_draw_pass_ex /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/intern/draw_manager_exec.c:1113\n - 14 0x597fe50 in DRW_draw_pass /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/intern/draw_manager_exec.c:1153\n - 15 0x5c51bc0 in EEVEE_velocity_resolve /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/engines/eevee/eevee_effects.c:466\n - 16 0x5c5202b in EEVEE_draw_effects /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/engines/eevee/eevee_effects.c:486\n - 17 0x5a21f42 in EEVEE_render_draw /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/engines/eevee/eevee_render.c:671\n - 18 0x59b1dbc in eevee_render_to_image /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/engines/eevee/eevee_engine.c:559\n - 19 0x5948820 in DRW_render_to_image /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/intern/draw_manager.c:1909\n - 20 0x14ff1d77 in engine_render_view_layer /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/engine.c:786\n - 21 0x14ff40a3 in RE_engine_render /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/engine.c:914\n - 22 0x1501aafb in do_render_3d /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/pipeline.c:1132\n - 23 0x1501c24a in do_render /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/pipeline.c:1209\n - 24 0x1501e036 in do_render_composite /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/pipeline.c:1349\n - 25 0x15022789 in do_render_all_options /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/pipeline.c:1615\n - 26 0x15027839 in RE_RenderFrame /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/pipeline.c:2015\n - 27 0xd7b42cf in render_startjob /home/dev/01-data/01-git/blender-git/blender/source/blender/editors/render/render_internal.c:630\n - 28 0x4fed4f1 in do_job_thread /home/dev/01-data/01-git/blender-git/blender/source/blender/windowmanager/intern/wm_jobs.c:389\n - 29 0x18de408d in tslot_thread_start /home/dev/01-data/01-git/blender-git/blender/source/blender/blenlib/intern/threads.cc:231\n - 30 0x7ffff7566608 in start_thread /build/glibc-ZN95T4/glibc-2.31/nptl/pthread_create.c:477\n - 31 0x7ffff6f46292 in __clone (/lib/x86_64-linux-gnu/libc.so.6+0x122292)\n\n0x61500034ec08 is located 8 bytes inside of 512-byte region [0x61500034ec00,0x61500034ee00)\nfreed by thread #15 here:\n - 0 0x7ffff76927cf in __interceptor_free (/lib/x86_64-linux-gnu/libasan.so.5+0x10d7cf)\n - 1 0x18e0eab0 in MEM_lockfree_freeN /home/dev/01-data/01-git/blender-git/blender/intern/guardedalloc/intern/mallocn_lockfree_impl.c:129\n - 2 0x16ef56c6 in blender::gpu::GLVertBuf::operator delete(void*) /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/opengl/gl_vertex_buffer.hh:57\n - 3 0x17018021 in blender::gpu::GLVertBuf::~GLVertBuf() /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/opengl/gl_vertex_buffer.hh:35\n - 4 0x16f8bb19 in blender::gpu::VertBuf::reference_remove() /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/intern/gpu_vertex_buffer_private.hh:94\n - 5 0x16f877f5 in GPU_vertbuf_discard /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/intern/gpu_vertex_buffer.cc:161\n - 6 0x5a0eab8 in EEVEE_motion_blur_cache_finish /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/engines/eevee/eevee_motion_blur.c:449\n - 7 0x59b1bef in eevee_render_to_image /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/engines/eevee/eevee_engine.c:542\n - 8 0x5948820 in DRW_render_to_image /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/intern/draw_manager.c:1909\n - 9 0x14ff1d77 in engine_render_view_layer /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/engine.c:786\n - 10 0x14ff40a3 in RE_engine_render /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/engine.c:914\n - 11 0x1501aafb in do_render_3d /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/pipeline.c:1132\n - 12 0x1501c24a in do_render /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/pipeline.c:1209\n - 13 0x1501e036 in do_render_composite /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/pipeline.c:1349\n - 14 0x15022789 in do_render_all_options /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/pipeline.c:1615\n - 15 0x15027839 in RE_RenderFrame /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/pipeline.c:2015\n - 16 0xd7b42cf in render_startjob /home/dev/01-data/01-git/blender-git/blender/source/blender/editors/render/render_internal.c:630\n - 17 0x4fed4f1 in do_job_thread /home/dev/01-data/01-git/blender-git/blender/source/blender/windowmanager/intern/wm_jobs.c:389\n - 18 0x18de408d in tslot_thread_start /home/dev/01-data/01-git/blender-git/blender/source/blender/blenlib/intern/threads.cc:231\n - 19 0x7ffff7566608 in start_thread /build/glibc-ZN95T4/glibc-2.31/nptl/pthread_create.c:477\n\npreviously allocated by thread #15 here:\n - 0 0x7ffff7692bc8 in malloc (/lib/x86_64-linux-gnu/libasan.so.5+0x10dbc8)\n - 1 0x18e0f510 in MEM_lockfree_mallocN /home/dev/01-data/01-git/blender-git/blender/intern/guardedalloc/intern/mallocn_lockfree_impl.c:276\n - 2 0x16ef5676 in blender::gpu::GLVertBuf::operator new(unsigned long) /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/opengl/gl_vertex_buffer.hh:57\n - 3 0x16ef70a2 in blender::gpu::GLBackend::vertbuf_alloc() /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/opengl/gl_backend.hh:121\n - 4 0x16f85d10 in blender::gpu::VertBuf::duplicate() /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/intern/gpu_vertex_buffer.cc:77\n - 5 0x16f8750d in GPU_vertbuf_duplicate /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/intern/gpu_vertex_buffer.cc:149\n - 6 0x5a0ee06 in EEVEE_motion_blur_cache_finish /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/engines/eevee/eevee_motion_blur.c:467\n - 7 0x59b1953 in eevee_render_to_image /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/engines/eevee/eevee_engine.c:507\n - 8 0x5948820 in DRW_render_to_image /home/dev/01-data/01-git/blender-git/blender/source/blender/draw/intern/draw_manager.c:1909\n - 9 0x14ff1d77 in engine_render_view_layer /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/engine.c:786\n - 10 0x14ff40a3 in RE_engine_render /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/engine.c:914\n - 11 0x1501aafb in do_render_3d /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/pipeline.c:1132\n - 12 0x1501c24a in do_render /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/pipeline.c:1209\n - 13 0x1501e036 in do_render_composite /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/pipeline.c:1349\n - 14 0x15022789 in do_render_all_options /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/pipeline.c:1615\n - 15 0x15027839 in RE_RenderFrame /home/dev/01-data/01-git/blender-git/blender/source/blender/render/intern/pipeline.c:2015\n - 16 0xd7b42cf in render_startjob /home/dev/01-data/01-git/blender-git/blender/source/blender/editors/render/render_internal.c:630\n - 17 0x4fed4f1 in do_job_thread /home/dev/01-data/01-git/blender-git/blender/source/blender/windowmanager/intern/wm_jobs.c:389\n - 18 0x18de408d in tslot_thread_start /home/dev/01-data/01-git/blender-git/blender/source/blender/blenlib/intern/threads.cc:231\n - 19 0x7ffff7566608 in start_thread /build/glibc-ZN95T4/glibc-2.31/nptl/pthread_create.c:477\n\nThread #15 created by T0 here:\n - 0 0x7ffff75bf805 in pthread_create (/lib/x86_64-linux-gnu/libasan.so.5+0x3a805)\n - 1 0x18de42b9 in BLI_threadpool_insert /home/dev/01-data/01-git/blender-git/blender/source/blender/blenlib/intern/threads.cc:245\n - 2 0x4fee516 in WM_jobs_start /home/dev/01-data/01-git/blender-git/blender/source/blender/windowmanager/intern/wm_jobs.c:484\n - 3 0xd7b88bd in screen_render_invoke /home/dev/01-data/01-git/blender-git/blender/source/blender/editors/render/render_internal.c:1041\n - 4 0x4f8f3ba in wm_operator_invoke /home/dev/01-data/01-git/blender-git/blender/source/blender/windowmanager/intern/wm_event_system.c:1293\n - 5 0x4f912f2 in wm_operator_call_internal /home/dev/01-data/01-git/blender-git/blender/source/blender/windowmanager/intern/wm_event_system.c:1500\n - 6 0x4f91551 in WM_operator_name_call_ptr /home/dev/01-data/01-git/blender-git/blender/source/blender/windowmanager/intern/wm_event_system.c:1548\n - 7 0x90f1c95 in ui_apply_but_funcs_after /home/dev/01-data/01-git/blender-git/blender/source/blender/editors/interface/interface_handlers.c:932\n - 8 0x917b8d0 in ui_handler_region_menu /home/dev/01-data/01-git/blender-git/blender/source/blender/editors/interface/interface_handlers.c:10809\n - 9 0x4f88a9b in wm_handler_ui_call /home/dev/01-data/01-git/blender-git/blender/source/blender/windowmanager/intern/wm_event_system.c:636\n - 10 0x4f9ec36 in wm_handlers_do_intern /home/dev/01-data/01-git/blender-git/blender/source/blender/windowmanager/intern/wm_event_system.c:2757\n - 11 0x4f9fc78 in wm_handlers_do /home/dev/01-data/01-git/blender-git/blender/source/blender/windowmanager/intern/wm_event_system.c:2865\n - 12 0x4fa537f in wm_event_do_handlers /home/dev/01-data/01-git/blender-git/blender/source/blender/windowmanager/intern/wm_event_system.c:3288\n - 13 0x4f732cf in WM_main /home/dev/01-data/01-git/blender-git/blender/source/blender/windowmanager/intern/wm.c:635\n - 14 0x34f58fa in main /home/dev/01-data/01-git/blender-git/blender/source/creator/creator.c:522\n - 15 0x7ffff6e4b0b2 in __libc_start_main (/lib/x86_64-linux-gnu/libc.so.6+0x270b2)\n\nSUMMARY: AddressSanitizer: heap-use-after-free /home/dev/01-data/01-git/blender-git/blender/source/blender/gpu/opengl/gl_batch.hh:112 in blender::gpu::GLBatch::verts_(int) const\nShadow bytes around the buggy address:\n 0x0c2a80061d30: fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd\n 0x0c2a80061d40: fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd\n 0x0c2a80061d50: fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd\n 0x0c2a80061d60: fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd fa\n 0x0c2a80061d70: fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa\n# >0x0c2a80061d80: fd[fd]fd fd fd fd fd fd fd fd fd fd fd fd fd fd\n 0x0c2a80061d90: fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd\n 0x0c2a80061da0: fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd\n 0x0c2a80061db0: fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd\n 0x0c2a80061dc0: fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa\n 0x0c2a80061dd0: fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd\nShadow byte legend (one shadow byte represents 8 application bytes):\n```\nAddressable: 00\nPartially addressable: 01 02 03 04 05 06 07 \nHeap left redzone: fa\nFreed heap region: fd\nStack left redzone: f1\nStack mid redzone: f2\nStack right redzone: f3\nStack after return: f5\nStack use after scope: f8\nGlobal redzone: f9\nGlobal init order: f6\nPoisoned by user: f7\nContainer overflow: fc\nArray cookie: ac\nIntra object redzone: bb\nASan internal: fe\nLeft alloca redzone: ca\nRight alloca redzone: cb\nShadow gap: cc\n```\n", "Boolean modifier - Intersect option - not working / artifacts while doing 3D Sectioning\nOperating system: Linux-5.4.0-72-generic-x86_64-with-glibc2.31 64 Bits\nGraphics card: Mesa DRI Intel(R) HD Graphics 4600 (HSW GT2) Intel Open Source Technology Center 4.5 (Core Profile) Mesa 20.2.6\n\nBroken: version: 2.93.0 Beta\nWorked: in Blender 2.92\n\nStrange artifacts using 3D sectioning using Boolean modifier - Intersect (see sectioning animation issues below)\n\nExpected behavior:\n3D slice working correctly in Blender 2.92 (to view internal section of object)\n```\n-Create Mesh Plane (larger than object)\n-Select Plane - Add Boolean Modifier \n\tOptions - Intersect\n\tObject - STL or Object\n\tSolver - Exact\n\n```\n\n\nActual behavior:\n3D slice issues using Blender 2.93 beta (to view internal section of object)\n1) Not showing object as a 3D section like in 2.92 (see animation above)\n2) Artifacts Showing up when moving cutting plane back and forth (see animation below) \n\n\n\n\nIn the blend file, move the `Plane_xy` along the z axis, or move the `Plane_xz` along the y axis.\n[section.blend](section.blend)\n\n", "Material glitches at joints when using Bevel modifier\nOperating system: Ubuntu 20.04.3 LTS\nGraphics card: GeForce RTX 2060\n\nBroken: 3.1.2, cc66d1020c3b, master, 2022-03-31\nWorked:\n\n\nI know people experience no end of unexpected Bevel behavior due to issues with their geometry. But I think this is a genuine bug - as do people who responded to my Blender SE question: 124535\n\nRobin Betts commented there that \"it looks like a bug. But at least a consistent, and therefore traceable bug. It seems to be in the 'Automatic' (-1) material offset setting, which should split the bevel down the middle at material boundaries. It's getting the boundary 1-off at the end of a patch/arc spread of faces.\"\n\nAt joints where two materials meet, the Bevel modifier seems to miscalculate which faces of the bevel should contain one material and which the other, as shown here:\n\n\n\nIf I apply the modifier, I can then correct the issue by hand.\n\nThe above picture is from Blender 3.1.2. The problem seems to have been there for a long time, things look different but are still incorrect in 2.93.7:\n\n\n\nThe complete mesh is fairly simple (I cut it out of a larger object for this bug report) and I've confirmed there are no overlapping vertices etc.:\n\n\n\nI actually started with a much simpler mesh but converted to triangles to ensure the issue wasn't somehow n-gon related (it isn't):\n\n\n\n\nCreate a simple shape like the one shown in the last image above and use two different materials for your object, e.g. above, I've applied one material to the darker colored faces and the other to the lighter colored faces.\n\nThen apply a *Bevel* modifier and then look at the object in a *Viewport Shading* mode that shows materials, e.g. *Material Preview* or *Rendered* (you can even see the issue with *Solid* viewport shading if you set the *Viewport Display* for one of the materials to be a different color to the default grey).\n\nHere are two small `.blend` files - the first created in Blender 3.1.2 and the second in Blender 2.93.7:\n\n[bevel-3.1.2.blend](bevel-3.1.2.blend)\n\n[bevel-2.93.7.blend](bevel-2.93.7.blend)\n\nYou can find a larger `.blend` file containing the original full mesh in my Blender SE question: 124535", "bevel modifier shading error\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce RTX 2070 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 457.51\n\nBroken: version: 2.91.0\n\nHarden Normals option inside the Bevel Modifier, generates shading error.\n\nOpen the file to check it.\n\n\n[bevel.blend](bevel.blend)", "if multiple splines on the same place... no selection is visable\nOperating system: Linuxmint 19.0\nGraphics card:Geforce GTX1050\n\nBlender Version\n81ea815dcb6, blender2.8, 2018-12-12\n\nI created a spline for modelling.\nThen I converted the spline into a mesh.\n\nI want to delete some vertices, but I could not select any.\nBy clicking an moving I saw there is 1 vertice selected.\n\nBut any other selection is blind.\n\nBy the way: I don't even know how these >10 splines were created...\nsuddenly they where there...\n\n\n[Bug-Multiple Objects.blend](Bug-Multiple_Objects.blend)\n", "Curves sculpting collision with surface.\n", "Cycles does not render geo with Multires modifier while in Sculpt mode\nOperating system: Windows 10, Linux\nGraphics card: Nvidia GTX 960\n\nBroken: (example: 2.80, edbf15d3c044, master, 2018-11-28, as found on the splash screen)\nWorked: (optional)\n\nAdding a Multires modifier with the Sculpt parameter > 0 (to further develop the model I'm working on) makes the geo disappear when using Cycles in the viewport, while in Sculpt Mode. As a first time Blender user, I'm not sure if this is the expected behaviour for this. I have checked with other users and they can reproduce the error. \n\nIt would be useful to render in viewport without leaving Sculpt Mode.\n\nCreate a geometry. Add a Multires modifier. Set Sculpt to a value greater than 0. Enter Sculpt Mode. Change the viewport render from Eevee to Cycles. The geometry disappears on rendering. I have a sample of the problem in this tweet I published several hours ago: 1150657198374080512\n\nThanks!", "Text Object -> Edit Mode, Many Short-cuts Showing Here & There, But Not Working \nOperating system: Windows-10-10.0.17134 64 Bits\nGraphics card: AMD Radeon HD 8400 / R3 Series ATI Technologies Inc. 4.5.13399 Core Profile Context 15.200.1065.0\n\n\nBroken: version: 2.80 (sub 74)\nWorked: (Probably Never)\n\n\nFor Text Object -> Many Short-cuts are showing in e.g. menus but not working. e.g. T (Tool Shelf), N (Numeric Pane) e.t.c\n\n\n\n\n*Wait:* I am checking a list the all short-cuts which are not working..." ]

[ "Modifiers on curves/text objects in 2.8\nWin 10 , GTX-1050/80\n\n\nBroken: 2.80-63150511a29-win64\n\n\n\nIt seems like there are a couple of issues for text/curve objects with modifiers in 2.8:\n### 1. deforming modifiers\n- I see a TODO comment in `MOD_get_mesh_eval(): /* TODO(sybren): get evaluated mesh from depsgraph once that's properly generated for curves. */\n - @dr.sybren A. Stüvel (sybren) : do you know the status of this?\n\n- Then for following modifiers:\n - we are asserting BLI_assert(mesh_src->totvert == numVerts);\n - if I remove the assert, the modifiers are actually working seemingly fine for e.g. text objects\n - @mont29: do you know the status of this (are these neccessary only for mesh objects)?\n - - [x] Cast\n - - [x] Curve\n - - [x] Shrinkwrap\n - - [x] Smooth\n - - [x] Warp\n\n### 2. generative modifiers on curves not working (basically all of them):\n- [ ] Array\n- [ ] Bevel (wasnt even there in 2.79)\n - What is the point here? Bevel of surfaces???\n- [ ] Build\n- [ ] Decimate\n- [ ] Mirror\n- [ ] Remesh (had bad results in 2.79, but not working at all in 2.8)\n- [ ] Screw\n- [ ] Solidify\n- I havent looked into the cause of this yet...\n- maybe @mont29 would know right away?\n\n### 1. e.g. Cast modifier asserts\nAdd text object\nAdd cast modifier -> assert\n### 2. e.g. Solidify modifier doesn't increase thickness of text objects in 2.8\nAdd text object\nAdd solidify modifier\nincrease thickness -> text remains flat \n\n\n" ]

[ "Texture Node Editor Using nodes causes weird issues in sculpt mode.\nOperating system: Linux-5.15.0-48-generic-x86_64-with-glibc2.31 64 Bits\nGraphics card: NVIDIA GeForce RTX 2060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 515.76\n\nBroken: version: 3.3.1\nWorked: perhaps never? (can redo in 2.83)\n\nAdding an image from the nodes of the texture editor for example, a sculpting brush, causes really weird issues. Disabling nodes and using the texture properties does not have this issue.\n\n- Check the blend file. Try sculpting, you will see weird noise being displaced with the texture.\n- Now disable the nodes and set the texture type from the texture properties to the same image. Sculpting again you see the issue is not happening.\n[bug.blend](bug.blend)\n\n\n\n", "object display as wire , invisible when overlay is turned off \nwin 10\nintel i5\namd radeon \n\nblender 2.8 \n6b39dc7672eb\n\n\nNot sure if it's abug or intended behaviour \n\nWhen an object is set to wireframe ( object properties , viewport display ) , when overlays is turned off , the object is not shown ", "AutoDepth ignores texture transparency\nOperating system: Win 10 64-bit\nGraphics card: 980 Ti\n\nBroken: 2.93.4 LTS, 3.0 alpha (including current latest build)\nWorked: 2.79\n\nWhen navigating using mouse with auto-depth enabled, the full-transparent parts of textures should be ignored when picking the depth.\nThis worked correctly in 2.79, but 2.93 and the latest 3.0 alpha seem to ignore this.\n\n\nEnable Auto-Depth, create an object, assign it a material with some full-transparent texture, then try to zoom / pan using mouse with the cursor on it.\n[autodepth-transparent-issue.blend](autodepth-transparent-issue.blend)", "Invert Masking behaviour for Sculpt/Paint Mode\nThis is a design related to the implementation of Paint Mode.\nIt outlines the possibility to invert the behavior & visualization of masks.\n**Current** = Masked vertices cannot be edited\n**Proposed** = Masks & Selections are the same -> They can be edited.\n\n# Current Issues\n\nThere are a few different issues that exist the with current way of how masks work.\n\n### An Inverted Workflow\n\nA typical workflow issue when using masks is that users mask a section of their mesh and then invert the mask.\nIn a lot of cases the areas you mask are the ones you DO want to edit.\nMasking is very much designed with this workflow in mind. But this leads to using `Invert` on the mask potentially hundreds of times per day.\n\n### Syncing Selections across Modes\n\nFor the implementation of Paint Mode it will be very beneficial to start syncing selections between various modes. \nSo if a selection is made in Edit Mode it will also appear as masked in Sculpt/Paint Mode (This would likely be a direct conversion every time you switch modes)\nThis will be less confusing and disconnected if both the selection in Edit Mode and the mask in Sculpt/Paint Mode are areas you CAN affect.\n\n### Masks vs Selections\n\nA current plan for the implementation of Paint Mode and the new Curves object type is to use \"Selections\". \nThe use case for sculpt mode masks and selections is essentially the same: \n**You define a temporary area that should be affected by your operations.**\n\nThe issue is that \"Masks\" also refer to the concept of for example creating an attribute (like a vertex group or color attribute)\n or a grayscale image texture and using it as a factor for shader/geometry nodes. The use case here is quite different:\n**You save an area for repeated use in materials and procedural networks.**\nFor example layering textures.\n\nReplacing the current concept of sculpt mode masks with selections would avoid confusion and make the use case of selections and masks very clear.\n\n# Proposal\n\n### Selection instead of Masking\n\nWe need to test the following behavior:\n- Rename \"Mask\" to \"Selection\" or \"Selection Mask\"\n- Unselected areas are darkened with an overlay\n- Selected areas are affected by brushes & other operations\n\nThere are use cases that are important to keep in mind with this change:\n- When entering sculpt mode, everything should already be selected.\n- Previous workflows (Mask brush) to create a deselected area and should still be accessible.\n\nOther areas of Blender will still refer to \"Masks\" like a Texture Mask in the brush settings, mask nodes, grease pencil layer masks and motion tracking masks.\n\n### Offer a preference to use the old behavior\n\nThere should be an option in the preferences to invert the selection behavior globally in sculpt/paint mode and get the old behavior back.\nTo what extend the inverted behavior would be done needs to be tested.", "Smear tool painting vertex colors corrupts color info: part 2\nvertex colors get corrupted under certain blend mode combinations. 3.0.1 master.\n\nusing the smear tool over blue colors introduce green, magenta, yellow, gray and orange not present inside the blue ranges.\n\npainting with blue results in grays instead. \n\ni had this problem in 3.2 alpha using sculpt vertex colors, but when i got back to a project in 3.0.1 it happened with vanilla vertex colors tool. \n\nincluded a file this time\n\n[vertex colors bug.blend](vertex_colors_bug.blend)\n\n\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1650/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 511.79\n\nBroken: version: 3.0.1\nWorked: (newest version of Blender that worked as expected) 2.93\n\nSmear tool painting vertex colors corrupts color info\n\nuse paint brush with blue color. paints gray instead.\nuse smear brush. it should smear blues and not oranges and greens.\n", "viewport shading > color > texture, textures not affected by transformation\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2060 SUPER/PCIe/SSE2 NVIDIA Corporation 4.6.0 NVIDIA 536.99\n\nBroken: version: 4.0.0 Alpha\n\n\nViewport solid mode color option \"Texture\" does not take mapping nodes into account.\n\n---------\n\nOriginal description:\n\nwhen switching to texture in viewport shading color properties, the image textures are only affected by UV editor and not by texture coordinates node, in My opinion this texture preview is much more faster and better than EEVEE, if only you can ditch EEVEE and make the workbench preview the main material preview that support all nodes, If you notice Cinema 4D ,Max and Maya, they all use a material preview similar to this Blender Workbench preview, I don't know if you noticed these days almost everyone is saying that they prefer previewing and editing their material in cycles rather than EEvee because its way more faster to load, I tested in a 2k scene cycles took a couple of seconds switching to render view on CPU, Eevee took 12 Mins to load all textures, I hope Eevee next solves this issue if not, I would really hope that you consider developing the workbench preview \n\n?\n\n", "Sculpt Mode Painting\nThis task outlines the remaining To Do's for fundamental sculpt mode painting.\n\n### Color Swapping\n\n- `X` for switching primary and secondary color\n- The current shortcut for `X` for the draw brush will be removed.\n\nIn one of the upcoming releases all brush shortcuts in sculpt mode will be removed anyway, for maximum user customization.\n\n### Canvas Picker & Viewport Color\n\nSince painting is something optional in sculpt mode we need to make sure that:\n- Color Attribute shading isn't enforced\n- Color attributes will be displayed when using painting tools\n- A fast way of choosing color attributes in the UI\n- An optional & discoverable way of switching the viewport shading for all objects (optimal painting scenario)\n\nThe proposal for this is:\n- Use [D16241](D16241) canvas picker UI in sculpt mode. Always visible but grayed out if viewport shading color is not `Attribute`\n- Whenever a painting tool is active, the canvas picker is no longer grayed out and the color attribute viewport shading is enforced in workbench shading on the active object only\n- In the canvas picker popover, add a message or button to switch the viewport color to Attribute if this is not yet the case, so that colors stay visible between brushes/tools. \n- Remove the current behavior of automatically switching the viewport color when a painting tool is used\n\nTo make sure the sculpting experience is already taking color attributes into account, the default sculpting template and workspace should use color attributes for viewport shading.", "Bug with environment texture\nOperating system: Windows 10\nGraphics card: Nvidia GeForce RTX 3050 Laptop GPU\n\n3.6.2\n\nAfter I checked and unchecked \"Open Shading Language\" checkbox Blender crashed, after that I opened my project and turned rendered viewport the environment texture become a one solid color, and as I was rotating my camera it was changing the color.\n\nBased on an attached .blend file you can go to the Render viewport and take a look at this error.\n\n", "Collada import/export: transparent textures don't work\nOperating system: Linux\n\nBroken: blender-3.3.0-alpha+master.dfb8c90324c0, probably all versions since 2.8\nWorked: 2.79\n\nCollada cannot import or export transparent (alpha) textures.\n\nImport this .dae\n\n[BlenderLogoDae.zip](BlenderLogoDae.zip)\n\nThe result (with 2.79 for comparison).\n\n\n\nManually hook up the alpha in the node graph and export. Open the .dae in a text editor and confirm there is no <transparent> element (which sets the alpha texture). You can confirm exporting works in 2.79 by exporting and reimporting.\n\n**Other**\nNote that OpenCollada's parsing of transparency is weird and buggy, see [D11259](D11259) and 651.", "4090 has problem with texture in render preview\nOperating system: Windows 11\nGraphics card: Nvidia Geforce 4090\n\nBroken: (3.4 and 3.5.1)\nWorked: none.\n\n**Short description of error** \nThis only started recently, has been fine up till today. My preferences are set to use both the CPU (AMD Threadripper 32) and the GPU (geforce 4090) in OptiX (or CUDA, doesnt seem to matter). Once the file opens, material preview works fine. (refer to \"materialpreview.png\") but when i click render preview (refer to renderprevier.png) all i see is my rig. Anything with a texture is a similar dark grey as the default blender background. I can move the camera around fine, but no textures are visible. If i click material preview again to switch back, blender freezes and must \"end task\" in task manager. If i turn off the GPU assistance and only have the CPU enabled, everything works as it should. \n\nYesterday i updated to latest windows 11 update. I also updated my gameready drivers for my 4090. When the issue started, i thought it might have something to do with gameready drivers so switched to Studio drivers. This did not fix it. \nAssuming that these updates may have broken my file, i rolled back windows 11 drivers to most recent previous version, and did the same with the nvidia studio drivers (uninstalled 5.35.... installed 5.31). This fixed nothing.\n\nI would love to know why my GPU can't be used with Blender suddenly. Games work fine. Furmark 3d app works fine, no issues. Heat is fine. Very frustrating to have such expensive gear suddenly stop working.\n\n", "No preview icon in the material list\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1060 6GB/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 536.40\n\nBroken: version: 3.6.1\nWorked: (unknown)\n\nI noticed that for this material in the list it is not previewed, but in the \"Preview\" panel the preview is correct.\nThis is a simple setup of mixing 2 shaders using the \"Alpha\" attribute as a factor.\nThis situation makes it difficult to work when only using the drop-down list of materials in various situations..\n\nThx.\n\n", "Allow user to change the shading parameters of the asset preview generation scene\nUsers should be able to have some control over the render settings of generated asset preview icons.\n\nPersonally, I want to be able to turn off the specular lighting because it makes my assets look ugly in the previews (note the big shiny surfaces on these bushes).\n\n", "Cycles: sRGB texture compression happens after interpolation \nBroken: 2.81\n\nCycles does sRGB->linear texture conversion after texture interpolation instead of before. This leads to incorrect results.\n\nExample scene is attached. Two squares with a fine checkerboard texture, one with linear interpolation, the other with closest interpolation.\n\nExpected result: Both textures should render to an identical shade of gray. When switching the texture color space to \"linear\" instead of \"sRGB\", the result should not change as the texture contains only pure black and white pixels.\nActual result: The textures show different shades of gray. Switching the color space changes the result.\n[checkerboard.blend](checkerboard.blend)\n\n", "Workbench/Solid : Texture interpolation not displaying properly in viewport (shared across all users of an image)\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 531.18\n\nBroken: version: 3.5.0\nWorked: --\n\nLow resolution textures are no longer being correctly displayed as a sharp graphic in the viewport when the texture interpolation is set to \"closest\" in the material shader.\n\nfeature works as intended in rendered view so long as Eevee or Cycles is selected. Workbench does not work as intended in rendered view.\n\nplace object into scene. Locate a low resolution asset such as pixel art and map the texture to the object. \n\ntextures have always appeared blurry at first in the viewport but common pixel art practice is to set the interpolation from \"linear\"(or whichever is set to default) to \"closest\"\n\nWe have 4 materials each having individual Image Texture nodes (but all using the same image).\n\nIf you take the example file, and set Interpolation to Closest on all of them, you'll get it working in the viewport as well.\n\n", "White Noise Glitch using Value Attribute of Particle Instances\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 970/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.71\n\nBroken: version: 2.93.0 Alpha\nWorked: (?)\n\nThe White Noise Texture (in 1D) does not display as expected using the Value Attribute of Particle Instances in both EEVEE and Cycles. Instead of showing a single random value or color on the surface of an instance there's a fuzzy mess of different values similar to the look of overlapping faces:\n\nThe issue also appears on non flat geometry not only on planes. The using the Index attribute shows the same result. It's the same on the current stable release (that being 2.91.2) and previous releases. I tried the same setup with a Noise Texture and it seems to work just fine:\n\n- Choose any object as the emitter of a particle hair system.\n- Adjust the settings of the particle system to display the desired quantity of hairs\n- Add another object and add a Particle Instance Modifier to it\n - under 'Object' set the emitter object\n - check 'Create Along Paths'\n - in the object data properties under vertex color create a new vertex color\n - choose the same vertex color as the 'Value' data of the particle instance modifier under 'Layers'\n- use the created data as the name in the Attribute Node\n- connect any output of the Attribute Node to a White Noise Node set to '1D'\n- display the results\n(same last 5 steps for Index Attribute)\n[ParticleInstanc_ValueAttributeBug.blend](ParticleInstanc_ValueAttributeBug.blend)\n\nThank you so much for looking into it! And keep up the great work." ]

[ "Switching to eevee turns all materials into white\nOperating system: Windows-10-10.0.18362 64 Bits\nGraphics card: GeForce MX250/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 442.19\n\nBroken: version: 2.81 (sub 16)\nWorked: (optional)\n\n[When switching to eevee, My materials turns white. Cycles is Fine]\n\n[My current version is 2.81 because I switched from 2.82. blender 2.82 also has the same issue when i turn on eevee it turns all my materials to white. \n\nI'm not sure but this is what I observed, When I try to switch on cycles and eevee for like 15mins when I'm doing a project It's all fine. But when I close blender and get back again, Switching to eevee will take all my materials to white. \n\nHope this can be fix if it is a bug.\n]\n\n", "Black material + render view with heavy scene + fog\nOperating system: Windows 10\nGraphics card: 1080 ti\n\nBroken: 2.82.7\nWorked: 2.82.7\n\n* Scene was working normally when the file was actively being put together, and renders were already made.\n* Scene now goes black when put into material preview or render view mode (eevee).\n\n\n* If the large cube covering the entire scene with fog material is deleted, it leaves behind this:\n\n\nSo far, I have tried:\n\n* Imported all collections into a new file (black).\n* Imported object per object until I got black (volume was seemingly the culprit).\n* Imported only the volume shader into a new file (looked fine).\n* Opened the original file again (everything looked fine).\n* Next day starting the PC, the file is once again black when openend.\n\nSeems to only happen in this very large file, and though it doesn't contain anything sensitive, it was made at work, so I can't just publicly post it. If a developer wants to look at it, please contact me in PM here or on devtalk/BA and I will send you the file.", "EEVEE scene with many instances opens with huge white glare, or black materials\nOperating system: Windows-10-10.0.17134-SP0 64 Bits\nGraphics card: GeForce GTX 1080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 436.30\n\nBroken: version: 2.82 (sub 7)\nWorked: (optional)\n\nOpen blender scene, blender scene shows blinding white light that fills the screen in some systems or black materials in another.\n\n- Open blender scene provided,\n- go into viewport rendering mode,\n- scene should glow bright white light or with black materials.\n\nIf you exclude main collection from view layer, then enable it again, the light should go away.\nAlso note that if you exclude, then enable it again, then revert the scene or re-open the scene, the error will not happen again. It only happens when blender if first opened. \n\n[eevee_white_scene.blend](eevee_white_scene.blend)\n\n" ]

[ "Video Editing: Strip Offset Start/End not consistent with adjusting the strip handles\nOperating system: Win 10, 64bit\nGraphics card: no (CPU)\n\nBroken: (v 2.82a) and also (v 2.83 Alpha, Date 2020-03-18 17:16, Hash c9c08dc7c873)\n\nFrom the Manual:\nstrip.html\nStrip Offset Start/End [...] This is the same as adjusting the strip handles.\n\n\nHowever currently it is NOT really the same, which adds a lot of confusion to the already complicated concept that includes several offsets:\n\n\n\n**Detailed description of error**\n- Good: Extend the Strip using Handles: The Strip gets a helpful color code. The \"animated\" part is colored. The extended \"NOT animated / last frame repeated\" part is transparent.\n- Bug: Extend the Strip using Strip Offset: \"Animated\" and extended parts of the Strip get the same color.\n\n- Good: Shorten the Strip using Handles: Strip Offset is correctly calculated.\n- Bug: Extend the Strip using the Handles (or Extend *Keyboard E* or Slide *Keyboard S*): Strip Offset is incorrectly reset to 0.\n\n- Good: Handles: The Strip can not be shortened past the last \"animated\" frame.\n- Bug: Strip Offset: The Strip can be shortened, until no \"animated\" frame is left, and even further until the Duration becomes negative.\n- Bug: Using Slide *Keyboard S*, the \"animated\" frames can be moved completely outside the Strip. At least one \"animated\" frame should remain inside the Strip.\n\n**To further improve clarity, I would**\n- rename **Hold Offset Start/End** to **Trim Source Start/End**\n- change **Hold Offset Start/End tooltip** to \"Shortens the Strip by skipping frames at the start/end of the source\"\n- add **Strip Offset Start/End tooltip** to \"Extends the trimmed strip by repeating it's first/last frame or shortens it further\"\n- make **End** editable (it should do the same, as moving the right strip handle)\n- add **End tooltip** \"X position where the strip ends\"\n- change **Duration tooltip** \"The length of this strip after applying Trim and Strip Offset\"\n\n**If you feel very courageous**\n- change **Start** so that it points to the start of the Strip (currently it points to the first \"animated\" frame)\nI think, Blender heavily depends on **Start**, while it happily ignores Strip Offset Start and Strip Offset End. So, changing the internal meaning of **Start** likely makes a huge mess. However, maybe you can introduce and display a new variable \"display start\" or \"strip start\" instead, that does the trick, without messing too much with the internals.\n\nHappy Coding!", "Cycles: Normal Edit modifier after displace modifier does not with motion blur (Eevee seems fine)\nTried on multiple systems\n\nBlender 3.0 (And the previous beta builds)\n\nThe normal edit mode will not work with motion blur enabled\n\nSee attached file - Using a NormalEdit modifier looks correct in viewport then fails at render time if motion blur is used in combination with the attached Geometry Nodes tree that is displacing the main geometry over time. \nWith the node tree disabled the NormalEdit modifier works even with motion blur.\n\n**Viewport Result:**\n\n\n**Rendered Result:**\n\n\n\n**Demo file:**\n[Bug (1).blend](Bug__1_.blend)\n[#93700 Cycles Normals Motion Blur Displace.blend](T93700_Cycles_Normals_Motion_Blur_Displace.blend)", "Eevee camera clipping inconsistent when using child of constraint on an armature.\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 471.41\n\nBroken: version: 2.83.15\nversion: 2.83.17\nversion: 2.83.18 Release Candidate\nversion: 2.93.2\nversion: 2.93.3 Release Candidate\nversion: 3.0.0 Alpha\n\nViewport's material preview and viewport's render has different camera cull than F12's render image, despite same clipping settings\n[2021-08-15_12-16-50.mp4](2021-08-15_12-16-50.mp4)\n\n[CullBugSimplified.blend](CullBugSimplified.blend)\nVerify viewport is in ACamera's view\nSwitch to viewport rendered and material preview to verify cubes on foreground are visible\nRender to verify cubes in foreground have disappeared\nSwitch to ACamera.001 and render to verify cubes in foreground are consistent with viewport\n\n", "Cycles: Motion blur artifacts\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 472.88\n\nBroken: version: 3.2.2\n\n\n**description of error**\nWhen rendering an animated rigged model on cycle **with motion blur** ; if the model has **auto smooth enable AND a subdivision modifier**, then it is possible that strange artefact will appear during the render. (doesn't happen in the viewport) \n- This bug do not happen if auto smooth is disable or doesn't have a subdivision modifier. \n- The bug also only appear during some specfic poses. So this is mainly noticiable when rendering an animation.\n- Those artefact often appear near sharp edge.\n- Only happen on Cycle, Eevee render seem fine.\n- [Edit] changing the shutter values of the blur will sometime remove the bug on some frames OR make the bug appear on frame that was fine before.\n\nHere 2 Render exemple with and without autosmooth , both has motion blur enabled :\n \nAnd here an animation where you can notice the bug.\n[Animated0035-0318.mp4](Animated0035-0318.mp4)\n\n\n[MotionBlurCycleArtefact.blend](MotionBlurCycleArtefact.blend)\nIn this blender file, rendering the frame 302 will make the bug happen. \n- If you disable auto smooth on the model (the main body and not the cloth) then the render will be fine.\n- If you disable the subdivision modifier on the model then the render will be fine.\n- If you disable the subdivision modifier AND the auto smooth on the model then the render will be fine.\n\nThank\n", "Multi-Object-Mode: EditMesh Tools\nNote, while these follow menus, duplicates have been removed.\n\n## Remaining\n\n- MESH_OT_select_similar_region ([D3672](D3672))\n- MESH_OT_knife_project (leave aside for now, too complex - see [D3637](D3637))\n- MESH_OT_knife_tool (leave aside for now, too complex - see [D3637](D3637))\n\n\n## Mesh Menu:\n\n- ~~MESH_OT_duplicate_move~~ 1ade071052\n- ~~MESH_OT_extrude_edges_indiv~~ d8f70a26fd\n- ~~MESH_OT_extrude_faces_indiv~~ 07774d4860\n- ~~MESH_OT_extrude_verts_indiv~~ 7c415e6fd7\n- ~~MESH_OT_extrude_edges_move~~ 1d7bdbd273\n- ~~MESH_OT_extrude_vertices_move~~ 1d7bdbd273\n- ~~MESH_OT_delete_loose~~ bfc9d426bb\n- ~~MESH_OT_decimate~~ 5b45d32f3a\n- ~~MESH_OT_dissolve_degenerate~~ (initial comsmit) bfc9d426bb\n- ~~MESH_OT_dissolve_limited~~ 09e14cb5bc\n- ~~MESH_OT_face_make_planar~~ 0e5a4f927a\n- ~~MESH_OT_vert_connect_nonplanar~~ b942106c73\n- ~~MESH_OT_vert_connect_concave~~4043e4b820\n- ~~MESH_OT_fill_holes~~ 69b95b45f4\n- ~~MESH_OT_symmetrize~~ 03a916e5b5\n- ~~MESH_OT_symmetry_snap~~ 7f9768cb80\n- ~~MESH_OT_bisect~~ f9eb2f29fe\n- ~~MESH_OT_sort_elements~~ ffb424c85f\n\n## Snap Menu:\n\n\n- ~~VIEW3D_OT_snap_selected_to_grid~~ 016ba0f38b\n- ~~VIEW3D_OT_snap_selected_to_cursor~~ ec3357e03a\n- ~~VIEW3D_OT_snap_selected_to_active~~ ec3357e03a\n- ~~VIEW3D_OT_snap_cursor_to_selected~~ ec3357e03a\n- ~~VIEW3D_OT_snap_cursor_to_center~~ (no change required)\n- ~~VIEW3D_OT_snap_cursor_to_grid~~ (no change required)\n- ~~VIEW3D_OT_snap_cursor_to_active~~ (no change required)\n\n\n## Vertex Menu:\n\n- ~~MESH_OT_merge~~ b73c7381cc\n- ~~MESH_OT_remove_doubles~~ 22b2bab702\n- ~~MESH_OT_rip~~ 9d0d7c375e\n- ~~MESH_OT_rip_edge~~ 6539c1fc41\n- ~~MESH_OT_split~~ fc8d903000\n- ~~MESH_OT_separate~~ 5e915baec4\n- ~~MESH_OT_mark_sharp~~ 0ac3d5f7db\n- ~~MESH_OT_vertices_smooth~~ d915aa57f8\n- ~~MESH_OT_vertices_smooth_laplacian~~ 1f30232aa7\n- ~~MESH_OT_blend_from_shape~~ 6ae09dfcad\n- ~~OBJECT_OT_vertex_group_smooth~~ 3dee8b4a12\n- ~~MESH_OT_shape_propagate_to_all~~ 3a43528274\n\n## Edge Menu:\n\n- ~~MESH_OT_edge_face_add~~ de2d642c84\n- ~~MESH_OT_subdivide~~ (initial patch)\n- ~~MESH_OT_subdivide_edgering~~ 853e55b043\n- ~~MESH_OT_unsubdivide~~ e245f7c09c\n- ~~TRANSFORM_OT_edge_crease~~ (initial patch)\n- ~~MESH_OT_mark_seam~~ 61f0608b4a\n- ~~MESH_OT_mark_sharp~~ 0ac3d5f7db\n- ~~MESH_OT_edge_rotate~~ 66ffbf9b7d\n- ~~MESH_OT_edge_split~~ 207b549d40\n- ~~MESH_OT_bridge_edge_loops~~ eb2226dde2\n- ~~TRANSFORM_OT_edge_slide~~ (wrapper to other operator, working)\n- ~~MESH_OT_loop_multi_select~~ df49155cfb\n- ~~MESH_OT_loop_to_region~~ 8b41999ea0\n- ~~MESH_OT_region_to_loop~~ 56d65719b3\n\n\n## Face Menu:\n\n- ~~MESH_OT_flip_normals~~ 4704f2687a\n- ~~MESH_OT_edge_face_add~~ de2d642c84\n- ~~MESH_OT_fill~~ b726ff4fc4\n- ~~MESH_OT_beautify_fill~~ e42b8c32ab\n- ~~MESH_OT_inset~~ 570455fb83\n- ~~MESH_OT_bevel~~ d9e04cb594\n- ~~MESH_OT_solidify~~ f309c34cfe\n- ~~MESH_OT_intersect~~ dffd37877d\n- ~~MESH_OT_intersect_boolean~~ 46aec45b2a\n- ~~MESH_OT_wireframe~~ a0a78f6da4\n- ~~MESH_OT_poke~~ 3a9669bfff\n- ~~MESH_OT_quads_convert_to_tris~~ (initial patch bfc9d426bb\n- ~~MESH_OT_tris_convert_to_quads~~ (initial patch bfc9d426bb\n- ~~MESH_OT_face_split_by_edges~~ b6a822d726\n- ~~MESH_OT_faces_shade_smooth~~ e4c4e8566f\n- ~~MESH_OT_faces_shade_flat~~ 9fc0c0c3de\n- ~~MESH_OT_uvs_reverse~~ 63951bc987\n- ~~MESH_OT_uvs_rotate~~ faaffc4011\n- ~~MESH_OT_colors_rotate~~ 42c5c20e54\n- ~~MESH_OT_colors_reverse~~ 753be0a2a5\n\n## Mesh Toolbar: (should be added to menu)\n\n- ~~MESH_OT_loopcut~~ e87dd9aa00\n- ~~MESH_OT_convex_hull~~ e1cee0ac71\n- ~~MESH_OT_delete~~ (initial patch bfc9d426bb)\n- ~~MESH_OT_delete_edgeloop~~ 49be803ad6\n- ~~MESH_OT_dissolve_edges~~ 7f5c469716\n- ~~MESH_OT_dissolve_faces~~ 848d9cda0e\n- ~~MESH_OT_dissolve_mode~~ (wrapper of vert/edge/face dissolve)\n- ~~MESH_OT_dissolve_verts~~ 5f59a8bb11\n- ~~MESH_OT_drop_named_image~~ b4d23818c8\n- ~~MESH_OT_dupli_extrude_cursor~~ 68a719f1ea\n- ~~MESH_OT_duplicate~~ 1ade071052\n- ~~MESH_OT_edge_collapse~~ 2349b9777a\n- ~~MESH_OT_edgering_select~~ 7fa77cd31d\n- ~~MESH_OT_edges_select_sharp~~ dd93c80747\n- ~~MESH_OT_extrude_region~~ 1d7bdbd273\n- ~~MESH_OT_extrude_repeat~~ 9a125ca5fe\n- ~~MESH_OT_faces_select_linked_flat~~ 1bb9ccf887\n- ~~MESH_OT_fill_grid~~ 1cbe3b6e03\n- ~~MESH_OT_hide~~ 5d8937e8ba\n- ~~MESH_OT_loop_select~~ 7fa77cd31d\n- ~~MESH_OT_normals_make_consistent~~ 266638d783\n- ~~MESH_OT_offset_edge_loops~~ efd843b51c\n- ~~MESH_OT_reveal~~ f4c1685a16\n- ~~MESH_OT_screw~~ badd88ca02\n- ~~MESH_OT_select_all~~ (initial patch bfc9d426bb)\n- ~~MESH_OT_select_axis~~ c899f21800\n- ~~MESH_OT_select_face_by_sides~~ 90c9458b2f\n- ~~MESH_OT_select_interior_faces~~ e161c51e7f\n- ~~MESH_OT_select_less~~ b6a7fa1880\n- ~~MESH_OT_select_linked~~ (initial commit) bfc9d426bb\n- ~~MESH_OT_select_linked_pick~~ (initial commit) bfc9d426bb\n- ~~MESH_OT_select_loose~~ eacb2eb423\n- ~~MESH_OT_select_mirror~~ 3ebe389605\n- ~~MESH_OT_select_mode~~ (initial patch bfc9d426bb)\n- ~~MESH_OT_select_more~~ 2309131b4d\n- ~~MESH_OT_select_non_manifold~~ 4da2aec2bb\n- ~~MESH_OT_select_nth~~ 63c2397171\n- ~~MESH_OT_select_random~~ 847f028b71\n- ~~MESH_OT_select_similar~~ 75b2091d42\n- ~~MESH_OT_select_ungrouped~~ 1d8e6c5cd0\n- ~~MESH_OT_shortest_path_pick~~ c32cc3e43e\n- ~~MESH_OT_shortest_path_select~~ 5a431be629\n- ~~MESH_OT_spin~~ ab8dded330\n- ~~MESH_OT_vert_connect~~ 944054fbb6\n- ~~MESH_OT_vert_connect_path~~ fc7316fb54\n", "Use toolbar for Clip Editor\nAs discussed with @Sergey, here's how we would like the Clip Editor to work:\n\n{[F6943021](Screenshot_2019-04-14_at_10.06.22.png), size=full}\n\nThis makes it compliant with the other editors in Blender 2.8.\n\n\n - - [x] {icon circle color=yellow} Add toolbar to the Clip Editor\n - - [x] {icon circle color=yellow} Add Marker tool to the toolbar\n - The Marker tool lets you add markers whenever you click. The tool settings for the next marker should then be added to the tool settings\n - - [x] {icon circle color=yellow} Add Selection tools to the toolbar\n - - [ ] {icon circle color=yellow} Add Cursor tool to the toolbar\n - - [ ] {icon circle color=yellow} Add transform tools (Move, Rotate, Scale, Transform) to the toolbar\n - - [x] {icon circle color=yellow} In Mask mode, add the masking tools to the toolbar", "imported scenes in video sequencer, don't use the scenes own dimensions\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3090/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 536.23\n\nBroken: version: 3.6.0\n\nmaking a video editing scene, and importing scene clips from different sizes all clips will have the same dimensions or bounding box despite them not actually being the same size\n\nLook at the attached blend. When selecting the scene 1 clip with the monkey, you can see that it's bounding box is much bigger than the actual clip. Which makes it just a bit harder to align correctly. And regardless of impact it seems wrong\n\n", "Application handler not executed when video sequencer contains a strip\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce RTX 3080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 462.59\n\nBroken: version: 2.93.4\n\nA registered handler for frame_change_post does not get executed while rendering an animation when there is a strip present in the video sequencer.\n\n1. Open the example file and turn on the system console\n2. Execute the example script present in the file\n3. Render an animation, note how you can see the handler getting called before each rendered frame in the console\n4. Add a strip to the video sequencer (for example a solid color)\n5. Render another animation, note how this time the handler doesn't get called per frame anymore\n6. Remove the strip again\n7. Render once more, note how it works again as expected\n\n**Example file**\n[bug_example.blend](bug_example.blend)\n", "Transform Tools influence on Hidden Face Sets with Multires Modifier\nOperating system: Linux-5.3.0-62-generic-x86_64-with-debian-buster-sid 64 Bits\nGraphics card: Quadro GP100/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 440.100\n\nBroken: version: 2.91.0 Alpha\n\nThere seem to be multiple strange effects that the Transfrom Tools (Move, Rotate, Scale, Transform) have on an object with Multires in Sculpt Mode.\nWhen hiding face sets the boundaries are still being effected but are lagging behind.\nIf the operation is cancelled (Esc) the boundary edges will move half of the way anyway.\nIf the operation is accepted it will unhide all face sets.\n\n[2020-07-27 14-06-58.mp4](2020-07-27_14-06-58.mp4)\n\n- In the default cube, add the Multires modifier\n- Subdivided the mesh a few times.\n- In sculpt mode, draw a Face Set for one of the faces.\n- Hide the rest\n- Enable the Transform tool and move the object.\n", "Add In Front/Visible option to lasso and box mask tools\nCurrently Box Mask and Lasso Mask create the mask through the object, which is not useful in some cases. There should be an in front option to change this behavior. \n\nImplementing this by checking the vertex normal and view direction is trivial, but it won't work as expected on models with occluded vertices facing toward the view. We should evaluate if it is possible to implement vertex occlusion test in sculpt mode. If it is not, then we can use the vertex normal implementation. ", "Invert Masking behaviour for Sculpt/Paint Mode\nThis is a design related to the implementation of Paint Mode.\nIt outlines the possibility to invert the behavior & visualization of masks.\n**Current** = Masked vertices cannot be edited\n**Proposed** = Masks & Selections are the same -> They can be edited.\n\n# Current Issues\n\nThere are a few different issues that exist the with current way of how masks work.\n\n### An Inverted Workflow\n\nA typical workflow issue when using masks is that users mask a section of their mesh and then invert the mask.\nIn a lot of cases the areas you mask are the ones you DO want to edit.\nMasking is very much designed with this workflow in mind. But this leads to using `Invert` on the mask potentially hundreds of times per day.\n\n### Syncing Selections across Modes\n\nFor the implementation of Paint Mode it will be very beneficial to start syncing selections between various modes. \nSo if a selection is made in Edit Mode it will also appear as masked in Sculpt/Paint Mode (This would likely be a direct conversion every time you switch modes)\nThis will be less confusing and disconnected if both the selection in Edit Mode and the mask in Sculpt/Paint Mode are areas you CAN affect.\n\n### Masks vs Selections\n\nA current plan for the implementation of Paint Mode and the new Curves object type is to use \"Selections\". \nThe use case for sculpt mode masks and selections is essentially the same: \n**You define a temporary area that should be affected by your operations.**\n\nThe issue is that \"Masks\" also refer to the concept of for example creating an attribute (like a vertex group or color attribute)\n or a grayscale image texture and using it as a factor for shader/geometry nodes. The use case here is quite different:\n**You save an area for repeated use in materials and procedural networks.**\nFor example layering textures.\n\nReplacing the current concept of sculpt mode masks with selections would avoid confusion and make the use case of selections and masks very clear.\n\n# Proposal\n\n### Selection instead of Masking\n\nWe need to test the following behavior:\n- Rename \"Mask\" to \"Selection\" or \"Selection Mask\"\n- Unselected areas are darkened with an overlay\n- Selected areas are affected by brushes & other operations\n\nThere are use cases that are important to keep in mind with this change:\n- When entering sculpt mode, everything should already be selected.\n- Previous workflows (Mask brush) to create a deselected area and should still be accessible.\n\nOther areas of Blender will still refer to \"Masks\" like a Texture Mask in the brush settings, mask nodes, grease pencil layer masks and motion tracking masks.\n\n### Offer a preference to use the old behavior\n\nThere should be an option in the preferences to invert the selection behavior globally in sculpt/paint mode and get the old behavior back.\nTo what extend the inverted behavior would be done needs to be tested.", "Grease pencil. Mask not working when masking layer has more than one keyframe\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce RTX 2060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 457.30\n\nBroken: version: 2.93.0 Alpha\n\nWhen my masking layer has only one keyframe, masked layers display as expected. When my masking layer has more than one keyframe expected masking doesn't happen.\n\n* open .blend (the mask layer has two keyframes, both having the same filled shape)\n* expected result: the strokes on the 'marks' layer that intersect the shape on the mask layer are always visible\n* actual result: the strokes on the 'marks' layer that intersect the shape on the mask layer are only visible after i delete the second keyframe on the 'mask' layer\n\nWhile dragging the timeline play head the marks layer appears as expected, correctly masked. But when i release the mouse button the marks layer disappears entirely again.\n\n[isthisamaskbug.blend](isthisamaskbug.blend)", "Regression: Armatures [in front] overlay doesn't update properly (toggling overlay ON and OFF)\nOperating system: Linux-5.15.0-52-generic-x86_64-with-glibc2.31 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 470.103.01\n\nBroken: version: 3.4.0 Alpha\nBroken: 2.92.0\nWorked: 2.91.2\n\nBisecting gives 0f95f51361 as the culprit commit\n\nif the armature viewport mode is set to \"in front\" it doesn't properly update when the overlays are turned off and on\n[Peek 2022-11-01 05-35.mp4](Peek_2022-11-01_05-35.mp4)\n\n1-add an armature\n2-set the \"viewport display\" to in front\n3-turn off and on overlays\n4-if you can confirm the bug the armature should appear after changing anything in the 3d view to update as in the video, move rotate or zoom \n\n", "render lagging behind tiles\nOperating system: win10\nGraphics card:rts 2070 super\n\n2.81\nwhen rendering, the picture doesnt appear for a while after the render tile already finished rendering that region, not sure if its a bug. didn't see anything like this online. here is a video\n\n\n[bug (2).blend](bug__2_.blend)\n\n- Open file and render\n\n[gnVknluHJZ.mp4](gnVknluHJZ.mp4)", "Sculpt Paint: Mask by color produces an empty mask with a threshold of 0.0\nOperating system: Windows-10-10.0.19043-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 512.95\n\nBroken: version: 3.2.0\n\nIt is impossible to make a precise color selection from two similar colors with Mask by Color alone, because a threshold near or at 0 produces an empty mask. I consider this a bug because according to the comment in the source code, this functionality isn't working as intended. Quote directly from source code:\n\nD8157\n```\n/* sculpt_mask_by_color_delta_get returns values in the (0,1) range that are used to generate the\n * mask based on the diference between two colors (the active color and the color of any other\n * vertex). Ideally, a threshold of 0 should mask only the colors that are equal to the active\n * color and threshold of 1 should mask all colors. In order to avoid artifacts and produce softer\n * falloffs in the mask, the MASK_BY_COLOR_SLOPE defines the size of the transition values between\n * masked and unmasked vertices. The smaller this value is, the sharper the generated mask is going\n * to be. */\n```\n\nFor completeness, here is a real world scenario where it is tedious to mask the pants, shirt, or skin because the colors are too close together for the Mask by Color tool.  \n\n\nApply two similar colors to a subdivided surface with the paint tool, and try to mask only one of them. The only way to produce a sharp mask is with the Increase Contrast operator when the Threshold is fine tuned to work with it.\n[blender_3Yh12TUXry.mp4](blender_3Yh12TUXry.mp4)\n" ]

[ "mask does not update correctly when changing frames\nOperating system: windows 10 x64\nGraphics card: intel hd4600\n\nBroken: (version: 2.80 (sub 61), branch: blender2.7, commit date: 2019-05-12 22:25, hash: 05f1451d7248, type: Release)\nWorked: (2.79)\n\nwhen changing frames the mask animated points does not move correctly every time, sometimes it works other dont!\nplease see attached video:\n[0001-0005.mp4](0001-0005.mp4)\n\nadd any mask\nmake animation on it\ngo back and forth with KB left arrow and wright\n" ]

[ "bpy.props.IntProperty fails to register if given float values\nOperating system: Windows-10-10.0.19042-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1650 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 496.76\n\nBroken: version: 3.1.0 Alpha\n\nWhen an addon tries to register an IntProperty, whilst not giving one of the numerical values an explicit int value, the property will return an error during registration.\n\n```\nprop = bpy.props.IntProperty(default=1.0) # prop is created okay\nbpy.types.Object.intprop = prop\nTypeError: integer argument expected, got float\nTraceback (most recent call last):\n File \"<blender_console>\", line 1, in <module>\nValueError: bpy_struct \"Object\" registration error: 'intprop' IntProperty could not register (see previous error)\n```\n", "`action.frame_range` span always >=1 when there's only 1 key frame in it.\nOperating system: Mint\nGraphics card: Nvidia\n\nBroken: 3.5 & main branch\n\n\nframe_range wrong value when only 1 keyframe\n\n- with default cube, go to frame 6\n- Insert a keyframe (for example, on location)\n- bpy.context.object.animation_data.action.frame_range returns (6.0, 7.0)\n\nMy expectation was it returns (6.0, 6.0)\n\nIf you then insert a keyframe at frame 7, it still returns (6.0, 7.0)\n\n", "Properly handle file links (as in symlinks)\nWhile it should be fairly straight forward, I didn't fully port the file linking support to the new file-list design yet. Most work should already be done however.", "Cycles Volume Anisotropy artifacts with intersecting volumes\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: AMD Radeon RX 5700 XT ATI Technologies Inc. 4.5.14761 Core Profile Context 21.30.23.01 30.0.13023.1012\n(CPU rendering)\n\nBroken: version: 3.0.0\nAlso tested in 3.1\nWorked: maybe never?\n\nI have a scene in which I have a smoking volcano, clouds, and an atmosphere, which was giving me trouble for a while.\n\nAs you can see, there's a horizontal line running through the smoke column.\nBy trial and error, I found out that it wasn't there when the atmosphere, which is a cube with a volume shader on it, was disabled.\n\nI tried to recreate the setting to figure out the cause of the issue, and I **boiled it down to this**:\n*If there are two volumes intersecting with different anisotropy values, things get a little weird.*\nTake a look:\n\n| | Column 0 | Column 0.1 |\n| -- | -- | -- |\n| Atm. 0 |  | \n| Atm. 0.1 | |  |\n\nNotice the horizontal line running across the smoke column when the two anisotropy values don't match.\nThis seems to happen at a specific density where the height fallof of the atmosphere crosses a certain threshold.\n\nThings I have found that have no effect:\n- *Volume step rate*\n- *Volume max steps*\n- *Clipping planes*\n- Disabling *ray visibility* (on both the atmosphere and the smoke column) to everything but *Camera*\n\n[BrokenAnisotropyCycles.blend](BrokenAnisotropyCycles.blend)\n\n\n\nPerhaps this is worth a look?", "Writing to 8 bit integer \"enum\" attributes in geometry nodes\nThis is needed to make writing `curve_type`, `normal_mode`, and other curve attributes work properly. There are two potential approaches:\n- Allow writing `int32_t` values as `int8_t` attributes transparently.\n - This is inconsistent with the way we handle byte colors and other types currently.\n- Add a `8 Bit Integer` mode to the store named attribute node\n - This should probably be done anyway\n - It's a little less intuitive, since it doesn't \"just work\"\n - It also doesn't solve writing with the modifier output attributes\n\nThere is also the problem of clamping the written values when they are out of range of the enum.", "Material Library VX causes permanent freeze with EEVEE rendering\nOperating system: Windows 10, 64 bit\nGraphics card: Intel HD Graphics\n\nBroken: 2.81\n\nUsing materials from Material Library VX addon, particularly Floor Wood and Rusty Metal, causes EEVEE to freeze all of Blender. They work fine with Cycles, so it's not the addon. Not even the partial \"developer view\" render can take the materials.\n\nOpen default blend file.\nEnable Material Library VX addon.\nApply \"Floor Wood\" to the cube.\nSwitch to \"render view\" without changing the engine off of EEVEE. Blender 2.81 will freeze.\n\nNote: If you try to view the material preview and EEVEE is the render engine even *that* will freeze Blender 2.81.", "Show small preview image even in non-Thumbnail Display Modes\nAs mentioned in #89729 (Add toggle for Vertical and Horizontal Display Mode) we should support the Vertical and Horizontal List Display Modes as well. But rather than showing the data-block type icon there (which is rather useless), show the preview, which is much more important typically.\n\nThis may actually be nice for the regular File Browsing as well, but there it may be more important to have an icon to quickly differentiate the different file types.", "AlembicObjectPath has a fixed-size char[]\nAlembicObjectPath (see `DNA_cachefile_types.h`) has a fixed-size char- [ ] buffer. \n\n> In 3128600a8a#142412, @LazyDodo wrote:\n> if the object path can be arbitrarily large it's gonna be a matter of time before someone comes with a file with 4097 byte object path, why not just allocate the ram you need? \n\n\nI think that's a good idea, and that's what this task is for.\n", "Cannot keyframe particle count property\nOperating system: Windows-7-6.1.7601-SP1 64 Bits\nGraphics card: Quadro 2000M/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 377.83\n\n\nBroken: version: 2.80 (sub 75)\nWorked: (optional)\n\n\n clicking keyframe button next to particle count responds with *Failed to resolve path to property, try manually specifying this using a Keying Set instead*\nsee screenshot attached\n\nClick on animate property next to particle count\n", "Cleanup: remove `Scene *` from dupli-list\n`SnapObjectContext` needs a `Scene *` to pass it to the dupli-list, which only needs to pass it to the particle system, which only needs it to obtain the gravity vector.\n\nThe gravity vector should only be read from the evaluated scene, but there are cases where the original scene is passed instead. Since the depsgraph is also passed to the dupli-list (`object_duplilist()`) & simulation code (`ParticleSimulationData`), that can be used to obtain the evaluated scene instead.\n\n- rainyDay\n", "Can't get vertex color in index layer of \"Particle Instance\" modifier\nOperating system: Mac OS ventura 13.1 \nGraphics card: Apple M1\n\n\nBroken: version: 3.5.1. 3.6.1\nWorked: 2.80.75\n\n\nI define a vertex color named \"test color\" in \"Object data properties\"-\"color attributes\"-\"+\"-\"domain: vertex\"-\"data type: color\" for a general plane mesh. Then I add a \"Particle Instance\" modifier for the mesh. In the \"layers\" section of the modifier, I want to assign the previous \"test color\" to the \"index\" field. However, I get an empty list prompted when tapping the field area. And I can't enter the name manually either.\n\n1. build a plane mesh\n2. Create a vertex color. \"Object data properties\"-\"color attributes\"-\"+\"-\"domain: vertex\"-\"data type: color\" for the plane\n3. Add a \"Particle Instance\" modifier for the plane\n4. Go to \"index\" field in \"layers\" section of the modifier.\n\n", "Blender 2.83 smoke or liquid simulation not showing up when switching to Modular mode\nOperating system: Windows 10pro 64 bit\nGraphics card: Nvidia GTX 960\n\nBroken: 2.83, v2.83.0 \nWorked: 2.82\n\nWhen I'm simulating smoke or liquid. If I switch to modular mode, the bake the simulation. I do not see any smoke or liquid after it has run the bake. I can see the domain moving as if particles are filling the domain, but during play back nothing is there.\nIf I change the mode to replay mode. I can see the smoke or liquid as it is simulating. But I don't get this in modular mode\nI have a youtube video link that shows the problem. \n\nIoqbkzs7h5w\n\nI like modular mode, because baking the simulation this was takes less resources by not updating the view port.\n\n- Create a box.\n- Duplicate the box. \n- In physics tab... Make one for the emitter and the other as domain.\n- Scale domain as necessary.\n- By default Blender is in Replay mode. Change it to modular to get the Bake buttons. \n- Bake simulation. \n- Then change scale of emitter and bake simulation again.\nor\n- open attached file\n- Try see simulation with Replay (it should work)\n- Then try with Modular\n[Name_animationV2.blend](Name_animationV2.blend)", "VSE: Crash using Retime-Tool on extended Movie-Clip\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 531.79\n\nBroken: version: 4.0.0 Alpha\nWorked: N/A\n\nBlender crashes when trying to use the Retime-Tool on a Movie-Clip where the start was extended.\n\nIn the VSE\n\n1. Import a Movie-Clip (.mkv encoded with NVENC, H.264 from OBS in my case).\n2. Extend the left edge of the clip so it begins before the actual content begins (the first frame is looped up to the point the actual video starts).\n3. With the Retime-Tool selected move the right edge of the clip to change the playback-speed of the whole (extended) clip.\n4. Blender crashes\n\nVideo and blender.crash.txt are attached.\n\nWhen the clip is retimed first and then extended before its actual start (do step 3 before step 2) Blender doesn't crash and the result looks as expected.", "Image Sequence data disappears after certain properties are set\nOperating system:\nGraphics card:\n\nBroken: 2.83.6 (and all previous 2.8x versions)\n\n\nCannot access image sequence pixel data from 'Image Sequence' editor after certain properties are set.\n\nBased on the following startup file and bundled image sequence:\n\n\n1. Open the 'image_sequence_issue.blend' file from the following zip folder:\n[image_sequence_issue.zip](image_sequence_issue.zip)\n2. Run the active script in the text editor area\n3. Notice that an error is thrown - the image sequence has no data.\n4. Now comment out lines 8 & 9 (or line 12), and re-run the script. Notice that no error is thrown and the image has data.\n\nFor reference; the script bundled in the above file is copied below:\n\n```\nimport bpy\n\n# get the image editor space in the current screen\nviewer_area = next(area for area in bpy.context.screen.areas if area.type == \"IMAGE_EDITOR\")\nviewer_space = next(space for space in viewer_area.spaces if space.type == \"IMAGE_EDITOR\")\n\n# set properties on the image user\nviewer_space.image_user.frame_offset = 1\nviewer_space.image_user.use_cyclic = True\n\n# set the display channels to 'color'\nviewer_space.display_channels = \"COLOR\"\n\n# viewer_space.image.reload() # this line does not fix the issue\nprint(\"Before update:\", viewer_space.image.has_data)\nviewer_space.image.update()\nprint(\"After update: \", viewer_space.image.has_data)\n```", "VDB sequences are rendered blocky here and there ( issues with Volume Displacement Modifier )\nOperating system: Windows 10 \nGraphics card: RTX 2070 CUDA\n\nBroken: blender-3.0.0-alpha+master.ef5a362a5b24-windows.amd64-release.zip\n\n\nBlender renders these blocky frames here and there during the actual render. The VDBs seem to be fine, the same botched render frames look fine in the viewport render. These anomolies are happening only during the actual render. See the attached image (middle right image) for what is happening please.\n\nThe same sequence renders without any issues in Houdini.\n\n(Cycles)\n\n\n\n\nHere is an actual movie render (with Eevee) that shows how horrible this issue is.\n[borkedframes_0001-0200.mp4](borkedframes_0001-0200.mp4)\n\n\n[#88966.blend](T88966.blend)\n[steam.4.0001.vdb](steam.4.0001.vdb)\n\n- Open .blend file\n- Render sequence with CUDA\n\nIssue should appear within 30 frames\n" ]

[ "OpenVDB cache files have extra \"_00\" at end.\nMacBook Pro (15-inch, Mid 2009)\nProcessor: 2.8 GHz Intel Core 2 Duo\nMemory: 4 GB 1067 MHz DDR3\nGraphics: NVIDIA GeForce 9600/9400M 256 MB\n\n2.77 RC \ndate: 2016-02-26 12:32 \nHash: b594b25\nBranch: Head\n\nWhen Smoke cache files are written, there is an extra \"_00\" at the end, just before the extension. So a files are usually called something like, SmokeCache_000010_00.vdb. Up till now this hasn't been a problem. But when using OpenVDB, it causes many problems when trying to load them into other programs. At this point you must manually rename the files by hand in order to get them to load. I have tested this with Modo and Vray in 3DSMax.\n\nNo need for steps. It's Pretty obvious." ]

[ "Weird Motion Blur with Simulation Nodes\nBroken: 3.6.2\n\nMotion Blur doesn't work properly with Geometry Nodes instances.\nAs shown in the rendered video, the motion blur seems to give weird results, the motion trails of the instanced particles from the Geometry Nodes (and Simulation Nodes) setup don't seem to follow their direction properly.\n\n<video src=\"attachment\" title=\"animation.mp4\" controls></video>\n\n[Test.blend](attachment)\n- Open file\n- Hit play to bake simulation until let's say frame 45\n- Render frame 45\n\n", "Geometry Nodes: Viewer overlay disappears when enabling In Front\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 960/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.94\n\nBroken: version: 3.4.0 Alpha\nWorked: Never\n\nCaused by: c55d38f00b\n\nWhen enabling \"In Front\" in the Viewport Display options of an object, which currently has a geometry nodes viewer overlay, that overlay disappears.\n\n| Viewer object: `In Front`: OFF | Viewer object: `In Front`: ON |\n| -- | -- |\n|  |  |\n\n1. Open the attached file in 3.4 or newer versions.\n2. Make sure the viewer node is enabled in the geometry nodes of `Plane.001`.\n3. Under Viewport Display, toggle `In Front` for `Plane.001`.\n\n[bug_report.blend](bug_report.blend)\n", "mathnode socket doesn't restore when voronoi signal is turned to \nOperating system: Linux-4.15.0-66201911071317-generic-x86_64-with-debian-buster-sid 64 Bits\nGraphics card: GeForce GT 630/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 390.116\n\nBroken: version: 2.82 (sub 6)\nWorked: (optional)\n\nthe mathnode after a voronoi tex node ejects the noodle when you switch voronoi to \"n-sphere radius\" and keeps hiding any values.\n\nopen blender\ncreate a material for an object of your choice\nadd voronoi and mathnode as shown in the picture\nswitch voronoi from f1 to n-sphere radius\nnotice that the mathnode disconnenćts, which is fine, butit should re-open the slot then, right?\n\n\n\n", "Curve to Mesh node causes other objects to edge split when not shaded smooth\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2070 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 472.39\n\nBroken: version: 3.0.0 Beta\n\nWhen using a Curve to Mesh node, it causes other objects joined to it to be read with each face as a separate mesh island by Cycles (using the Random per Island socket of the Geometry node in shaders). This only happens while using Flat shading AND the Curve to Mesh is plugging into the first position of the Join Geometry node. In any other order it will not happen.\nAlso happens with more mesh objects as long as the Curve to Mesh object is the first on in the Join node.\n\nThe edges aren't actually split, it just seems to be Cycles that thinks they're individual mesh islands.\n\nAdd a grid node and join it to a curve going through a Curve to Mesh node.\nEnsure the Curve to Mesh is the first connection into the Join Geo node.\nAssign a material to the grid and in the shader, view the Random Per Island output of the Geometry node.\n\n[2021-11-06 18-08-29.mp4](2021-11-06_18-08-29.mp4)\n\n[Mesh Island Bug.blend](Mesh_Island_Bug.blend)", "Geometry Nodes socket is not showing relevant value info for current modifier\nOperating system: win 10\nGraphics card: 1050 ti\n\nBroken: 3.3.0 beta\n\nUsing many modifiers with the same GN node tree (multiple users) has some issue with data updating.\nIf you bring cursor over socket to see its value, then it shows only data from first GN modifier, no matter what modifier is selected at the moment. Seems to be only interface problem, as internally processed data is correct.\n\nIn attached file bring cursor over orange socket (Object type) and it will show same object for both modifiers, when in fact they are different.\n[socket_info_issue.blend](socket_info_issue.blend)", "Sculpting laggy and weak from some directions.\nOperating system: Windows 10\nGraphics card: Nvidia GTX 970\n\nBroken: 3.6.1\nWorked:Never(2.80+)\n\nWhen i try to sculpt from from below, using radius unit: scene, sculptings get laggy and the stroke is weak.\n\nI'm trying to sculpt stones for a wall in place.\nThe sculpting works fine if i first move the stone i'm working on to the scene origin. This adds a lot of work as i want to for example put bevels on all objects first, then go on with cracks and other features!\n\nTo reproduce:\nOpen file, sculpt from top and bottom\n\nFile is default sculpt template with sphere moved up by 5 and using scene radius.\n\n", "Geometry Nodes + Shader Editor + Asset Browser bug\nOperating system: macOS-13.1-x86_64-i386-64bit 64 Bits\nGraphics card: AMD Radeon Pro 5500M OpenGL Engine ATI Technologies Inc. 4.1 ATI-4.9.51\n\nBroken: version: 3.5 `7bf75231e8c5` , 3.4.1 `55485cb379`\nWorked: (latest 3.3.2 for Mac Intel)\n\nGeometry Nodes Editor + Shader Editor loads in all marked, related assets from Asset Browser causing crash and memory overload\n\n- Extract Asset.zip\n- Open default blender scene and assign asser library path as shown in the image\n- Open node editor\n- SHIFT+A action cause immediate memory overload from Asset Browser\n\n\n\n[Assets.zip](Assets.zip)\n\n[asset-library-indices.zip](asset-library-indices.zip)", "Cannot paginate shape key list\nSteps to reproduce:\n* add any mesh (e.g. a cube)\n* in Object Data Properties > Shape Keys click '+' twenty times\n* clicking above or below the scroll bar does not paginate the list\n\nBroken: 3.5 alpha, 50c7eb14f455, master, 2022-12-19\n", "Geometry Nodes: Rendered glitch with Image Texture and Scene Time\nOperating system: Linux-5.15.0-76-generic-x86_64-with-glibc2.35 64 Bits, X11 UI\nGraphics card: NVIDIA GeForce GTX 1060 6GB/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 535.54.03\n\nBroken: version: 3.6.0\nWorked: Never\n\n**Short description**\nGeometry nodes that use single images or image sequences are rendered inconsistently but usually look normal in the viewport.\nHigh-res images seem to be more unstable.\nHaving a connected/unconnected scene time node in the geo node group while using an image sequence with any number of frames also makes the rendered image/animation more unstable.\n\n|Expected render|Occasional render glitch|\n|--------------------------|--------------------------------------|\n|| |\n\n\n**Reproduction steps**\nThis is the most unstable setup:\n\n1. Create new project\n2. in the image editor window add a new image with resolution of around 1080x1080 pixels and set the source an image sequence \n3. add geometry nodes modifier to the default cube\n4. follow this node setup; in the image texture node use the created image sequence \n5. Render animation\n6. some of the rendered images will be glitched\n\nI've also tested this approach with face extrusion or instance selection instead of deleting faces. The resulting renders were also glitched. \n\nInside of this zip file is a blender project with a similar setup along with some sample images:\n[geo node groups that use images are rendered inconsistently.zip](attachment)\n\nThis reddit post seems to be describing a similar issue:\ngeometry_nodes_rendering_glitch_info_in_comments/\n\nThe workaround with disabling geo nodes modifier in the viewport mentioned there seems to be working in a simple setup. In the above project from the .zip file it is less effective.\n\n", "Point Clouds only support one material\nOperating system: macOS-13.4-arm64-arm-64bit 64 Bits\nGraphics card: Metal API Apple M1 Pro 1.2\n\nBroken version: Current main.\nWorked: Never\n\nWhen using the `Join Geometry` node in Geometry nodes to join two points together, their material is reset to default. This only happens when joining two or more \"lone\" points.\n\n1. Recreate the geometry node setup found in the image below\n2. Change the render engine Cycles (necessary to see points with a material)\n3. Render the scene.\n4. Notice that the materials you assigned to the points are not applied.\n\nI have also included a .blend file with steps 1 and 2 done.\n[Missing Material - Join Geometry.blend](attachment)\n\n\n", "Bevel modifier/tool expands vertex groups into surrounding geometry\nOperating system: Windows-10-10.0.19043-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3060 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 497.09\n\nBroken: version: 3.1.0 Alpha\n\nWhen beveling geometry with avertex group assigned, some of the surrounding vertices are added to that group after beveling. This creates a lot of problems when deforming a mesh in Geometry Nodes based on vertex groups.\n\n\n\n - Assign part of the mesh to a vertex group.\n - Bevel entire mesh either with a modifier or a bevel tool.\n\n\n", "Mesh Boolean and Bounding Box node joined together make lose the UVs\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3090/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 526.47\n\nBroken: version: 3.5.0\n\nApparently, joining 2 branches of geometry node that contain a Mesh Boolean node and a Bounding Box node make the UV gone. On image attached, the cube on the left with the voronoid is the original cube and on the right it's the same cube with the Geometry node turned on. Maybe other node combination recreate this issue.\nWhen inspecting the branches independently the UV are still working but stop working when adding the Join Geometry.\n\n- Have an object with texture applied. \n- Have a geometry node modifier assigned.\n- The geometry node consists of a Mesh Boolean node joined with a Bounding box node. (I used a transformed node to move the bounding box away and see our object)\n\n", "Nodes: Group input values are lost when ungrouping\nOperating system: Linux-5.15.0-72-generic-x86_64-with-glibc2.31 64 Bits\nGraphics card: NVIDIA GeForce GTX 980/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 470.182.03\n\nBroken: Curent main.\nWorked: Never.\n\nParameters are lost when ungrouping a geometry node NodeGroup\n\nUsing the given blend file with a NodeGroup that displays 2 cubes :\n1. Ungroup the NodeGroup.\n2. Parameter (a translation value) is lost (translation is now zero and the 2 cubes overlap).", "Extremely Long time delay using geometry nodes with simple mesh boolean node 37706ms\nOperating system: Linux-5.13.0-41-generic-x86_64-with-glibc2.31 64 Bits\nGraphics card: Mesa DRI Intel(R) HD Graphics 4600 (HSW GT2) Intel Open Source Technology Center 4.5 (Core Profile) Mesa 21.2.6\n\nBroken: version: 3.2.0 Beta\n\nExtremely Long time delay using geometry nodes with simple mesh boolean node and difference (it takes **37706ms** for the mesh boolean node to do a cylinder difference)\n\n\nI've included the blend file which shows the times when you change the item to do a boolean difference 1 cylinder/count to 2 cylinder/count. \n\n**When I use the Boolean Modifier this time lag doesn't occur**. I've included this in the blend file for testing also.\n\n**PS: this time lag also occurs in Blender 3.1.2**\n\n\n\n[prego lid gn.blend.zip](prego_lid_gn.blend.zip)\n", "Some areas do not use modified (by geometry nodes) curve\nOperating system: Windows 10 - Probably irrelevant.\n\nBroken: 3.0.0 (not checked earlier versions).\nWorked: (none)\n\nThe Curve modifier on mesh objects does NOT use the modified version of a curve\n\nSee attached blend file for easy example: [bug - curve with modifier.blend](bug_-_curve_with_modifier.blend)\nExample screenshot of that blender: \nThe thin line is the curve, which has geometry nodes modifying it.\nThe thick (square) line is the mesh, which targets the curve but doesn't use the modified version.\n\n - Add BezierCurve, with a few points.\n - BezierCurve: Add geometry nodes (or any other modifier, like wave or whatever) that changes the curve.\n - BezierCurve: (for visibility) add geometry.\n - Add Cube .1m \n - Cube: Add Array modifier and Curve modifier, targeting the BezierCurve\n\n**Expected result**: Mesh follows modified curve.\n**Actual result**: Mesh follows original (unmodified) curve.\n\nThe blender file shows this very easily, the modified curve is round, the mesh is in the \"old location\" that the curve would have without the modifier on the curve.\n\n\nYes, this is a duplicate of: 93702/ - Where apparently there is confusion on what the issue is. I could not find a way to un-archive a ticket.\n\n\n**Areas Affected**\n- Curve modifier\n- Spline IK" ]

[ "Difficulties in pulling out noodles from a socket of a node\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 511.79\n\nBroken: version: 3.2.0 Alpha\n\n[Hi! lately it has become hard to me to pull out noodles from a socket in a node: now I must be very precise and click almost right in the center of a socket to start the interaction]\n\nHope this can be useful\nkind regards\nFabrizio\n", "Node socket click detection is broken in some cases\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 511.79\n\nBroken: version: 3.2.0 Alpha\nbroke recently between eabdcdcd44b9 - 0a4fdcbb5fe3\nPossibly caused by {aa71414dfc}?\n\nThe click detection for nodes does not always work correctly.\n\n1. Set Blender's interface scale to 2.0 (not required)\n2. Zoom in and out on any node tree while attempting to click and drag the sockets.\n3. Notice sometimes the click is not registered.\n\nDoes not seem to affect other node trees apart from Geometry Nodes.\n\n[demo.mp4](demo.mp4)" ]

[ "stack_overflow exception while rendering with viewport render on\nOperating system: win10\nGraphics card: 2070super\n\n[blender_system_info.txt](blender_system_info.txt)\n\nBroken: 2.83.4, 2.91 6fe609d4c3\n\n\nwhen pressing f12 in render mode, blender crashes,\nwhen pressing render in solid mode, blender doesnt crash\n\n[blender_debug_output.txt](blender_debug_output.txt)\n```\nError : EXCEPTION_STACK_OVERFLOW\nAddress : 0x00007FFF42963FDB\nModule : C:\\Windows\\system32\\DriverStore\\FileRepository\\nv_dispsi.inf_amd64_6af46268c19e4980\\nvcuda64.dll\n```\n[2.blend](2.blend) \n1. switch to render mode\n2. press f12 & watch task manager.\n3. blender will close", "Blender Opens Window then Closes/crashes\nOperating system: Windows 10\nGraphics card: AMD FirePro W2100\n\nBroken: 3.5.1 2023-04-24 18:11 Hash:e1ccd9d4a123 Branch: Blender -v3.5-release\nWorked: first version used on this computer\n\n-First attempted to run blender, found error by running the debug, error:\n\n_EXCEPTION_ACCESS_VIOLATION_\n\n- Did specific driver updates for AMD FirePro W2100 and Intel CPU\n\n- Then ran dell command update\n\n- New error appeared after driver updates and running blender:\n\n_A graphics card and driver with support for OpenGL 3.3 or higher is required.\nInstalling the latest driver for your graphics card might resolve the issue._\n\n- Then downloaded GPU-Z to figure out OpenGL version\n- Open GL version was 1.1 when monitor DisplayPort was plugged into the motherboard\n- Open GL version was 4.5 when monitor DisplayPort was plugged into GPU\n\n- Now blender window would appear but closes immediately\n- running blender_debug_gpu_glitchworkaround.cmd in the Blender file worked to make Blender run but is NOT the overall fix\n- updated python\n- did windows updates\n\nunsure\n\n", "Cycles Rendering of project causes crash of program.\nOperating system: Windows-10-10.0.22621-SP0 64 Bits (Windows 11)\nGraphics card: NVIDIA GeForce RTX 3070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 537.13\n\nBroken: version: 3.6.2\n\nTrying to render, pressing F12 of the file to render project it crashes. I'm using Blender 3.6.2\n\nTrying to render, pressing F12 of the file to render project it crashes. I'm using Blender 3.6.2\nNot sure if you need textures, can provide if needed.", "AMD: Rendering Instances On Points using a Collection crashes Blender on Linux\nBroken\nOperating system: Linux-6.2.0-060200-generic-x86_64-with-glibc2.35 64 Bits\nGraphics card: AMD Radeon RX 580 Series (polaris10, LLVM 15.0.7, DRM 3.49, 6.2.0-060200-generic) AMD 4.6 (Core Profile) Mesa 23.1.0\nversion: 3.5.1\n\nBroken\nOperating system: Linux-6.1.18-200.fc37.x86_64-x86_64-with-glibc2.36 64 Bits\nGraphics card: AMD Radeon Graphics (renoir, LLVM 15.0.7, DRM 3.49, 6.1.18-200.fc37.x86_64) AMD 4.6 (Core Profile) Mesa 22.3.7\nversion: 3.5.1\n\nFine (same system as above, but using nvidia)\nOperating system: Linux-6.1.18-200.fc37.x86_64-x86_64-with-glibc2.36 64 Bits\nGraphics card: NVIDIA GeForce RTX 3080 Laptop GPU/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 525.89.02\nversion: 3.5.1\n\n\nOn Linux, using Geometry Nodes to have a Collection Info node feed instances of like items to an Instance on Points node going to Group Output will work properly, but will crash Blender when Rendering.\n\nDoing the obligatory Donut Tutorial. When creating the sprinkles using the Collection Info node to feed Instance on Points will create a nice set of sprinkles. The sprinkles share a Linked Material to randomize colors. Using Render Preview in the Layout (or Shading) tab correctly displays randomly colored sprinkles. However, actually rendering crashes Blender. Using the Realize Instances geometry node before the Output node will allow it to render, but then every sprinkle has a single, random color.\n\nReproduce (using Linux): \n1. Using the attached file, render using the Realize Instances node on the Icing's Geometry Nodes (file's current setup). Blender will not crash, but all sprinkles have same color. \n2. Remove the Realize Instances geometry node, and the sprinkle colors will be properly randomized. Render, and blender will crash.\n\nWarning: This is a hard system freeze, not a simple Blender crash. Typically requires a system reset to recover.\nUpdate: AMD graphic cards only. Eevee rendering only.", "Regression: Crash when animation is playing\nOperating system: Windows 10\nGraphics card: 2x RTX3080\n\nBroken: 3.6 and 3.6.1 release canditate\n\nPress the space bar, the animation starts and Blender closes. The console says : EXCEPTION_ACCESS_VIOLATION\nI also test the file with an independent portable version of Blender, but the crash still occurs.\n\nPress space bar to play the animation and wait the frame 48.\n\n[Fluent_Testing_7_for_blend_dev.blend](https://projects.blender.orgattachment)", "Segmentation fault when selecting context.area.type before applying operator (scripting)\nOS: Ubuntu 16.04.4 LTS\nHW: Dell XPS 9550 notebook, using Intel HD GPU (nvidia deactivted)\n\nBroken: 2.83\n\nExecuting the operator `bpy.ops.transform.translate` causes a crash if `bpy.context.area.type` is set manually before.\nCrash in `setTransformViewMatrices` because `t->ar->regiondata` is `NULL`\n\nRunning the following script causes the crash:\n```\nimport bpy\nimport bmesh\n\n# Create object\nbpy.ops.mesh.primitive_plane_add(size=1, enter_editmode=False, location=(0, 0, 0))\n\n# Enter edit mode\nbpy.context.area.type = 'VIEW_3D' # If this is not set, crash does not occur\nbpy.ops.object.mode_set(mode='EDIT', toggle=False)\n\n# Modify vertices\nmesh = bmesh.from_edit_mesh(bpy.context.active_object.data)\nmesh.verts.ensure_lookup_table()\nmesh.verts[0].select = True\n\n# Next call causes crash:\nbpy.ops.transform.translate(value=(0, 0, 1), constraint_axis=(False, False, True),\n mirror=False, use_proportional_edit=True,\n proportional_edit_falloff='SHARP')\n```\n", "Crash (Seg Fault) When Rendering Grease Pencil Stroke on Render Farm\nOperating system: Darwin-19.5.0-x86_64-i386-64bit 64 Bits\nGraphics card: AMD Radeon Pro 5500M OpenGL Engine ATI Technologies Inc. 4.1 ATI-3.9.15\n\nBroken: version: 2.83.1\n\nOur render farm of 70 iMacs has been running for 3 years through the 2.7x series up till 2.81a.\nWhen we switched to 2.83 last week it started crashing consistently when rendering Grease Pencil strokes.\nThe crash occurs when using Blender to do command line rendering via an SSH session on a machine without the user physically logged in. (Crashes on Mac and Linux)\n\nCommand Line Rendering on my own machine works fine.\nCommand Line Rendering on a remote machine via SSH works fine if I'm also logged into that machine.\nBut as soon as I log out of the machine so that the SSH session is the only active user on the machine, then the error shows up.\n\nI tested this on 2.81a and the error (below) shows up there too, but in 2.81a it doesn't also lead to the Segmentation Fault that crashes Blender.\nIt's only in 2.83 (and 2.83.1) that the error is followed by the Seg Fault 11\n\n\n**_RegisterApplication(), FAILED TO establish the default connection to the WindowServer, _CGSDefaultConnection() is NULL.\nWriting: /var/folders/jk/7b488ncn6c13t515bnt6q_8w68xcj1/T/test.crash.txt\nSegmentation fault: 11**\n\non Linux the error message is slightly different, but still a crash:\nUnable to open a display\nAborted (core dumped)\n\nContents of test.crash.txt (from the Mac crash)\n# Blender 2.83.1, Commit date: 2020-06-25 09:47, Hash 8289fc688b3e\n\n# backtrace\n0 Blender 0x00000001079eb9d7 BLI_system_backtrace + 55\n1 Blender 0x000000010159ddb8 sig_handle_crash + 392\n2 libsystem_platform.dylib 0x00007fff5db5fb3d _sigtramp + 29\n3 ??? 0x00007f8594802010 0x0 + 140211698802704\n4 Blender 0x0000000101a39da2 DRW_opengl_context_create + 66\n5 Blender 0x00000001018f47d5 WM_init_opengl + 37\n6 Blender 0x0000000101a374ff DRW_render_gpencil + 127\n7 Blender 0x0000000105de5ef4 RE_engine_render + 1156\n8 Blender 0x0000000105df06d3 do_render_all_options + 387\n9 Blender 0x0000000105df2a14 RE_RenderAnim + 2260\n10 Blender 0x000000010159c9dd arg_handle_render_animation + 141\n11 Blender 0x000000010797880d BLI_argsParse + 173\n12 Blender 0x000000010159a497 main + 855\n13 libdyld.dylib 0x00007fff5d974ed9 start + 1\n14 ??? 0x000000000000000a 0x0 + 10\n\n\n\n1) Make a new default file.\n2) Delete the Cube\n3) Add->Grease Pencil->Monkey\n4) Save as test.blend\n5) Copy test.blend onto a remote machine\n6) Make sure the machine is at the login prompt (showing the username password boxes).\n7) ssh into the remote machine\n8) command line render one frame from test.blend something like:\nBlender -b /test.blend -s 1 -e 1 -a\n\n", "use-after-free crash after setting scene's render.engine inside custom render engine context\nOperating system: Ubuntu 20.04.2LTS\n\nBroken: 2.93.0 Beta, branch: master, commit date: 2021-05-19 10:14, hash: a294670bacb1, type: release\nAlso Broken: 0efb627bbd3ac3e53afb95591ae4e244bd1f8af4\n\nBlender crashes when setting some scene's render.engine property inside `__init__` or the `view_update` context of a custom render engine to a different one than the custom engine.\n\n[T88404_asan_report.txt](T88404_asan_report.txt)\n**Code to reproduce the error**\n```\nimport bpy\nimport bgl\n\nclass CustomRenderEngine(bpy.types.RenderEngine):\n bl_idname = \"CUSTOM\"\n bl_label = \"Custom\"\n bl_use_preview = True\n \n def __init__(self):\n return\n scene = bpy.data.scenes.new(\"foo\")\n scene.render.engine = \"CYCLES\"\n bpy.data.scenes.remove(scene)\n\n def __del__(self):\n pass\n\n def view_update(self, context, depsgraph):\n scene = bpy.data.scenes.new(\"foo\")\n scene.render.engine = \"CYCLES\"\n bpy.data.scenes.remove(scene)\n\n def view_draw(self, context, depsgraph):\n pass\n\ndef get_panels():\n exclude_panels = {\n 'VIEWLAYER_PT_filter',\n 'VIEWLAYER_PT_layer_passes',\n }\n\n panels = []\n for panel in bpy.types.Panel.__subclasses__():\n if hasattr(panel, 'COMPAT_ENGINES') and 'BLENDER_RENDER' in panel.COMPAT_ENGINES:\n if panel.__name__ not in exclude_panels:\n panels.append(panel)\n\n return panels\n\n\ndef register():\n bpy.utils.register_class(CustomRenderEngine)\n\n for panel in get_panels():\n panel.COMPAT_ENGINES.add('CUSTOM')\n\n\ndef unregister():\n bpy.utils.unregister_class(CustomRenderEngine)\n\n for panel in get_panels():\n if 'CUSTOM' in panel.COMPAT_ENGINES:\n panel.COMPAT_ENGINES.remove('CUSTOM')\n\nif __name__ == \"__main__\":\n register()\n```\n\n1) Copy code from above into a new script file inside of Blender\n2) Run the script\n3) Select \"Custom\" as Render Engine inside Render Properties\n4) Select \"Material Preview\" as display method in the viewport.\n\n[crash.blend](crash.blend)\n\nCrash log:\n```\n# Blender 2.93.0, Commit date: 2021-05-19 10:14, Hash a294670bacb1\nEngine 'CUSTOM' not available for scene 'Scene' (an add-on may need to be installed or enabled) # Error\nbpy.ops.text.run_script() # Operator\n\n# backtrace\n./blender(BLI_system_backtrace+0x20) [0xa63c950]\n./blender() [0xf6082a]\n/lib/x86_64-linux-gnu/libc.so.6(+0x46210) [0x7f7ce17be210]\n./blender(GPU_framebuffer_blit+0x2f) [0x8664faf]\n./blender() [0x1485fc3]\n./blender() [0x14597fe]\n./blender(DRW_draw_render_loop_ex+0x242) [0x1459a72]\n./blender(view3d_main_region_draw+0x8f) [0x1d3570f]\n./blender(ED_region_do_draw+0x851) [0x189c771]\n./blender(wm_draw_update+0x52e) [0x130cc9e]\n./blender(WM_main+0x30) [0x130aa20]\n./blender(main+0x319) [0xe881c9]\n/lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xf3) [0x7f7ce179f0b3]\n./blender() [0xf5d1bc]\n\n# Python backtrace\n```", "3D Texturing: Limit Eevee samples to 1 during paint stroke.\nGhosting effects appear when painting. We should limit the number of samples to improve the performance and reduce the ghosting effect.", "Ctest: Crash in cycles_image_colorspace_cpu\nOperating system: Windows\nGraphics card: Irrelevant\n\nBroken: current master (c32c6a5f85) \nWorked: Unknown\n\n\nDuring testing of [D10212: Cmake/deps: Update OSL to 1.11.10.0](D10212) I noticed a single test failing... once... by repeating the test over and over I got it to misbehave once every 5-45 minutes , so... that's not gonna be super fun to track down.\n\nI attached 3 stack traces which all point to the same location, a null check there just mask the issue, personally I'd take a crash over a random bad frame being output. \n\n\n`ctest -C Release -VV -R cycles_image_colorspace_cpu --repeat-until-fail 9999 -O Fail.log`\n\n\n[image_non_color.crash.txt](image_non_color.crash.txt)\n\n[image_log_osl.crash.txt](image_log_osl.crash.txt)\n\n[image_alpha_blend.crash.txt](image_alpha_blend.crash.txt)\n\nI mentioned this on chat earlier today, but tracking ticket is probably nicer cc: @brecht ", "Random crashes with NVidia + GPU Subdivision (sometimes when activating X-Ray or Wireframe overlay)\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\n| Graphics card: | Driver : | User:\n| --- | --- | ---\n| NVIDIA GeForce RTX 3090/PCIe/SSE2 | 4.5.0 NVIDIA 516.94 | @GabrielMoro\n| NVIDIA GeForce GTX 1050 Ti/PCIe/SSE2 | 4.5.0 NVIDIA 512.15 | @STYXtheMUTT\n| NVIDIA GeForce RTX 3070 Ti/PCIe/SSE2 | 4.5.0 NVIDIA 516.94 | @Jojolix\n|NVIDIA GeForce RTX 2070 SUPER/PCIe/SSE2|NVIDIA 516.59|@MassimilianoPuliero\n|NVIDIA GeForce RTX 3090|NVIDIA 516.94|@The5\n\nBroken: version: 3.2.2\n\nOn some seemingly random occasions, Blender crashes with the following stack:\n```lines=10\nStack trace:\nblender.exe :0x00007FF77C1458B0 BKE_subdiv_free\nblender.exe :0x00007FF77C147880 BKE_subsurf_modifier_subdiv_descriptor_ensure\nblender.exe :0x00007FF77BFFC3D0 <lambda_c000e58b9a8caf72c54e0e02a46effdb>::operator()\ntbb.dll :0x00007FFDBAF64FD0 tbb::interface7::internal::isolate_within_arena\nblender.exe :0x00007FF77BFFC6F0 BKE_mesh_wrapper_ensure_subdivision\nblender.exe :0x00007FF77CB38FA0 raycast_obj_fn\nblender.exe :0x00007FF77CB375E0 iter_snap_objects\nblender.exe :0x00007FF77CB3E530 transform_snap_context_project_view3d_mixed_impl\nblender.exe :0x00007FF77CB3F0E0 ED_transform_snap_object_project_view3d\nblender.exe :0x00007FF77CB5E7C0 applyFaceProject\nblender.exe :0x00007FF77CB5EE00 applySnappingIndividual\nblender.exe :0x00007FF77CB72F70 recalcData_objects\nblender.exe :0x00007FF77CB81010 applyTranslation\nblender.exe :0x00007FF77CB46BE0 transformApply\nblender.exe :0x00007FF77CB41920 transform_modal\nblender.exe :0x00007FF77C229D30 wm_macro_modal\nblender.exe :0x00007FF77C20F9E0 wm_handler_operator_call\nblender.exe :0x00007FF77C210BD0 wm_handlers_do_intern\nblender.exe :0x00007FF77C210000 wm_handlers_do\nblender.exe :0x00007FF77C216590 wm_event_do_handlers\nblender.exe :0x00007FF77C1FC380 WM_main\nblender.exe :0x00007FF77B2F12A0 main\nblender.exe :0x00007FF7814128B0 __scrt_common_main_seh\nKERNEL32.DLL :0x00007FFDC7B67020 BaseThreadInitThunk\nntdll.dll :0x00007FFDC85A2680 RtlUserThreadStart\n```\n```lines=10\nStack trace:\nblender.exe :0x00007FF7A50C5B10 BKE_subdiv_update_from_mesh\nblender.exe :0x00007FF7A50C7880 BKE_subsurf_modifier_subdiv_descriptor_ensure\nblender.exe :0x00007FF7A52A1110 draw_subdiv_create_requested_buffers\nblender.exe :0x00007FF7A52A2080 DRW_create_subdivision\nblender.exe :0x00007FF7A52D1870 DRW_mesh_batch_cache_create_requested\nblender.exe :0x00007FF7A52C6000 drw_batch_cache_generate_requested\nblender.exe :0x00007FF7A529F1A0 drw_engines_cache_populate\nblender.exe :0x00007FF7A529B8C0 DRW_draw_render_loop_ex\nblender.exe :0x00007FF7A529C9B0 DRW_draw_view\nblender.exe :0x00007FF7A5D3FA00 view3d_main_region_draw\nblender.exe :0x00007FF7A5551070 ED_region_do_draw\nblender.exe :0x00007FF7A51A3700 wm_draw_window_offscreen\nblender.exe :0x00007FF7A51A3560 wm_draw_window\nblender.exe :0x00007FF7A51A2FB0 wm_draw_update\nblender.exe :0x00007FF7A517C380 WM_main\nblender.exe :0x00007FF7A42712A0 main\nblender.exe :0x00007FF7AA3928B0 __scrt_common_main_seh\nKERNEL32.DLL :0x00007FFD1C4A54D0 BaseThreadInitThunk\nntdll.dll :0x00007FFD1D524830 RtlUserThreadStart\n```\n\nOccasions observed:\n- try to snap objects on faces (the object that shall be snapped to is usually a complex mesh with modifiers like subdivisions, mirror, solidify, bevel and booleans)\n- toggle the shading mode between shaded and wireframe or between X-Ray and Wireframe Overlay.\n[intro.crash.txt](intro.crash.txt)\n\n\n- open attached file\n- disable and enable X-Ray multiple times\nIt happens specially when working fast.\n\n[BLENDER_Youre_Alive_v67.blend](BLENDER_Youre_Alive_v67.blend)\n\n", "Workbench: Random object color disables transparency in solid view\nOperating system: Windows 10 64bit\nGraphics card: Nvidia RTX 2080ti\n\nBroken: 2.83.1 official\n\n\n\n\nWhen setting the color of the material in the viewport to be semitransparent (Alpha value = 0.1) it works if the color of the viewport objects is set to \"Material\" it works, while if the color is set to \"Random\" the object is not semitransparent anymore.\n\nBlend file to test:\n[bug.blend](bug.blend)", "Themes→3dViewport→Active vertex color do nothing\nOperating system: Windows-10-10.0.17763-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 512.95\n\nBroken: version: 3.5.0\n\nThis one influence nothing. Should it be deleted?\n\n\n[2023-04-13_07-48-03.mp4](attachment)\n\n\np.s. Dull colors and other issues about wireframe [described here](426) and I am sad, that it looks abandoned and ignored…\n\n", "Blender development todo list -- Render\nNOTE: this is a direct dump of Render (without cherrypicking the valid items)\n\n### Render & UI\n\n- [ ] While rendering, manipulating images (like reload, resizing generated) can still crash Blender. Better image \"owner\" locking is required for 2.5 (#28477)\n- [ ] While rendering, several UI changes can cause crashes (#24758)\n- [ ] Rendering can't really use normal scene data if ui is interactive without causing all kinds of trouble, for examples and related issues see the item above this one.. and the reports (#26535)\n- [ ] Renaming scene leads to missing rendered result in image editor and compositor (see (#30222))\n- [ ] Rendering with File Output nodes could produce unwanted output images, see (#38605)\n- [ ] Shaded draw mode (using render code for vertex colors) is most certainly not coming back\n- [ ] Color Management can be communicated better, like when sampling colors in Image Window it confuses (could indicate such). Color Management also wasn't added backward compatible, disabling it gives dark results.\n- [ ] Compositor \"file out\" node doesn't work for FSA, and as color management troubles too.\n- [ ] FSA render uses temp files with names preventing other Blender instances to render too. (#28314)\n- [ ] Texture nodes have been implemented by recursively (backwards) calling parent nodes, this gives issues for viewers, previews and threads in general.We will try to keep it stable, but the functionality for this is going to replaced with a better shader system in 2.6. (compositor and material nodes call nodes forward, sorted on dependency. Each node is only called once this way).\n- [ ] Render display in Image window, for a window previously showing UV editor, doesn't restore correctly. (#26828). Code for this is too much spread around and fuzzy, needs full revision.\n- [ ] Object transform is not possible during rendering #37019\n- [ ] Don't render and warn when a driver or pre-render script can't be executed, a texture is missing, physics are not baked. Needed e.g. when renderfarms have scripts blocked.\n\n### Node Textures\n- [ ] Grouping texture nodes affects on data range #45696\n\n### Baking\n- [ ] add option to smooth vertex colors as a post process after baking #35066.\n- [ ] Baking to a low-res textures might miss pixels which lies on the UV island boundary #37681\n- [ ] Vertex baking might cause AO artifacts #44178\n- [ ] Support dupligroups #43553\n- [ ] Modifier produces non deterministic effects #42015\n- [ ] Update Blender side baking to support different topologies needed for adaptive subdivision #49405", "Crash when shading\nModel: HP Probook x360 435 G7 (independently upgraded SSD and RAM)\nOperating system: Windows 11 Pro 22H2 22621.1555\nGraphics card: AMD Radeon Graphics (Built-in graphics of AMD Ryzen 5 4500U with Radeon Graphics)\n\nBroken: 3.5.1 and 3.3.1 were tested, both had this issue.\nWorked: 3.1.2 – At the time of issue creation, there are no newer versions that have worked, including the latest build of 3.6.0 Alpha.\n\nBlender freezes for a second or so, and then crashes.\n\nI don't know if it *is* reproducable, but I have tested multiple conditions under which it crashes. These conditions are accompanied with the crash logs of that test. The problem does not seem to be (V)RAM usage, as the usage remains constant from before Blender is opened to long after.\n- Press F12 to Render Image: Crash [Log](itihmb)\n- Select Default Cube, Click on Shading workspace: Crash [Log](nxyeck)\n- Select Default Cube, Go to Materials tab: Crash [Log](ixgrhp)\n- Change Viewport Shading mode to Material Preview: Crash [Log](diwmma)\n- Change Viewport Shading mode to Render Preview: Crash [Log](ztajes)\n- DEL to delete Default Cube, SHIFT+A M P to create a Plane, Go into Materials Tab, Create Material: Crash [Log](gcaded)\n- Change Render Engine to Cycles, Press F12 to Render Image: **No** crash\n- Change Render Engine to Cycles, Click on Shading workspace: Crash [Log](chmkgj)\n- Change Render Engine to Cycles, Go to Materials tab: **No** crash\n- Change Render Engine to Cycles, Change Viewport Shading mode to Material Preview: Crash [Log](fhfyfo)\n- Change Render Engine to Cycles, Change Viewport Shading mode to Render Preview: **No** crash\n- Change Render Engine to Cycles, DEL to delete Default Cube, SHIFT+A M P to create a Plane, Go into Materials Tab, Create Material: **No** crash\n- Change Render Engine to Workbench, Press F12 to Render Image: **No** crash\n- Change Render Engine to Workbench, Click on Shading workspace: **No** crash\n- Change Render Engine to Workbench, Go to Materials tab: **No** crash\n- Change Render Engine to Workbench, Change Viewport Shading mode to Material Preview: **No** crash\n- Change Render Engine to Workbench, Change Viewport Shading mode to Render Preview: **No** crash\n- Change Render Engine to Workbench, DEL to delete Default Cube, SHIFT+A M P to create a Plane, Go into Materials Tab, Create Material: **No** crash\nThank you for your help.\n\n" ]

[ "Ctrl right mouse click on mesh in weight paint mode crashes Blender\nOperating system: Linux-4.9.0-9-amd64-x86_64-with-debian-9.9 64 Bits\nGraphics card: GeForce GTX 1060 6GB/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 390.116\n\n\nBroken: version: 2.80 (sub 61)\nalso version from 11.05.2019\nWorked: I didn't test it\n\n\nCtrl right mouse click on the mesh to collect weight sample crashes Blender immediately.\nsee attached simple file.\n\n1. start Blender with --factory-startup\n2. open attached file\n3. ctrl right click the mesh to collect weight painted sample\n - > Blender crashes immediately\n[ctrlrightmouse-bug.blend](ctrlrightmouse-bug.blend)\n\n", "Texture Paint mode color picker hotkey crash when object has UV\nOperating system: Windows-10-10.0.17134 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 425.31\n\n\nBroken: version: 2.80 (sub 61)\nDelete all the UVs and the color picker hotkey stop crashing\nClick the color picker on the color palette and sample a color on the viewport works fine\n\n\nWhen painting a texture, place the mouse on the viewport and hit S will crash Blender. For some reason, deleting all the UVs and the hot key stop crashing\nThis only happen when using the hotkey S to pick a color, manually select the color picker and pick a color in the viewport works but then it pick the shaded color of the viewport instead of the actual color of the texture being painted.\n\nUnwrap UV for the default cube\nGo to texture paint mode (You don't even need to add a texture)\nPlace the cursor anywhere in the viewport\nHit S\nBlender crashes\n\n", "Texture Paint Crash\nOperating system: Linux-4.15.0-48-generic-x86_64-with-debian-stretch-sid 64 Bits\nGraphics card: GeForce GTX 1050 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 418.56\n\n\nBroken: version: 2.80 (sub 61)\nWorked: (optional)\n\n\nBlender crashing when pressed **S** key in Texture Paint tab. Also tested in factory settings.\n\nGo to Texture Paint tab > Press **S** key in 3D viewport\n\n\n\n", "Blender 2.8 crash\nOperating system: Windows 10 \nGraphics card: GTX 1070\n\nBroken: (example: 2.80, edbf15d3c044, master, 2018-11-28, as found on the splash screen): 2.8 ebc44aae9897\nWorked: (optional)\n\n\nBlender crashed frequently when undoing an action, so I reinstalled the application but now it crashes every time I try to sample a color in texture paint. \n\nBased on the default startup or an attached .blend file (as simple as possible).\n1.) UV unwrap object\n2.) set it up so you can texture paint on the object\n3.) hit 'S' key to sample color on object - this causes the crash \n[Fish.blend](Fish.blend)", "Sample in texture paint causes instant shutdown\nOperating system: Windows 7 64 Bit\nGraphics card: Geforce GTX 570\n\nBroken: (example: 2.80, edbf15d3c044, master, 2018-11-28, as found on the splash screen)\nWorked: (optional)\n\nHit (s) in texture paint mode to sample color, instant crash.\n\nBased on the default startup or an attached .blend file (as simple as possible).\n\nTake the cube, smart UV unwrap, give it a texture to start texture painting. Draw a color. Then hit (s) to sample the color, Blender quits." ]

[ "GPUOffScreen example give strange \nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2080 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 512.96\n\nBroken: version: 3.1.2\n\n\n\n1. copy the code from here gpu.html#rendering-the-3d-view-into-a-texture\n2. run it \n\n", "2.80 Pose Breakdowner doesn't work in Add NLA Strips\nOperating system: ubuntu 18.04\nGraphics card: gtx1080ti\n\nBroken: (2.80)\n\nPose Breakdowner doesnt work when editing NLA Tracks \n\n\nOpen attached file and use Pose Breakdowner\n[broken .blend](broken_.blend)", "eGPU doesn't get used by EEVEE on Windows (Mac / Bootcamp)\nOperating system: Windows 10 1903 (Bootcamp)\nGraphics card: Radeon RX 560, Radeon RX Vega 64\nDevice: Macbook Pro 15\" 2018\n\nBroken: 2.82\n\nWhile using a eGPU on Windows 10 via Bootcamp, Blender doesn't use the external GPU to accelerate and power EEVEE. No matter if the Blender Graphic Setting preferences are set to High Performance, ergo utilizing the RX Vega 64 in any scenario. Missing usage of the external GPU can be tracked via the Task Manager -> Performance Tab. (See screenshots).\n\n**Note 1:**Cycles on Windows/OpenCL recognize and uses the external GPU with no issue as expected.\n\n**Note 2:** On Mac OS Catalina, EEVEE can use the external GPU as expected by enabling the option 'Prefer External GPU'.\n\nSo the problem affects only Windows when an external GPU is used for EEVEE only.\n\n\n\n\n\n\n\n- Open Blender and launch a rendering with EEVEE\n- Check GPU usage through Task Manager \n- On any operating during rendering the external GPU is not used at all.", "No error prompted when Cuda fails during GPU bake\nOperating system: Linux Debian 10 (buster) - kernel 4.19.0-19-amd64\nGraphics card: GeForce GTX 1060 Mobile\n\nBroken: 3.1.0\nWorked: -\n\nWhen you try to bake, having GPU set as compute device, and CUDA has an error, no error is shown.\nYou only notice the error when you try to render and get \"*Failed to create CUDA context (unknown error)*\"\n\nIt is useful to show an error, in the same way that, for example, \"Circular dependency for image...\" is shown during the bake.\n\nNote that this report is not about the error itself. It is about Blender not reporting any errors during baking, which causes user confusion.\n\nThank you,\nRiccardo", "Asset Browser don't show my asset's and crash Blender\nOperating system: Windows-10-10.0.18363-SP0 64 Bits\nGraphics card: Radeon RX 560 Series ATI Technologies Inc. 4.5.0 Core Profile Context 22.11.2.221130\n\nBroken: 3.4, 3.5.0\nWorked: \n\n\nI created my asset browser and at first it worked well. But after it's worked strange. Then i opened asset browser i don't see my assets, but after refreshing Blender was crashed. I tried to re-specify the path to the library, but was changed nothing. I tryed to install new Blender (3.5) and it take change but don't fix it. I can't see assets in asset browser, but then a refresh it, Blender don't crash.\n\n\n- Download and extract the attached .zip file\n- In Preferences -> File Paths -> Asset Libraries, add the directory of the unzipped file\n- Go to Asset Browser an look for the assets in the added library - You cannot find then\n- Refresh - crash", "activate addons via command line does not active it in the preferences\nOperating system: Linux-3.10.0-1127.18.2.el7.x86_64-x86_64-with-centos-7.2.1511-Core 64 Bits\nGraphics card: GeForce GTX 1070 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 440.36\n\nBroken: version: 2.90.1\nWorked: Never (2.8+)\n\nwhen launching blender with --addons argument, it correctly registered the scripts but do not activate it in the preferences, which makes addon options not available unless checkbox is manually enabled.\n\nfor example, make sure node_wrangler addon is not enabled in userpref.blend, then launch blender with \"blender --addons node_wrangler\". the scripts will be registered but not the addons and its properties.\n\n", "Lots of animation-related operators do not check if the data is editable or not\nMany animation-related operators do not do any check whether the data-block they operate on is linked, liboverride or fully local data. This seems to affect in particular operations from the 3DView (like key insertions), and in the NLA (like adding new tracks, duplicating strips, etc.).\n\nE.g. one can insert keyframes on a linked object (ie create animdata for that linked object, a new local action, and assign that local action to the linked object!).\n\nRelated to issue reported in #98686 (Cant rename local NLA tracks within a local library override data-block.).", "GPU Compute is not available\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: GeForce GTX 660/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.71\n\nBroken: version: 3.6.2\n\nGPU Compute is not available in render device untill recheck\n\n\nAn interesting circumstance was discovered:\nIf open `preferences/System/CUDA` and just uncheck and check the checkbox of GPU Device, the problem is solved, then GPU Compute is accessible and fully functional.\n\nhad a similar problem: 94699\nThe solution is supposedly installing a newer version of nvidia driver. However, in fact, Blender works fine with the existing version.\n\n", "System UI Freeze and blender UI blinks\nOperating system: Linux-6.4.12-zen1-1-zen-x86_64-with-glibc2.38 64 Bits, WAYLAND UI\nGraphics card: Mesa Intel(R) HD Graphics 630 (KBL GT2) Intel 4.6 (Core Profile) Mesa 23.1.6-arch1.4\n\nBroken: version: 3.6.2\n\nblender often freezes (with the entire UI of the System), sometimes blender comes back to control, but, there are times when doesn't, the blender UI blinks to come entirely black and unresponsive.\n\nThis happens when I was using a graphic tablet to sculpt something.\nJust opened blender and started to sculpt, eventually the situation occur.\n\n", "Request for Telugu Translation of Blender \nI want to translate blender into my native language. Telugu. Can anybody help in this regard. I can translate the words but somebody has to tell me what to translate and how it works. Kindly somebody guide me or tell the developers.. ", "Rendering animation freezes with particle system\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce RTX 3090/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 461.40\n\nBroken: version: 2.91.2\nWorked: 2.83LTS (not sure, this is said in most of the comments)\n\nI am trying to render out an animation based on the CGBoost Blender Launchpad Tutorial. Every time I try to render, it will render around 50-400 frames and then it will just stop rendering new frames. My CPU usage goes straight up to 100% when that happens and my GPU usage drops down to 0%. The Blender UI is still responsive, it is just that the render seems to get stuck on a frame. If I cancel the render, then Blender becomes totally unresponsive and I have to force quit via the Task Manager.\n\nIssue happens in both EEVEE and Cycles\n\n**Workarounds (picked up from comments)**\n- Render animation from command line\n- Check lock interface: {nav Render > Lock interface}\n\n1. Open attached Blend file \n2. Press Ctrl-F12 to render out animation.\n3. Wait for render to freeze on a frame. This can happen within the first few frames or after 400-500 frames. It seems random.\n4. Examine CPU and GPU usage.\n\n[Blender Launchpad.blend](Blender_Launchpad.blend)\n[ING BG.blend](ING_BG.blend) (File provided by @Megaseppl: #85267#1122053)", "Blender 4 Alpha Won't Open\nOperating system: Linux-5.15.0-79-generic-x86_64-with-glibc2.31 64 Bits, X11 UI\nGraphics card: AMD JUNIPER (DRM 2.50.0 / 5.15.0-79-generic, LLVM 12.0.0) X.Org 3.3 (Core Profile) Mesa 21.2.6\n\nBroken: I tried opening Blender 4.0.0 - Alpha\n August 30, 13:25:10\n e071288ab290\n tar.xz\n 250.91MB\n\nWorked: Blender 4 Alpha from July 15 opens, though.\nBlender 4.0.0 - Alpha\n July 15, 02:35:02\n d8e4fe320784\n tar.xz\n 249.01MB\n\n\n\n- I tried opening Blender 4 Alpha, latest version, and nothing happens.\n- I also tried an earlier version from August 15, and another from July 15.\n- August 15 wouldn't open. I made sure it was set to \"run as executable\".\n- July 15 opens, though.\n- I might check other versions. They take a while to download, though (I live in Canada).\n\n\nBased on the default startup or an attached .blend file (as simple as possible).\n - Not sure how others can reproduce this unless they have my pc, or are having the same problem.\n - If there's some kind of terminal window thingy that writes down info while trying to open Blender, let me know how to do that, and I'll copy and paste whatever it says.\n- Note: sometimes Blender 3.6 doesn't open at first. But if I try again, it opens just fine.\n\n", "Menus & Context Menus: What to do if operators aren't valid\nThis is a design doc to clarify how we should handle cases where operators aren't applicable. \n\nCurrently, we deal with this poorly:\n\n - In the context menus, we often show properties that won't work (such as when nothing is selected)\n - In regular menus, sometimes operators don't work in certain criteria, but it isn't communicated\n\nHere's how I think we should solve it:\n\n## Context menus\nRule: Only ever show operators here that actually work in the current context.\nThis means that, if nothing is selected, we should not show operators to act on the selection, since this will fail.\nIn general the context menus are always meant to show the user relevant items for the current situation, so this fits with that concept.\n\n**When nothing is selected**\nFor the case where nothing is selected, there often won't be very many applicable operators. In that case, we should add operators to add new items (objects, meshes, strips etc).\n\n## Regular menus\nThese we should keep constant, and not hide or show items, since it breaks muscle memory. Instead, we should use greying out. Preferably we should improve operator polling so that operators get greyed out when not applicable. When nothing is selected, or the wrong kind of item is selected, operatoers that don't do anything then should become greyed out.", "Link-drag-search unsupported in texture node editor\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3060 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.59\n\nBroken: version: 3.3.1\nWorked: 3.0.0, since there is no such feature.\n\n\"Auto-completion/search\" is not working as intended with Texture Node Editor. For some reason, only reroutes are being displayed. [2022-11-23 22-42-21_1.mp4](2022-11-23_22-42-21_1.mp4)\n\nOpen Texture Node Editor -> Drag from any socket.\n\nNote: sorry for the incorrect spelling of word \"reroute\"\n", "Assigning shortcut to Sequencer data is wrongly 'hooked' the Scene instead of Scene's VSE data\nThis error shows all the time when I right click on the viewscreen, and I have to right click and press ESC, so the message disappears and I keep working this way, right click, a message appears, ESC.\nIt suddenly started to happen and even deleting and downloading blender again, even with the other 2.8 version it happens.\n\nThank you, I love your work guys\nJust trying to help\n\nOperating system: Windows 10 Home, Core i7 - 7500U @ 2.70GHz, 8.00 GB RAM\nGraphics card: AMD Radeon R7 M440 32GB, 4.00 Gbps\n\nBroken: (example: 2.80, edbf15d3c044, master, 2018-11-28, as found on the splash screen)\nWorked: (optional) " ]

[ "All LTS download links on blender.org/download/lts/ lead to 404\nOperating system: Linux-5.10.22-200.fc33.x86_64-x86_64-with-fedora-33-Thirty_Three 64 Bits\nGraphics card: Mesa DRI Intel(R) HD Graphics 4600 (HSW GT2) Intel Open Source Technology Center 4.5 (Core Profile) Mesa 20.3.4\n\nBroken: version: 2.83.12\n\nTried to download 2.83.13, \n\n\nclick on down load long term\nblender-2.83.13-linux64.tar.xz\nNot found\n\n\n" ]

[ "Rendering Error in viewport for Intel Iris and Xe GPU\nOperating system: Windows-10-10.0.22621-SP0 64 Bits (but I am using windows 11 ????)\nGraphics card: Intel(R) Iris(R) Xe Graphics Intel 4.5.0 - Build 30.0.101.1692\n\nBroken: version: 3.5.1\n\nViewport rendering errors appear on my new laptop for semi transparent objects. I assume the issue is related to the Intel Iris GPU. This issue never appeared on other PCs. All of them were using Nvidia GPUs.\nThe issue occurs when a semi transparent object is overlapping another object. See image/video. \n\n1. Use a computer with an intel iris GPU (Issue did never appear on my other pcs with Nvidia GPUs).\n2. Make sure that the Viewport is set to SOLID and the Viewport Color Type is set to Object.\n3. Create an opaque cube \"CubeA\".\n4. Duplicate the cube and make it bigger \"CubeB\".\n5. Go to Object Preferences > Viewport Display > Color.\n6. Change the alpha value of CubeB to be partially transparent.\n\nThis results in rendering errors of the overlapping objects. \nExample file and video are attached.\n\n", "rendering crashed\nOperating system:\nGraphics card:\n\nBroken: (example: 2.80, edbf15d3c044, master, 2018-11-28, as found on the splash screen)\n\n\nBased on the default startup or an attached .blend file (as simple as possible).\n\n", "Font rendered incorrectly with text object\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1060 3GB/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.94\n\nBroken: version: 3.4.1\n\nThe font is displayed incorrectly when added to 3d text object\n\nFont and blend file in attachment\n[untitled.blend](untitled.blend)\n\n[roadrage.ttf](roadrage.ttf)\n\n", "Workbench: Shadow Bug\nOperating system: MacOS Catalina\nGraphics card: Intel Iris Pro 1536 MB\n\nBroken: 2.83\n\nWhen trying to add a shadow, all screen \"cracks\" into different-shadow pieces\n\n\n\nViewport Shading> \"checkbox\" Shadow ", "Design issue: Icon next to overridden property does not perform the function it strongly suggests\n3.0.0\n\nWhen a propertry is overridden, the \"animate property\" button next to it is replaced by this \"reset\" icon:\n\nThese really seem to suggest that clicking them will remove the override. Instead it just adds a keyframe as usual. This seems misleading.\n\n - Override a property of some library override asset\n - Click the button to the right of that property\n\n", "Redo panel overlaps part of the interface\nIn the image editor, the redo panel covers part of the pixel values. \nWould be nice if such panel could be dismissed, or if it would take into account the possible presence of the pixel value subheader.\n\n[open_image_overlay.mov](open_image_overlay.mov)", "Missing Letter V and Black Screen\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: AMD Radeon(TM) Graphics ATI Technologies Inc. 4.5.0 Core Profile Context 23.7.1.230626\n\nBroken: version: 3.6.0\n\nWhenever I open Blender, the viewport is black and the letter v is not displayed in any of the menus. Here is a link to a screenshot: VZui9.png \n\nIt simply appears like this on startup.\n\n", "The display will be damaged when the node editor in Fullscreen Area mode (ctrl alt spacebar)\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: AMD Radeon(TM) R7 Graphics ATI Technologies Inc. 4.5.14736 Core Profile Context 20.11.1 27.20.12033.2007\n\nBroken: 2.93 as well as version: 3.0.0 Beta\nWorked: Haven't tested before 2.93.\n\nThe display will be damaged when the node editor in Fullscreen Area mode (ctrl alt spacebar).\n\n* Switch workspace to Geometry Nodes\n* Add a new Geometry Nodes modifier/datablock\n* Ctrl+Alt+Space to go fullscreen (not Ctrl + Space)\n* Select or move a node\n\nFile:\n[bug.blend](bug.blend)\n\nGIF：\n", "Greasepencil z fighting alpha issue\nHi,\nnot too sure if this is known issue or any fix but I am getting this alpha issue from certain angle when I have 2 different grease object. \n\n", "Hair particle system altering display percentage crashes Blender.\nOperating system: Windows-10-10.0.17134 64 Bits\nGraphics card: GeForce GTX 760/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 431.36\n\n\nBroken: version: 2.80 (sub 74)\nWorked: (optional)\n\n\nChanging the percentage value in the Viewport Display section causes Blender to instantly crash.\n\nSet the display percentage with the currently active particle system\n[field_3400_autosave.blend](field_3400_autosave.blend)", "Regression: GPU Subdivision: EEVEE doesn't refresh viewport correctly with Armature modifier\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.94\n\nBroken: version: 3.3LTS, master\nWorked: 3.2.2\n\nGPU Subdivision doestn't refresh viewport correctly an object when Armature and Subdivision Surface Modifiers are used.\n\n1) Start Blender.\n2) Add Armature object\n3) Switch to \"Shading\" workspace.\n4) Select the Cube and add `armature modifier` then `Subdivision Surface Modifier`.\n5) Switch to Properties Editor to \"Modifier Properties\"\n6) Add armature in armature modifier's object property. \n7) Switch between object-edit mode\nIt works fine when \"GPU Subdivion option\" is turned OFF (Prefrences > Viewport > Subdivision > GPU Subdivision)\n\nI added \"BeforeGPUSubDBug.blend\". You can reproduce this issue by loading the blend file and doing step 7)\n", "Wintab rarely gets into a bad state and prevents Blender startup\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070 Ti/PCIe/SSE2 NVIDIA Corporation 4.6.0 NVIDIA 536.40\n\nBroken: version: 4.0.0 Alpha\nWorked:\n\nFor a few months now, I have been getting a fairly rare, but reoccurring issue (maybe once a week on average) where Wintab gets into some erroneous state and prevents Blender from starting if the Tablet API setting of the Input preferences is set to `Wintab` or `Automatic`.\nWhen Wintab gets into this erroneous state and Blender is already open, Blender will crash if I attempt to open the User Preferences. I'm not sure what else might cause a crash when Blender is already open.\n\nBy adding extra logging to a custom build of Blender, I've determined that the crash when starting Blender occurs when calling the `open` (`WTOpenA`) function: GHOST_Wintab.cc#L85\n\nMy drawing tablet's software also reports an exception in `WTOpen`, the value of which appears to equal 0xC0000005 which I think indicates a `STATUS_ACCESS_VIOLATION` according to 596a1078-e883-4972-9bbc-49e60bebca55.\n\n\nIn the drawing tablet's software, this exception only occurs if I actually connect the tablet to my PC, whereas Blender crashes even when the tablet is not connected.\n\nCurrently, if I don't actually need my tablet, I'm working around the issue by starting Blender headless, changing the Tablet API setting to Windows Ink, and then starting Blender normally.\n\nI have already tried updating the drivers for my tablet to the newest version and it made no difference.\n\nIf I start Blender with `--debug-wintab`, all that's printed is:\n```\nPS D:\\Downloads\\temp\\blender-4.0.0-alpha+main.dd9a92785acc-windows.amd64-release> .\\blender.exe --debug-wintab\nRead prefs: \"C:\\Users\\Thomas\\AppData\\Roaming\\Blender Foundation\\Blender\\4.0\\config\\userpref.blend\"\nLoading Wintab for window 00000000000910C4\nPS D:\\Downloads\\temp\\blender-4.0.0-alpha+main.dd9a92785acc-windows.amd64-release>\n```\n\nEvery version of Blender I've tried between 2.79 and 4.0.0a crashes unless I have the Tablet API set to `Windows Ink`.\n\nI don't know the specifics of how Wintab is handled, but could it be possible to either catch the exception in `WTOpen` and allow Blender to continue loading, or to check for a connected tablet before attempting to call `WTOpen`, similar to what my tablet's software appears to be doing?\n\nUnfortunately, Wintab getting into this erroneous state where it crashes in `WTOpen` occurs too rarely for me to figure out how to reproduce it and I'm not even sure if it's Blender that is causing it to get into this state in the first place.\n\n", "object monitor and menu bar flickers when i move move my cursor\nOperating system: Windows-11-22H2\nGraphics card: NVIDIA GeForce GTX 1650/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 528.92\n\nBroken: version: 3.6.2\n\n\n\n", "Rendered object in displayed as a transparent png\nOperating system: Windows-10-10.0.19043\nGraphics card: Radeon AMD Radeon HD 8500M (4.5.14831 Core Profile Context 21.5.2 27.20.20903.8001)\n\nWorked: (3.2.0)\n\nidk if it is even a bug but when i render my object, it shows the object as a transparent png. how do i fix it?\n\n[donut.blend](donut.blend)", "Blender stops updating display after GPU spike\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: Intel(R) Iris(R) Plus Graphics Intel 4.5.0 - Build 27.20.100.8681\n\nBroken: version: 2.92.0 Beta\nWorked: I have not found a version that works as expected on this computer. I have tried 2.79, 2.8, 2.91\n\nI was premature in saying the problem is gone for [#85061 ](T85061).\n>After a few minutes of using Blender, the window freezes. Windows does not detect that it is not responding, but Blender does not respond to any input, including clicking on the close button. All elements that are drawn by Blender remain as they were at the moment of the crash except if I resize the window smaller.\n\nIf I watch the performance tab on Windows Task Manager, the freeze coincides with a 100% spike on the GPU.\nAs long as that spike doesn't happen, there is no problem, but once it does, Blender stops responding to any input, but it does not crash.\nI have updated my graphics drivers.\n[blender_debug_output.txt](blender_debug_output.txt)\n\nI have not determined exact steps. It has occurred on multiple files doing multiple different tasks. As far as I can tell, there is no correlation between what I am working on or what task I am performing, and the error occurring.\n" ]

[ "Graphics glitch in viewport; making window very thin works fine\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: Radeon 535 Series ATI Technologies Inc. 4.5.13497 Core Profile Context 23.20.826.3072\n\nBroken: version: 2.82 (sub 7)\nWorked: (optional)\n\nThe content is displayed incorrectly on the screen. when you try to squeeze the application window horizontally, the content starts to display correctly. any settings in the application itself do not help in solving the problem. attempts to start the application with updated drivers did not give any result.\n\nOpening blender gives this\n\n\n" ]

[ "Crash after copy pasting a material onto itself and switching to render\nOperating system: ArchLinux\nGraphics card: NVIDIA 3090\n\nBroken: Current main.\nWorked: Never.\n\n\nBlender crashes when copy pasting a material onto itself (via UI), then creating a new material slot, and then switching to the rendered view.\n\n1. Open a new scene in blender.\n2. Select the cube and go to its material tab.\n3. Next to the material list, on the right, click the down arrow and do Copy Material.\n4. From the same menu now do Paste Material.\n5. Add a new material slot using the + in the materials tab.\n6. Switch to rendered view.\n\nAttached a crash stacktrace taken with gdb, and main build with debug.\n\nIf it's useful, the crash was taken with a manually built blender, so it was a factory startup with Eevee as a renderer, but it also crashes when using Cycles.\n", "Flashes in Eevee render with Mantaflow viscosity enabled and Motion Blur\nWindows 7 Pro 64-bit\nNvidia GeForce GTX 1060 6gig\n\nBroken: Blender 2.92.0 alpha / c7e92e379d62 / 2021-01-02\n\nExperimented with the new viscosity setting in Mantaflow.\nRendered the test with Eevee and ended up with output where the surface of the fluid was occasionally 'flashing' in certain frames.\nMy only suspicion is that something it going wrong with the normals of the mesh created by Mantiflow.\nPlease see attached video clip and Blend file.\n[viscosity flashes.mp4](viscosity_flashes.mp4)\n\n- Open attached file\n- Bake simulation (maybe you need to set `modular` and `Is Resumable`)\n- Bake Mesh (maybe you need to set `modular` and `Is Resumable`)\n- Render Animation in Eevee\n\n[viscosity flashes.blend](viscosity_flashes.blend)\n\n\n", "Normal map bug in Blender 2.9 / 3.0\nOperating system: Windows 10\nGraphics card: NVIDIA Geforce GTX 1050 Ti\n\nBroken: Blender 2.93 / 3.0 (stable version)\nWorked: Blender 2.83 (stable version)\n\nWhen applied a precedural texture (noise, Voronoi etc.) as a normal map in Node editor, the boundaries of the polygons stand out and it will look like a block. \nThere was no such problem in 2.83.\n\nCreate a mesh. Add a procedural texture on the Shading tab and apply it as a normal map.\n[#94139.blend](T94139.blend)\n\n\n\n", "Broken material assigment after changing mesh-linking to object-linking\nWindows 11\nNvidia RTX 2060 super\n\nBroken: blender-3.2.0-stable+daily.e05e1e369187\n\n\nOn objects with multiple material slots, the currently selected slot gets bugged when switching from object-linking to mesh-linking. It allows the user to select a different material for each link type, which seems to trigger the bug and has lingering effects. For example, in geometry-nodes setting the material becomes unpredicatable, with material names seemingly swapped.\n\nFrom the provided blend file or startup scene:\n1. create a new material slot\n2. pick a material for this slot\n3. swap the material link type between object-link and mesh-link\n4. select a different material for the same slot\nThe material or slot is now bugged. Geo nodes 'set material' will become unpredictable.\n\n[bug_slots.blend](bug_slots.blend)", "Eevee Motion Blur not applied to AOVs\nOperating system: Ubuntu 18.04\nGraphics card: GTX 2080 Super\n\n2.92 Official\n\nWhen rendering with eevee, motion blur is only applied to the beauty pass. Other passes such as cryptomatte or normal dont have mblur applied, but you can see the steps of the accumulation blur in some passes. This is a huge problem for compositing, it makes all the passes mostly useless.\n\n- animate an object\n- turn on passes\n- render with eevee with mblur on\n- check the aovs, there will be no mblur on them\n\n[aov_motion_blur.blend](aov_motion_blur.blend)", "Wrong result mixing 2 nodes of an image texture\nOperating system: Windows 10\nGraphics card: AMD\n\nBroken: 3.3, 2.93\nWorked: unsure\n\nWhen using 2 nodes of the same texture and connecting the Color output from one and the Alpha from another, the result looks different in comparison when using both outputs from one node. Broken in EEVEE\n\n1) Open the file, switch to material preview/rendered\n2) Reconnect the color (or the alpha) link to another texture\n[Bug1.blend](Bug1.blend)", "Texture properties not working with OSL enabled and unpacked textures\nOperating system: Linux-5.11.7-arch1-1-x86_64-with-glibc2.33 64 Bits\nGraphics card: Mesa DRI Intel(R) HD Graphics 520 (SKL GT2) Intel Open Source Technology Center 4.6 (Core Profile) Mesa 20.3.4\n\nBroken: version: 2.92.0\nBroken: version: 2.93.0 Alpha\n\nI have a texture with RGB channels set to 1 and alpha set to 0. When trying to use this texture with OSL, the alpha property (at least) of the image texture node does not work when the texture is unpacked. I set it to channel packed, expecting the color data to be always white. The problem is that when the texture is unpacked, the color data is black.\n\n1. Open the attached file.\n2. Set viewport shading to rendered.\n3. The plane shows as white (as it should).\n4. Unpack the texture.tga file.\n5. Toggle between preview and rendered viewport shading to refresh.\n6. The plane now shows as black.\n7. Pack the texture again.\n8. The plane shows as white again.\n\n[bug.blend](bug.blend)", "Remesh Modifier in Voxel mode produces surface artifacts when Adaptivity enabled\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce RTX 2080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.71\n\nBroken: \n - 2.83.9\n - 2.91.0 Beta\n - 2.92.0 \nWorked: never\n\nRemesh modifier produces artifacts on mesh which look like a result of bad triangulation.\n\n1. Open attached blender file\n2. Open modifier panel and select testMesh\n3. Change adaptivity value from 1 to 0 cm and back\n4. Observe black shading artifacts \n[remesh_bug.blend](remesh_bug.blend)\n", "A combination of shader nodes causes lack of material emission at any brightness (Cycles)\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 528.49\n\nBroken: version: 3.4.0(tested as well on 3.3.0 and 3.5.0 alpha)\nWorked: (none)\n\nA difference of bevel node and normal coordinates produces no light at any emission level. Clipping the result due to suspicion about negative brightness has no result.\nNote: adding any value to the result adds corresponding emission, but despite visuals, brightness from combination above does not.\n\n\n- Open attached file\n[cycles_where_is_emission__bug.blend](attachment)\n\n\n", "Rendering in cycles causes objects with materials to go black and shiny when using the GPU\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: AMD Radeon RX 7900 XT ATI Technologies Inc. 4.5.0 Core Profile Context 23.7.1.230626\n\nBroken: version: 3.6.1\nWorked: N/a\n\nWhen my render engine is set to cycles and device is set to GPU Computer (with HIP and my 7900 XT selected), if I look at an object with a material applied in either viewport shading or as a rendered image the model will appear black and shiny or occasionally will just show as a flat blue colour. \n\nThis issue does not occur when using the CPU as the device instead and also did not begin to occur until a few days ago (prior to which it was working fine).\n\nI've checked the same blender file on a laptop with a 3060 GPU which worked correctly so I don't think the file itself is corrupted.\n\nI have also checked multiple blend files, some of which work fine, others do not, one file has some things working and other things broken (I will attach it to this bug report)\n\n\n1. Load the attached blend file \"ArkenV1R1_Interior.blend\".\n2. Go to edit > preferences > system > cycles render devices, set it to HIP with RX 7900 XT selected.\n3. Under scene settings, change the render engine to cycles and device to GPU compute.\n4. Switch to Viewport shading.\n5. Observe that the bullets and floor panel are shown as black and shiny while the table renders correctly (grey and blue).\n6. Press F12 and observe that the rendered image also shows the bullets and floor panel as black and shiny.\n\n\n**Additional**\nIncorrectRender.png was rendered using GPU\nCorrectRender.png was rendered using CPU\n\nI will be happy to provide anything else needed to figure this out, thanks!\n\n", "Attribute Node Doesn't Function Under World Shader Context\nOperating system: Windows 10 19044.1826 64-bit, Linux 5.18.14.arch1-1\nGraphics card: NVIDIA GeForce GTX 1080 Ti\n\nBroken: 3.2.1\nWorked: never\n\nThe `Attribute` node (both in Cycles or Eevee) doesn't work under World materials.\nAlso the `Geometry` and `Instancer` types in the node are irrelevant in the Word context and should be hidden.\nMaybe the `Object` type should be renamed to 'World', but it currently doesn't seem to be able to pass world properties as uniforms.\nThis essentially means that:\n - There are redundant elements in the World shader node editor that currently serve no purpose\n - There is no way to pass custom uniform attributes to World shader nodes, although they should be available (similarly to how it is done for objects)\n\n\n\n- Open attached .blend file\n- Note that the \"test_color\" attribute of the World is not read by the material.\n[AttributeWorldShader.blend](AttributeWorldShader.blend)", "Weird artefacts from emission surface\nOperating system: MacOS 12.6.5\nGraphics card: AMD Radeon Pro 5500M (but this occurs also on CPU)\n\nBroken: 3.6.2\nWorked: don't know\n\n\nIn the attached .blend file I have set up an emission shader to emit light from one end of a long thin surface. The screenshot illustrates the problem: a bunch of weird artefacts appear.\n\nIf I change it to emit from the other end of the surface, by swapping the 'Map Range' node in the screenshot for the disconnected one, the problem goes away.\n\n\n\n\nLoad the attached .blend file and switch the view to rendered.\n\n", "Material setting \"show_transparent_back\" (Show Backface) gives weird results\nOperating system: Windows-7-6.1.7601-SP1 64 Bits\nGraphics card: GeForce GTX 980/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 461.40\n\nBroken: version: 2.92.0\n\n\nMy materials are messed up by having blend mode set to \"Alpha Blend\" and \"Show Backface\" set true.\n(The checkbox for \"Show Backface\" pops up if a materials blend mode is set to Alpha Blend\")\n\nIs this normal behaviour?\nThe intend here was to see every side of the cube with a bit of transparency.\n\n[ShowTransparentBackBug.blend](ShowTransparentBackBug.blend)\n1. Add a Cube (subdivide a few times) and give it a material\n2. Go to material settings and set the materials...\n- Blend mode to \"Alpha Blend\"\n- Backface Culling to False\n- Show Backface to True", "Viewport: anti-alising on normal lines\nNormals on faces/vertices are not antialiased:\n\n\n\nTo see them, enable them at the bottom of the 3D Viewport Overlay popover:\n\n\nThis issue is similar to what was covered in #107394", "compositor blur node set to fast gaussian produces black output when fed a colour output from an EXR image.\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.6.0 NVIDIA 536.99\n\nBroken: version: 4.0.0 Alpha\nWorked: (4.0 alpha from a few weeks ago not sure of version as deleted now)\n\nCompositor blur node set to fast gaussian produces black output when fed a colour output from an EXR image.\n\n1. Open a new blend file.\n2. Go to the compositor.\n3. Drag in the image file attached below.\n4. Connect the resulting image node's 0_Image output to a viewer node and turn on the backdrop.\n5. Image will be shown.\n6. Now place a blur node between the image node and the viewer node.\n7. Set the x and y values of the blur node to 1 and the size to 1.\n8. Image will blur.\n9. Now change the blur mode to fast gaussian.\n10. Image will go completely black.\n\n[test-image](https://projects.blender.orgattachment)" ]

[ "NormalMap input causing Material Viewport issue in dual 3DView Layout\nOperating system: Ubuntu 17.10 and Windows 10\nGraphics card: Nvidia GTX 1050 and 1060\n\nBroken: 2.80, 69b2f5268114\n\nWith a dual 3DView layout, switching from Solid to Eevee shading, will cause the second 3D View to not draw the material properly, if its shader has a normal map input.\nPrincipled shaders draw white, Glossy shaders draw black.\n\nIt seems to be specifically the Normal Map node, that is the cause here, and specifically the Tangent Space,\n\nVideo demo: watch?v=xhMjiKs7LPI\n[Blend file ](machin3_material_update_dual_3dview.blend)." ]

[ "Theme: Inconsistent solid color for the background in the 3D viewport\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1070 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 445.75\n\nBroken: version: 2.90 (sub 0)\nWorked: 2.79b\n\nThe color of the 3D viewport background set to theme seems to not match with the color set in the theme setting.\n\nCheck the examples below with the color #36393F\n\n2.79b: color is exactly like in the theme setting.\n\n\n2.80 to 2.82: color looks good but seems grainy.\n\n\n2.83 to 2.90: color is slightly changed, when sampled not the expected theme color.\n\n\n- Open blender\n- In the initial layout change the Outliner editor to Preferences\n- Go to background color option\n- Use the colorpicker to get a sample of the background color\nThe color shouldn't change, but it does", "Alpha bug on Principaled BSDF with GPU\nOperating system: Windows-10-10.0.19043-SP0 64 Bits\nGraphics card: NVIDIA RTX A2000 Laptop GPU/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 512.92\n\nBroken: version: 3.2.1\n\nDecreasing alpha makes material black in the Principaled BSDF shader when using GPU compared to CPU \n\n\n\n - Decrease \"Alpha\" on the Principaled BSDF shader\n - Change to device to either \"GPU Compute\" or \"CPU\" \n - Adjust alpha value to discern difference if not apparent when switching Device property\n\n\n**Notes**\nCPU appears to work correctly with this value, it's only on GPU on my machine that causes alpha to make the material black. Bug appears in rendered images too and not just viewport, and is present in both CUDA and Optix. \n\nMake my alpha 0% if duplicate.\n\n[Blender GPU Alpha Bug.blend](Blender_GPU_Alpha_Bug.blend)", "Screw Modifier Generate Incorrect Normal for Anisotropic Shading\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 531.41\n\nBroken: version: 3.5.0\n\nScrew modifier seems generating incorrect normal when used with anisotropic shading/bsdf. The only workaround i can find is to used Edges + Screw Modifier and convert it to mesh which kinda destructive.\nI'm not sure if this a bug or limitation.\n\n", "Knife Tool produces bad geometry in certain cases (disconnected edges/vertices, overlapping faces)\nif i use tools knife i see this bug.\n\n\n\n", "Artifacts when composing render passes with SSS, EEVEE and SSR\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.94\n\nBroken: version: 3.3.0 Beta\nWorked: I'm not sure this behavior is new. Blender 3.2 displays the same behavior.\n\nWhen compositing render passes that have been generated using EEVEE, and an object has Subsurface Scattering enabled, and Screen Space Reflections are enabled, and Half Res Trace is disabled, artifacts are generated. The combined pass, however, correctly renders.\n\n\n*What I get (demo file included below)*\n\n\n*What I should get*\n\nNote: if Half Res Trace is enabled, the problem disappears.\nAlso, if Cycles is used to render the same scene, the passes do not display these artifacts.\n\n\n\n\n- Create a file that uses EEVEE as renderer.\n- Enable Screen Space Reflections. Disable Half Res Trace.\n- Create an object with Subsurface Scattering\n- Enable the following necessary render passes: Diffuse Light, Diffuse Color, Specular Light, Specular Color\n# In compositing, mix as follows the passes: multiply Diffuse Light and Diffuse Color; multiply Specular Light and Specular Color; add them.\n\nChanging the roughness can make the effect more extreme. Having a very contrasty roughness map exacerbates the problem:\n\n\n*a contrasty roughness map*\n\n\n*the render showing the problem more extremely*\n\n\n*how it should be*\n\nThe final result should be akin to the one posted above. You can compare the composited result with the one that is already combined: the combined one does not display the issue. Using Cycles for the same setup also works without problems.\n\nThe inconsistency of results mentioned above makes me believe this is a bug, and hence my report.\n\nDemo file: [sstest.blend](sstest.blend)", "dual monitor will sometimes cause blender to freeze\nOperating system:win10\nGraphics card:2080s\n\nBroken: 3.4.1\nWorked: \n\ni have two monitors (4k and 2k) scrolling timeline or doing anything on one monitor and displaying cycles render viewport on the other monitor will sometimes cause blender to freeze forever\n\nit is not specific to this scene it will freeze on all kinds of scenes, might be related to the muti-monitor bug i have been reporting T96248\nhere is a screenshot of what it looks like when freezing \n\nmake some animation\nhave 2 monitors and use cycles rendered view on one monitor and scrolling timeline on the other monitor", "3D Texturing: Limit Eevee samples to 1 during paint stroke.\nGhosting effects appear when painting. We should limit the number of samples to improve the performance and reduce the ghosting effect.", "Texture Paint: Wrong active Texture Slot\nOperating system:win10\nGraphics card:GTX 740m\n\nBroken: version: 2.92.0, branch: master, commit date: 2021-02-24 16:25, build date: 2021-02-25, 09:31:14, hash: 02948a2cab\n\nVpqy7BmMKaA\n\n- Open attached file or:\n - Go to \"texture painting mode\" (in default workspace)･\n - Create 2 base color textures\n - Open Shader Editor\n- In the Shader Editor, select non -active texture (\"Material Base Color 2\")\n- in the View 3D, paint texture\n- Nothing happen in visual.\n- The other texture slot is changed instead of current active texture.\n[tex_slot_problem.blend](tex_slot_problem.blend)\n\n", "Looks like an eevee limitations with UV maps and attributes\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 522.30\n\nBroken: version: 3.5.0 Alpha\n\nWhen working with multiple UVs and attributes, eevee disables some normal/bump maps.\nThere is an error in the console:\n```\nGPU_material_compile: Attribute \"nor\" overlap one or more index from another attribute. Note that mat4 takes up 4 indices. While merging eevee_legacy_material_surface_frag_opaque\n```\nIn 3.4, the shader cannot compile at all (pink texture).\n\nThis is not documented anywhere and it is not clear to me how it works, because until I connect nodes with attributes in Base Color, all normals are displayed.\n\n1. Open .blend file.\n2. Turn on the eevee render.\n3. Look, all normals are displayed.\n4. Connect nodes with attributes to Base Color.\n5. One normal map is missing. An error appeared in the console.\n\n", "Eevee 'Principled Normal' connected to 'Normal Map' node is disabling 'Displacement'\nOperating system: Win 10\nGraphics card: 2x1080Ti\n\nBroken: blender-2.91.2-windows64, blender-2.93.0-420f538fadfd-windows64\nNever worked as long as I remember\n\nBasically connecting normal map to principled shader is disabling any displacement map connected to the output. Just unplugging normal from principled turning it on. I don't know if its a bug or is it intentional \n\nNormal and displacement plugged in - displacement not working\n\nNormal unplugged - displacement working\n\n\n- Open attached file\n- Compare the result in Eevee with Cycles\n[bump_test2.blend](bump_test2.blend)", "Mix/Add nodes breaks Eevee AO render pass if one socket is unfilled.\nOperating system: Windows-10-10.0.16299-SP0 64 Bits\nGraphics card: GeForce RTX 2070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 419.67\n\nBroken: version: 2.92.0, 2.93.0, 3.0 0efb627bbd\n\nIn Eevee, any material which contains an add shader node will be broken in the AO pass, if second socket is *invalid*, leaving the material either entirely black or white.\nThis is annoying if you're creating entirely black materials by leaving a socket empty.\n\n- Add a mesh, give it a material.\n- Put either an an add shader node or a mix node into the material somewhere, but only fill one of the input sockets\n- Enable AO with Eevee, and look at the AO render pass in render view. In 2.92, the object will be completely black, in 2.93 it will completely white.\n\n----------------\n- Render the scene with the file given.\n- For the second case, render by swapping the socket inputs of the \"Add Shader\" node.\n[T88779_mine.blend](T88779_mine.blend)\n\n\n\n\n\nAdditional example images.\n(2.93)\n[F10155389](image.png)\n[F10155391](image.png)\n\n(2.92)\n[F10155385](image.png)", "error in custom view3d.move Keymap Setting\nOperating system: Windows11\nGraphics card: It doesn't matter\n\nBroken: up to 3.5.1\n\n\n\nBecause I often use the 2D features in Blender, and the pan view function in Blender is not very convenient (requires using Shift + MMB), I made the settings as shown in the picture to add a new pan view keymap to Spacebar + LMB. In the normal object mode, I can indeed pan the view by using the new shortcut, but once I switch to any other mode, this shortcut key will be disabled. \n\nBased on the default startup or an attached .blend file (as simple as possible).\n\n", "Texture paint in viewport is not using the color space of the image.\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: AMD Radeon RX 5700 XT ATI Technologies Inc. 4.5.14802 Core Profile Context 22.2.1 30.0.14023.7007\n\nBroken: version: 3.2.0 Alpha\n\nIf you paint a noncolor texture like a normal map, painting it directly in the 2D image correcly uses the non-color space, but painting in the viewport doesn't.\n\n\n- With any non-color texture, try to paint both in viewport and in the 2D image\n[2022-02-15 10-34-07.mp4](2022-02-15_10-34-07.mp4)\n\nAditional note: This seems to happens when you sample from the 2D image. By contrast, if you sample the color from the 3D view, you can correctly paint on it, BUT in the 2D viewer the color becomes brighter.", "Unwanted behavior when modifying the normal vector of the refraction node\nOperating system: Windows 10\nGraphics card: RTX 3090\n\nBroken: Blender 3.3\n\nA subtle but sharp bending effect can be seen on refractive material with a modified normal vector.\n\nWhen modifying the \"normal\" vector of the refraction node (or any node with refraction), when viewing at a certain angle, a \"sharp\" bend appears.\n\nA series of bends in the red circled area can be seen where it should be a smooth or straight line.\n\nIt's more obvious if the normal vector is not normalized, but it should be normalized anyway, or the refraction/reflection calculation will not work correctly:\n\n\n\n**Possible causes**\nThe \"ensure_valid_reflection\" could be the cause.\n\nThe commit that added \"ensure_valid_reflection\" is to fix the reflection black spot issue when using a normal map.\nrBd6e769d32e7939e3bbd1986cdc4abd2b13135eab\nWhere the normal vector is being regularized so the reflection ray will not penetrate the surface.\n\nThe code \"ensure_valid_reflection\" is currently here:\nbsdf_util.h#L109\n\nIt is being used in many places where \"refraction\" is present and modifies the normal vector:\nclosure.h#L105\n\nI couldn't track down where exactly causes the issue, but I can show the bending effect aligns where \"ensure_valid_reflection\" kicks in.\n\nThe sharp line that appears on the \"reflection\" side is expected; without it, the ray will penetrate the surface.\nWhile on the \"refraction\" side, the ray should not need to be \"ensured.\"\n\n**Expected behavior**\nThe ray should behave the same as there is a surface with the same normal vector.\n\nNote:\nIn most cases, this doesn't make a significant appearance difference.\nBut this could open a lot of creative uses where light rays can be directed to any angle procedurally;\nFor example, a procedural fresnel lens surface can be used to create something like magnifiers.\n\n- Open attached blend file\n- Enable `Viewport Shading` `Rendered`\n[RefractionBendingEffect.blend](RefractionBendingEffect.blend)\n\nThe bending effect can be seen in the middle\n\nI have provided two mix nodes for quickly examining normalize/non-normalized normal vector, and refraction/reflection behavior.\nNote that when switching to a non-normalized vector, the bending effect will appear on a different viewing angle.", "Tangent space normal mapped normals change inappropriately with deformation\nOperating system: Windows-10-10.0.19042-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 511.65\n\nBroken: version: 3.1.2\nBroken: version: 2.83.0\nBroken: version: 3.2.0 Beta\nI don't think this ever worked right.\n\nCertain deformations, on certain meshes, lead to inappropriate output from normal mapping: output changes even when normalized normal, tangent, and normal map color remain the same.\n\nI'm attaching a file demonstrating the issue:\n\n[normalmapbug.blend](normalmapbug.blend)\n\nThis file contains a smooth shaded cube, with top and bottom faces deleted, with packed baked tangent space normal map. The cube has a shapekey which scales one (obscured, +Y) face. It contains a plane, from which the normal map was generated (selected plane to active cube.) There is a circle mesh and a camera so we can focus on the central sample of the cube's -Y face, which has properties which remain invariant under our shapekey deformation: normalized normal, tangent, and normal map color will all remain invariant. I have a world, but haven't packed the image to save space; it's just a default color grid, created so we can keep track of the direction the normal is pointing more easily.\n\nWhen we look at the normal without any deformation, we see that the reflection vector is pointing at the north pole, as it should, since we baked from a plane rotated 45 degrees from the horizontal:\n\n\n\nWhen we set the shapekey (shown on right), we see that the normal map output has rotated toward the unmodified normal, by about 12 degrees:\n\n\n\nNormalized normal for central sample is 0,-1,0 (interpolated from +-sqrt(2)/2, -sqrt(2)/2, 0 or from nearly +-0,-1,0 under deformation; X component cancels out for central sample). Tangent is 0,0,1 in both shapes. Normal map color, of course, remains the same. The only thing that changes here is the *un-normalized* normal, interpolated from vertex normals, which changes from 0, -sqrt(2)/2, 0 to 0,-1,0, growing about half again as long.\n\nThere's nothing magical about the central sample. It just happens to be my spherical cow here, which demonstrates the issue with the fewest complications. All samples are being affected in similar ways.\n\nWhat appears to be happening here is that Blender is not using an orthonormal tangent space. The interpolated normal changes length with deformation, and it does so at a different rate than the interpolated tangent. This means that Blender is not evaluating normal maps in a consistent space, but one that is non-uniformly scaled (and sheared, at least in the case of Eevee) on the basis of variables that should be unrelated. The normal map waxes and wanes in strength as the local area changes pointiness, although how big of an effect this has should depend on the specific orientation of the tangent-- if we were to rotate our tangent 30 degrees, we'd see a different effect, still not right, but different. In this way, this makes the normal map also dependent on specifics of the tangent that shouldn't matter.\n" ]

[ "Adding Normal Map to object in EEVEE results in emissive material\nOperating system: antergos (Arch current)\nGraphics card: GTX 770\n\nBroken: 2.80, 6d89337257b6, 2018-12-24\n\nAdding Normal Map node to Normal slot in Principled BSDF produce emissive material on Text object in EEVEE. Converting Text to Mesh doesn't change this behavior.\nSame material used on different mesh works properly. Also it is working ok in Cycles.\n\nUsing different shader (Diffuse, Glossy, SSS, Translucent) produce black material.\n\n1. Add Text object.\n2. Create new material.\n3. Add Normal Map (Tangent Space) node and connect to Normal socket in Principled BSDF.\n\n[text_normal_eevee_bug.blend](text_normal_eevee_bug.blend)", "Normal Map on Tangent Space makes some objects emissive\nOperating system: Windows 10 64x\nGraphics card: Geforce 1070\n\nBroken: 957b4547aeeb\n\n\n\nI have been having this problem for a while:\n\n\nThe normal space seems to be the problem, as i change it, things stop glowing : \n\n\nSometimes it only shows on render and not on the viewport : \n\n\n\nBy changing the Normal Space on tha black material, it doesn't turns into emissive anymore \n(there is still on emissive object but it might be another material)\n\nI have made a report about this previously, but i am updating with new images and info about the Normal Space, since it has not yet been fixed.\n\nI have saved a simple scene with only the objects using this material:\n\n**Viewport**\n\n\n\n**Render**\n\n\n[glowing normal.blend](glowing_normal.blend)\n" ]

[ "Keyed Particles don't render as expected\nOperating system: Windows-10-10.0.17763 64 Bits\nGraphics card: Intel(R) HD Graphics 4000 Intel 4.0.0 - Build 10.18.10.5069\n\nBroken: version: 2.81 (sub 16)\nWorked: (optional)\n\n[KeyedParticlesJL.blend](KeyedParticlesJL.blend)\n\n- Open file\n- Go to frame 105\n- Render\n- Compare render to viewport\n\nRender is different to viewport\n\nThat's what I expected:\n[ViewportRender.mp4](ViewportRender.mp4)\n\nThat's what I get when rendering with Eevee:\n[EeveeRender.mp4](EeveeRender.mp4)\n\nThat's what I get when renderng with Cycles:\n[CyclesRender.mp4](CyclesRender.mp4)\n", "Extremely frequent crashing when using the compositor. Especially when using the glare node.\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3070 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 536.99\n\nBroken: version: 3.6.2\nWorked: I believe it was far more stable before the realtime compositor implementation. (Blender 3.3-ish)\n\nBlender crashes extremely frequently when using the compositor on my system, especially when the glare node is being used at all. Does not seem related to vram or system specs, and looking through logs has not come up with any errors - it just closes.\n\nAdjust compositor glare node values with the viewer enabled. It seems to crash within a minute of doing so, usually sooner.\n\n", "Viewport Render (and Render Animation) color looks completely off from the Viewport\nOperating system: Darwin-19.0.0-x86_64-i386-64bit 64 Bits\nGraphics card: AMD Radeon Pro 560X OpenGL Engine ATI Technologies Inc. 4.1 ATI-3.2.24\n\nBroken: version: 2.82 (sub 6)\n\nRendering the viewport animation gives a very different result from the actual viewport.\n(Also: Viewport Anti-Aliasing isn't taken into consideration on the Viewport Render).\n\n||\n| -- | -- |\n\nThe problem is more noticeable when the Viewport Shading is set to \"Flat\", even though it also shows in \"Studio\". It delivers very dull colors with less contrast and a blue-ish tone.\nChanging between Cycles and Eevee doesn't affect the Viewport Render (from my tests).\n\n||\n| -- | -- |\nflat:\n||\n| -- | -- |\n\n- Open attached file.\n- Go to View -> Viewport Render Image\n[Viewport Colors.blend](Viewport_Colors.blend)", "Cycles and Eevee Significant Difference in Texture Output\nOperating system: Ubuntu 20.04\nGraphics card: GTX 1060 3Gb\n\nBroken: 2.80 and on (3.0.0 is still affected)\nWorked: 2.79\n\nFrom Blender 2.80, Cycles produces incorrect output for some procedural textures, while Eevee produces correct ones.\n{[F12696991](Cycles.png)}\n\nTest file (I assume there might be other files affected too). [Carbon_90.blend](Carbon_90.blend)", "Viewport Selection is getting slow depending on Scene Hierarchy depth / amount of objects\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 531.61\n\nBroken: version: 3.6.0 Beta\nThis was also tested in 3.5.1 / 3.4 (not working as fast as expected)\n\nSelecting objects in the viewport and in the outliner is taking seconds and produces high CPU load.\nThe spikes on the CPU graph are only produced by selecting a geometry (single left click on geometry in the viewport).\n\nThe scene from the screenshot contains an agency export model with a lot of Emptys that are used for the scene hierarchy / CAD. 33K objects (22K emptys) / 55 Mio Triangles.\nAfter deleting the emptys the selection is getting a little faster but still lacks behind. (from 3 to 1.5 seconds)\nIn Isolation view (1 single mesh) deselecting and selecting a mesh again it also takes around 1 second to select the mesh again. In edit mode selecting a vert / edge is instant as usual.\nThe geometry had Custom Split Normals Data but removing the normals made no difference. \n\nTurning off Emptys selection in The outliner also brought not the same feeling like selecting in Edit Mode. \n\n- Open attached [15000_empties.blend](attachment) file (It is a simplified one)\n- Select any object from the 3D View\n\nNote: The more Empties hidden, the faster the selection becomes.\n\nThanks a lot making Blender this awesome <3\n\n[15000_empties.blend](attachment)", "Solid mode in EEVEE looks pixelated (NEW)/ CYCLES in GPU 3D AMD driver problem (OLD KNOWN Problem)\n**Xeon E3 1230 V2 - 16Gb RAM**\nOperating system: Windows 10\nGraphics card: AMD RX 570 8Gb VRAM\n\n**Blender Version 2.92.0**\n\n\nI opened an old work of mine that I was mading on same 2.92.0 before. Today when I opened the solid mode as default and appears very pixelated and like matrix effect, as AMD error was giving in Cycles before... \nMaybe is AMD driver problem or materials render data. \nSome months ago was Cycles incompatibility. Cycles is still working only on CPU and now Eevee have problems in solid mode. \nI've tested in 2.83.5 version and worked. \n\n\nattached photos. In 2.92.0 isn't \n\n**Work in a project\nfews days later (about 20 days) open the archive and solid mode was pixelated  \n\n[estilizada unir cabeca retopo.blend](estilizada_unir_cabeca_retopo.blend)\nBased on the default startup or an attached .blend file (as simple as possible).", "Eevee slower than cycles when using a lot of particles or collection instances\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce RTX 2070 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 441.66\n\nBroken: version: 2.83 (sub 17)\n\nWhen using a lot of collection instances with particles in them, or with a lot of objects, Eevee seems to be very slow in evaluating the scene. \neevee: 2min and 2sek \ncycles (gpu): 9sek\n\nopen the file switch to cycles - hit render\nswitch to eevee hit render\n", "EEVEE crash while rendering animation, cycles works.\nOperating system: Win10\nGraphics card: GTX1070MAX-Q\nCPU : i7-6700HQ\nRAM : 32GB DDR4\n\nBroken: 2.90\nWorked: -\n\nBlender crashes w/o any errors when I started to rendering animation in Eevee. \nIt happens on 2 laptops and it always happen in 154 or 155 frame while rendering.\nIn attachment you can download the project which gets this error. \nWhen I use cycles rendering it works perfect.\n\n[#80740.blend](T80740.blend)\n\n- Open file\n- Render frames 154 and 155\n", "Crushed Velvet Eevee render problems in Cycles\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce RTX 2070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 445.75\n\nBroken: version: \n- 2.93.0 Alpha\n- 2.82 (sub 7)\nWorked: ?\n\n\nThe *Crushed Velvet* material from the *Material Library* does not render properly with Cycles when setting the device to *GPU*.\n\n- Load the simplified material attached below\n- Set the render device to *CPU* and render. The output should be all black\n# Set the render device to *GPU* and render.. Depending whether you have only GPU(s) activated in the preferences or also CPU(s) you should see one of the following. When only GPU(s) are enabled you should get a proper render. When CPU(s) are also enabled you should get dark render tiles that are rendered on the CPUs.\n\n[VelvetTestSimplified2.blend](VelvetTestSimplified2.blend)\n\n---\n\nOriginal description:\nI'm attaching the .blend file I used to test this issue. I'm not even sure this is a bug or simply me not being overly familiar with materials yet, but it feels like a bug. An dit may be a Blender bug and nothing to do with this add-on. Obviously, this material renders fine in Eevee, but it also renders fine in Cycles when using CPU to render, but has major artifacts when using GPU to render. I've attached a render to show what I'm getting in case you don't experience it.\n\nThe steps I took to get here are, \n\nBasically, I have a default scene I use to do my modeling or as a start point. It's very basic. I was experiencing this issue in another project I was working on and noticed it went away when I changed the material on one object. \n\n- I loaded up my basic start file\n- selected the sphere\n- Scrolled down to the Material Library VX panel in the Material Properties tab\n- Loaded the Sample Library\n- Chose the Crushed Velvet Eevee material and clicked Apply to Selected\n\nThat's it. I had initially thought it was simply a problem rendering in Cycles but for my first render, I forgot to switch from CPU to GPU and noticed it rendered fine. So I did a few more tests and it renders fine with CPU every time, but not with GPU.\n\n[VelvertTest.blend](VelvertTest.blend)\n\n", "EEVEE - Screen space reflections artifacts\nOperating system: Linux-5.3.0-7625-generic-x86_64-with-debian-buster-sid 64 Bits\nGraphics card: GeForce GTX 1080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 440.44\n\nBroken: version: 2.83 (sub 2)\nWorked: ?\n\nSSR shows some strips when I have bad data for normals.\n\nOpen the file:\n[ssr_sripts.blend](ssr_sripts.blend)\n\nIf you click anywhere multiple times, viewport gets darker and darker.\n\nBefore:\n\n\nAfter clicking the glossy node header a few times:\n\n", "Subdivision Surface modifier performance issues\nOperating system: Xubuntu 18.04\nGraphics card: Nvidia Titan X (Pascal)\n\nBroken: 2.80 26d5a3625ed\nWorked: Before the new \"Quality\" setting was introduced in the Subdivision Surface modifier\n\nThe Subdivision surface modifier has slowed down significantly, especially noticable with multiple Subdivision Surface modifiers on the same object.\nCreating the same setup in the builds before the new setting is as fast as 2.79.\n\n- Create Monkey object\n- Add Subdivision Surface modifier\n- Add Wireframe Modifier\n- Add second Subdivision Surface modifer\n[slow-subsurf.blend](slow-subsurf.blend)", "Poor performance in Timeline, Dope Sheet, etc when at the top of the editor.\nOperating system: Linux-5.3.0-7648-generic-x86_64-with-Ubuntu-19.10-eoan 64 Bits\nGraphics card: GeForce GTX 1080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 440.100\n\nBroken: version: 2.79, 2.83, 2.90.0 Alpha\nWorked: unknown, possibly never.\n\nSeems like some animation files cause really poor performance in Dope sheet when trying to scroll on top of it. Scrolling down and making the top section not visible seems to fix the performance. \nI presume the editors draw keyframes for channels on top of each other without checking if the key already exists.\n[2020-07-16 17-54-48.mp4](2020-07-16_17-54-48.mp4)\n\n1. Open attached file\n[slow.blend](slow.blend)\n2. Try to click around on timeline, dope sheet editor.\n3. Open the key dropdown in dope sheet and scroll down. Notice how performance goes back to normal.\n\n", "Motion Blur missing in the Render Preview of viewport\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 451.67\n\nBroken: version: 2.91.0 Alpha\n\nIn 2.83 we had the access to realtime motion blur with Eevee (which was limited to camera deformation).\nIn 2.90 and above this was replaced with a new motion blur system that is render only. \nEevee is amazing for having realtime screen space reflections, ambient occlusion, depth of field and motion blur, which made it perfect for screen capture or viewport rendering an animation without excessive time consumption.\nI was wondering wether this is a decision or a bug, since having the camera deformation motion blur back in the viewport would allow for the same realtime rendering results as in 2.83\n\n- Go to camera view\n- move the camera around\nIin 2.90 and above and youll see any motion blur previously existent in 2.83 is gone. \n", "Eevee randomly causes system to crash when rendering\nOperating system: Windows-10-10.0.19042-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2070 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 511.23 (8GB memory)\nCPU: Intel Core i9 9900k\nRAM: 32GB DDR4\n\nBroken: 3.0.0\n\n\nI'm new to Blender and currently trying to switch from my usual C4D and Redshift workflow, just to let you know.\n\nI have a fairly simple scene with a 250 frame animation. Everytime I render the animation with Eevee as a .png sequence with an .exr file for the layers, the process causes my system to crash.\nThe amount of samples seems to correlate with the amount of frames I'm able to render before failure. So I guess it might have something to do with memory.\n\nI'm not sure how that can be as more complex scenes never failed me before using Redshift as a render engine.\n[ocean_final.blend](ocean_final.blend)\nThe scene makes use of a lot of volumetric light and I also calculate the bloom.\n\nThere aren't a lot of vertices involved in the scene, but I'm not sure how I can display a value to give you here.\n\nAll drivers are up to date and I also reinstalled Blender.\n\nNot sure, but I attached the .blend file as well as logs created in debug mode.\n\n[blender_debug_output.txt](blender_debug_output.txt)\n\n[blender_system_info.txt](blender_system_info.txt)", "The Viewport Display setting of Particles affect the Render result with Eevee \nOperating system: Linux-4.15.0-55-generic-x86_64-with-debian-buster-sid 64 Bits\nGraphics card: GeForce GTX 1080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 390.116\n\n\nBroken: version: 2.80 (sub 75)\nWorked: (optional)\n\n\nThe Viewport Display setting of Particles affect the Render result with Eevee \n\nWith Particles selected Change \"Display as\" to \"Path\" in Viewport Display setting and Render with Eevee . error : Pathes will be dislpayed in render result \n[partical_render.blend](partical_render.blend) press render\n\nFrom other reports what is also affected:\n- `Viewport Display` > `Strand Steps`\n- `Viewport Display` > `Amount`" ]

[ "Shading Viewport, deadly slow or crashing in 3.2 or above\nOperating system: macOS 12.4\nGraphics card: Radeon Pro Vega 56 8Gb\n\nIntel Imac Pro 2017\n\nBroken: (3.2.0 also 3.2.1 and 3.3)\nWorked: 3.1.0 (excellent)\n\nView port works good, render (F12) works good. The problem begins with shading viewport in **material preview** takes ages to show or in some more complex scenes crashes after a while. All works and opens fine in 3.1. Tested with clean install with factory settings (no add ons, etc) and also importing 3.1 settings. All textures are procedural, tested without world texturing in all cases. \n\nBased on the default startup or an attached .blend file (as simple as possible).\nThe attached file it does work just takes ages to show compared with 3.1 that works really good, the ones that crashes are large scenes and after several tries (appending only different collections) I wasn't able to manage what makes it crash.{[F13228801](test.blend)}[Crash 3-3 on base Intel IMac 2017.txt](Crash_3-3_on_base_Intel_IMac_2017.txt)" ]

[ "Crash Python API - GPUOffScreen.draw_view3d() crashes blender with draw_type = POST_VIEW but works with draw_type = POST_PIXEL\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: GeForce RTX 3070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 457.51\n\nBroken: version: 3.3.1\n\nWhen using [GPUOffScreen.draw_view3d() ]] in a [[ https:*docs.blender.org/api/current/bpy.types.Space.html?highlight=post_view#bpy.types.Space.draw_handler_add | draw_handler ]], blender crashes when running the [[ gpu.shader.html?highlight=gpu+shader+from_builtin#built-in-shaders | 3D_IMAGE builtin shader ](https:*docs.blender.org/api/current/gpu.types.html?highlight=draw_view3d#gpu.types.GPUOffScreen.draw_view3d) if the draw_type was set to POST_VIEW but works if the draw_type was set to POST_PIXEL.\n\n\n\n - Open CrashScriptDrawView3D.blend\n - Run the script Crash3D and blender will crash\n\n\n\n\n\n - Open CrashScriptDrawView3D.blend\n - Run the script NoCrash2D, and it will draw the camera view in the viewport without crashing\n\nAttachments:\n\n[CrashScriptDrawView3D.blend](CrashScriptDrawView3D.blend)\n[blender.crash.txt](blender.crash.txt)\n\n\n\n\n\n\n", "Avoid graphical glitches during threaded preview generation\nWhen previews are generated using an OpenGL engine, there can be a number of glitches until the preview is done. That's because some data is accessed/modified that is not thread safe. E.g. `object_preview_render()` modifies `U.pixelsize`. I think other glitches are caused by overriding the active theme in `ED_view3d_draw_offscreen()` - although when I avoided this some time ago for similar work on threaded offscreen drawing for VR, this didn't solve the color flickering.\nThe active theme should probably use thread local storage.\n\nBesides that, the preview generation should be thread safe with OpenGL:\n* It uses the draw-manager's own context, which is mutex protected.\n* The non-thread-safe immediate mode API is not used inside the draw-manager.\n* The draw-manager already has workarounds for GPU driver issues with threaded drawing.\n\nA note: We could also speed up Eevee preview generation significantly by using viewport render quality, not final render quality. The `_imbuf` offscreen rendering functions don't expose the `is_image_render` option for that, which they probably should.", "Blender 3.1 CPU/GPU rendering denoise is broken on M1 Mac / Monterey\nOperating system: MacOS Monterey (M1 Max)\nGraphics card: Integrated GPU\n\n**Blender Version** 3.1.0 release for M1 Macs\nBroken: (example: 2.80, edbf15d3c044, master, 2018-11-28, as found on the splash screen)\nWorked: (newest version of Blender that worked as expected) 2.93.3\n\nOn all my projects, when rendering via the CPU or GPU, the render works as expected, but on completion the entire screen blacks out to a solid color and the render is LOST! I suspected it must be a post render step, and I have confirmed that it is the DENOISER that is broken in this version of blender. When I go into the scene and remove the denoiser, the render completes, and I can save the image!\n\nBased on the default startup or an attached .blend file (as simple as possible).\nYou need a project that uses denoising as a final compositing step. Attached is a simple matt object set to denoise at the end. Render it, and at the end you get a gray screen. Go into compositing and remove the denoise node, and you get the render.[TestRenderWithDenoise.blend](TestRenderWithDenoise.blend)\n", "Regression: Image rendering fails while viewport rendering is in progress\nOperating system: OSX Ventura Version 13.3\nGraphics card:AMD Radeon Pro Vega II Duo 32 GB\n\nBroken: 3.5.0 Date 2023-03-29 02:56 Hash: 1be25cfff18b \nWorked: 3.4.1 Date 2022-12-19 17:00 Hash 55485cb379f7\nCommandBuffer Failed: cycles_metal_integrator_compact_shadow_states\n\n* Load file provided\n* Set ViewPortShading to MaterialPreview\n* Then select ViewPortShading to Rendered\n* While the viewport is rendering Select Render Image.\n\nThis will often cause the error. It may take an attempt or two but more likely to happen after at least one attempt.\n\nIn some case blender will throw the error in both the viewport and in the image window. In some cases...Blender will hang and require you to force quit. At other times it will crash the OS. \n\nThe larger the scene the more drastic the failure. In this simplified file it typically will recover.\n\n", "Cycles crashes on viewport rendering in CUDA driver\nOperating system: Windows-10-10.0.19043-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.94\n\nBroken: version: 3.3.0\nWorked: 3.2\n\nThe crash happens while navigating the viewport in shading mode, using Cycles.\n\nThis crash reappeared in the latest 3.3 release. It happened in older releases, then it got stable, but now it's back again.\nApparently, it has to do with Cuda.\nIt's not happening in one particular scene only, but rather randomly regardless of the complexity of the scene.\n\n[Cave-CloseUp-v006.crash.txt](Cave-CloseUp-v006.crash.txt)\n\n[Cave-CloseUp-v005.crash.txt](Cave-CloseUp-v005.crash.txt)\n\n[blender.crash.txt](blender.crash.txt)", "Viewport Render (and Render Animation) color looks completely off from the Viewport\nOperating system: Darwin-19.0.0-x86_64-i386-64bit 64 Bits\nGraphics card: AMD Radeon Pro 560X OpenGL Engine ATI Technologies Inc. 4.1 ATI-3.2.24\n\nBroken: version: 2.82 (sub 6)\n\nRendering the viewport animation gives a very different result from the actual viewport.\n(Also: Viewport Anti-Aliasing isn't taken into consideration on the Viewport Render).\n\n||\n| -- | -- |\n\nThe problem is more noticeable when the Viewport Shading is set to \"Flat\", even though it also shows in \"Studio\". It delivers very dull colors with less contrast and a blue-ish tone.\nChanging between Cycles and Eevee doesn't affect the Viewport Render (from my tests).\n\n||\n| -- | -- |\nflat:\n||\n| -- | -- |\n\n- Open attached file.\n- Go to View -> Viewport Render Image\n[Viewport Colors.blend](Viewport_Colors.blend)", "Realtime Compositor - Crash on Mac when using Map UV node\nOperating system: macOS Ventura 13.3.1\nGraphics card: AMD Radeon Pro 5700 XT 16GB\n\nBroken: 3.6 alpha cef128e68af5 (2023-04-13 16:06)\n\nConnecting a Map UV node in Compositing will crash on macOS\n\nIn Blender's startup scene, enable Compositing by checking \"Use Nodes\".\nConnect a Map UV node between the Render Layers node and the Composite node.\nIn a 3D view, turn on Rendered viewport and set Viewport Compositing to Always.\nThis will crash (100% reproducible).\n\nPlease contact summary if more information is required.\n\n", "Blender Crash when deleting images too fast\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: AMD Radeon(TM) Graphics ATI Technologies Inc. 4.5.14761 Core Profile Context 21.30.44 30.0.13044.0\n\nBroken: version: 3.2.0\n\nBlender crashes when deleting images in the Layer blend file libery\n(3.3 alpha is crashing also and i didnt found any report about this issue)\n\n\n* go to outliner -> select blend file\n* make sure to be in the material preview mode\n* delete the first image by right click and then click on delete image\n* (now u only need to rightclick and leftclick to delete next images, because u dont need to aim for the delete anymore)\n* delete images while blender tries to preview the last change after deleting one of the images\n blender crashes\nwhen deleting images too fast, my blender just crashes (happens on my bad pc often, because i download models and rework them)\nIf you wait for the new material preview of the model after deleting nothing happens, but if you dont blender crashes when continuing to delete images\n\nVideo: CvO9nw4qilc\n\n[pharaoh x suite.blend](pharaoh_x_suite.blend)\n", "Workbench refactor causes significant performance regression with workbench shadows\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1060 6GB/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 445.75\n\nBroken: version: 2.83.0\nCaused by c476c36e40\n\nWorked: 2.82a\n\n2.83 solid viewport shading work slowly than 2.82 when shadows is on\n\nReduction of performance in viewport very high in heavy scenes as example scene with sculpts 6 millions of triangles\n\n", "Blender freezes with Metal render device\nOperating system: macOS-13.2.1-arm64-arm-64bit 64 Bits\nGraphics card: Metal API Apple M1 Pro 1.2\n\nBroken: version: 3.5.0\nWorked: None\nAlso Tested: 3.5.1 RC (April 01, 01:59:41, `6fa464d18971`)\n\nBlender freezes after a few seconds when I enable the Metal Render Device in Preferences, switch to Cycles with GPU Compute and switch Viewport Shading to Rendered.\n\n1. **Do not** delete the default cube\n2. Switch to Metal (GPU) Cycles Render Device in Preferences > System\n3. Set GPU Backend to Metal\n4. Switch Render Engine to Cycles and device to GPU Compute\n5. Switch Viewport Shading to Rendered\n6. Wait for Cycles to compile shaders\n7. Look around for a few seconds\n8. Blender Freezes\n\n", "texture in solid mode not updated\nOperating system: Ubuntu 20.04\nGraphics card: Geforce MX230\n\nBroken: 3.2, 2.93\nWorked: probably never\n\nif the shader editor is located in another workspace, in which there is no viewport in solid mode, then the texture on the object will not be updated\n\n- Open .blend file\n- In both layout an shading workspace: make sure skin texture is selected\n- switch to shading workspace and select blue texture\n- Now switch back to layout workspace (notice `blue` texture is selected in node space but viewport is not updated)\n\n[#99598.blend](T99598.blend)\n\n---\noriginal file and video provided by the reporter\n[jupiter.blend](jupiter.blend)\nVideo [F13283900](2022-07-11_00-19-08.mp4)", "Blender 3.6.3 SegFault Crash at startup on MacOS Ventura 13.5.2\nOperating system: MacOS Ventura 13.5.2 (22G91)\nGraphics card: Intel Iris Plus Graphics 650 1536 Mo\n\nBroken: 3.6.3 (3.6.3 2023-09-21)\nWorked: None tried on this machine\n\nCrash on startup with `Segmentation fault: 11`\n\nEither:\n- Double-click on Blender application icon\nOr:\n- command line : `/Applications/Blender.app/Contents/MacOS/Blender` with or without option `--gpu-backend`\n\n```\nThread 0 Crashed:: Dispatch queue: com.apple.main-thread\n0 Blender \t 0x109d2903d (anonymous namespace)::Local::Local() + 29\n1 Blender \t 0x109d28fe9 memory_usage_init + 25\n2 Blender \t 0x109d26f09 MEM_init_memleak_detection + 9\n3 Blender \t 0x101e098c3 main + 227\n4 dyld \t 0x7ff815a5241f start + 1903\n```", "Rendering large scene causes viewport shading rendered to go out of memory\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2060 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 531.41\n\nBroken: version: 3.6.0 Alpha\n\nwhen dealing with large files and scenes in Cuda or optix, rendering the scene then stopping and closing the render, causes the viewport shading rendered view to give out of memory when switch back to it, it renders normally before hitting F12, after hitting render and everything compile and render, when closing the rendered and switching back to viewport rendered view it gives out of memory and its stuck with it and I need to restart blender to make the viewport render work again.\n\nseems it keeps the memory full even after I close the render so there is no memory for viewport I have no idea (I'm not a programmer :D) \n\nWorks normally in 3.4.1\n\nThank you\n\n- Open large files (like view?usp=sharing)\n- set Cycles Render to Cuda or Optix\n- Press F12 to Render and ESC to cancel\n- Close render window\n- Set viewport shading to rendered\n\nSee \"out of memory\" issue\n\n", "Cycles rendered 3DViewport massive performance drop, using lights with node groups. --Intel MacOS--\nOperating system: macOS-13.4-x86_64-i386-64bit 64 Bits\nGraphics card: Metal API AMD Radeon Pro 5700 XT 1.2\n\nBroken: version: 4.0.0 Alpha\n\nSee screenrecording video! \nWhen switching to a light with nodegroups in Cycles rendered 3DView, performance drops a big time. (~50%).\n\n1. Open the performance.blend file with Intel MacOS blender 4.0, switch to cycles rendered 3DView, select the cube and rotate the view.\n2. Then switch to the caustic light and try to rotate the view again.", "dual monitor will sometimes cause blender to freeze\nOperating system:win10\nGraphics card:2080s\n\nBroken: 3.4.1\nWorked: \n\ni have two monitors (4k and 2k) scrolling timeline or doing anything on one monitor and displaying cycles render viewport on the other monitor will sometimes cause blender to freeze forever\n\nit is not specific to this scene it will freeze on all kinds of scenes, might be related to the muti-monitor bug i have been reporting T96248\nhere is a screenshot of what it looks like when freezing \n\nmake some animation\nhave 2 monitors and use cycles rendered view on one monitor and scrolling timeline on the other monitor" ]

[ "3.2 eevee bad performance\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: Radeon RX Vega ATI Technologies Inc. 4.5.14800 Core Profile Context 22.4.2 30.0.15021.7000\n\nBroken: version: 3.2.0\nWorked: 3.1.2\n\nEevee is slower when compiling shaders and on viewport.\nCompiling time has doubled and viewport feels laggy when zooming on small details.\n\n - Open attached file\n - Change viewport to material preview\n[slow.blend](slow.blend)\n\n\nMy compiling times went from ~10 seconds (3.1.2) to ~20 seconds (3.2) when limiting blender to one thread.\nWhen disconnecting the bump node both compile in ~3 seconds.\nI think it might be a combination of many nodes and bump node.\n" ]

[ "Apply Base (Multiresolution Modifier) on mesh with shape keys distorts the mesh\nBroken: version: 2.83 (sub 8)\nWorked: Doesn't Work even in Blender 2.79b official release. I haven't tested versions earlier than that.\n\nPressing Apply base using the multires modifier on a mesh with a stored Basis shape key causes artifacting and mesh corruption. Undo still works though.\n\n1. Delete Default Cube\n2. Add Plane mesh (Shift+A > Mesh > Plane)\n3. Add Basis Shape Key.\n4. Add Multiresolution modifier\n5. Subdivide multiple times. (Either Catmull-Clark or Simple type)\n6. (optional) Apply brush strokes in Sculpting mode. \n7. Press Apply Base in Modifier Properties.\n\nor\n\n1. Open attached file\n2. Press Apply Base in the Modifier Properties.\n\n[Shape key Multires Bug.blend](Shape_key_Multires_Bug.blend)\n\n", "Extra face dots in UV Layout editor (GPU subdivision OFF, On Cage OFF in Subdivision modifier make a difference)\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: Intel(R) HD Graphics 4600 Intel 4.3.0 - Build 20.19.15.4835\n\nBroken: version: 3.4.1\nBroken: version: 3.1\nWorked: not since GPU subdivision was introduced (so in a way, it was working in 3.0)\n\nExtra random faces appear in the UV Layout editor\n\n- I can mosty repro this if GPU Subdivision in Preferences > Viewport > Subdivision is OFF\n- It does not happen at all if I turn off `On Cage - Adjust edit cage to modifier result` on the SubSurf modifier.\n\nIn the attached file, toggle from face select mode to vertex select mode and back to face select mode. (Might have to repeat 5-6 times) Extra faces appear in the UV Layout editor. Also see here: [1457172]\n\n", "Sporadic boolean issues\nOperating system: Windows-10-10.0.19043-SP0 64 Bits\nGraphics card: GeForce GTX 1080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 431.07\n\nBroken: version: 3.2.0\nWorked: 3.0\n\nMeshes using the boolean modifier sporadically break in Blender 3.2.0, while they still worked in 3.0.\nThis would result in meshes that lack faces or even crash Blender\n\n\nUnfortunately I wasn't able to isolate the cause of the issue, as it seems to occur sporadically and randomly. What I have noticed is that sometimes meshes that were fine suddenly break once you enter exit edit mode.\n\n", "Transform: Snap to the visualized cage of the edited mesh\n## Motivation\nCurrently in editing mode the snap is always made on the values of the bmesh.\nBut intuitively the user tries to snap on the displayed cage which can be affected by the `On Cage` option.\n\n## Proposal\nUse `mesh_eval_final` or similar when the cage does not match bmesh.\nVertices and edges originating from modified faces can be ignored in snapping.", "Viewport Z-Bias & retopology overlay related issues\nSince this is being reported frequently and retopology overly has caused some weird issue with mask display as well previously, it's best we keep track of them in a central place so we could check if any particular fix is working for all situations:\n\n------\n\nProblem still presents:\n\n- Retopology overlay affecting other parts of the drawing/picking operation: #109115 #107768 ~~#109277~~ ~~#107411~~\n- (Potentially related) Selecting when X-Ray turned on is not visually effective and then switching to face mode crashes: #110136\n\n------\n\nFixed:\n\n- ~~Wireframe visible in close distances: #110048 #109040 #109514 #109499 #109685~~\n- ~~Retopology overlay 0 z offset problem: #109662~~\n\n-------\n\nProbable cause:\n\nAs suggested in 107768#issuecomment-937545 , I believe the problem here is that\n\n1. `DRW_object_is_in_edit_mode` does not exclusively check if object is in edit mode, but it combines other logics that affects drawing and stuff.\n2. `XRAY_ENABLED` marco is not always accompanied with `RETOPOLOGY_ENABLED`, this leads to inconsistencies in the user interaction and the display.", "Bevel Modifier: Odd amount of segments causes UV propagation issues at seams\nOperating system: Linux-5.17.15-76051715-generic-x86_64-with-glibc2.35 64 Bits\nGraphics card: Quadro RTX 6000/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 515.48.07\n\nBroken: version: 2.83, 3.4.0 Alpha\nWorked: unknown\n\nAny odd number of bevel segments greater than 1 fails to preserve seams in the propagated UV data on intersections. Instead UV data is interpolated, which can for example cause nasty artefacts when baking.\n[bug_bevel_uv_seams_interpolation-2022-09-01_15.26.11.mp4](bug_bevel_uv_seams_interpolation-2022-09-01_15.26.11.mp4)\n\nFile:\n[baking_artefacts.blend](baking_artefacts.blend)\n\n", "can't use boolean modifier to mask out Geometry Node text, bleed text & flickering in some frames\nOperating system: Windows-10-10.0.19042-SP0 64 Bits\nGraphics card: Quadro P4000/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 397.93\n\nBroken: version: 3.1.0 Alpha\n\ncan't use boolean modifier to mask out Geometry Node text, bleed text & flickering in some frames.\n\nin frame 139, 148, 149, 306, 353 text bleed out of mask_text box Boolean object\n\n[2021_Ecard_1920x976_23d.blend](2021_Ecard_1920x976_23d.blend)\n\n[2021_Ecard_1920x976_23d0001-0420.mp4](2021_Ecard_1920x976_23d0001-0420.mp4)\n\nThanks for your help!", "straighten edges and unfold uv island\nStraighten selection and unwrap unselected vertices\n\n**Starting point**\n\nThe uv has pretty wonky topology. \n\n\n\n\n**Desired result**\n\nThe selection is straightened and the non selected uv’s of the UV island is unwrapped. Should work on multiple uv islands\n\nSuggested approach:\n- Store\n - Original pinning\n - Original seams\n- Run “seams from islands”\n- Remove all pins\n- Pin selection\n- Align auto on selection\n- Unwrap\n- Restore \n - Original pinning\n - Original seams\n\n\n", "Cycles: improve subsurface scattering for some meshes shapes\nIn cycles-x, the Christensen-Burley dipole approximation was fully replaced by random walk.\n\nHowever for some mesh shapes results are not as good, or could be improved regardless.\n* Self-intersection meshes give dark borders. Counting back/front facing intersections could help determine inside/outside better to avoid such artifacts.\n* Sharp corners \"glow\", which may be physically correct but is not necessarily desired. It may be possible to add user control with an option somehow, using some minimum throughput\n\nIf these can't be solved, we may consider adding back the dipole approximation, but since path splitting and the memory usage is a difficult fit for the new kernels we'd really like to avoid it.", "Add \"Topology Mirror\" as option for relevant operators\nThis feature request was discussed with Campbell Barton during meeting 2021-02-23\n\nThe following mesh operators behaves differently when \"Topology mirror\" == True in the edit mesh options.\n\n- Select mirror\n- Symmetrize\n\n---\n\nThe topology mirror option is located here \n\n\n---\n\n\nIn order to make this option more clear for the user, it should be added as an option on each operator\n\nOperator mockup\n\n\n", "Bevel Modifier and Normal Bevel Skews\nOperating system: Windows 10\nGraphics card: GeForce GTX 3080\n\nBroken: All Blender versions including the latest 2.92\nWorked: Probably on previous versions\n\nHello guys,\n\nThe Bevel modifier and the normal Bevel skews the higher the offset, and does not keep a similar thickness from the outside and inside wall.\nThe current \"solution\" is to bevel a plane and then add a Solidify modifier however the shape alters, which will require applying the modifier and further altering the shape, not allowing for modifier stack workflow. I asked a friend to try bevel in the same way in 3DS Max using my exported FBX and it beveled without skew.\n\nLoad up the scene and you will see the applied modifiers, some of the different offset types have less skew effect but ultimately they all skew quite extensively\n\n[Bug.blend](Bug.blend)\n\n\n\n\n\n\n\n\n\nAs you can see the right shape is only with a Bevel modifier, it keeps the top flush to the shape however gets severely skewed and does not keep a similar thickness on the more rounded off corners. The version with a Solidify is more accurate however does not keep the top part flush with the rest of the shape and requires further modifying and loosing the stack workflow.\n\n", "Scrape brush with constant falloff creates a hole when scraping edges\nOperating system: Windows-10-10.0.22616-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3050 Ti Laptop GPU/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 512.78\n\nBroken: version: 3.2.1\nWorked: somewhere in 2.7\n\n\nScrape brush with constant falloff creates a hole when trying to flatten an edge.\n\n\nIn the attached file scrape a lot of times along the drawn line.\n\n[scrape.blend](scrape.blend)\n\n", "snap to face broken with mirror modifier\nOperating system: Linux-4.18.0-25-generic-x86_64-with-debian-buster-sid 64 Bits\n(same behavior on Windows 8.1)\n\nGraphics card: GeForce GTX 1060 6GB/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 390.116\n\nBroken: version: 2.81 (sub 11)\nAlso broken in 2.80 stable\n\n* open demonstration blend file with no config dir\n* note that snap-to-face (target:closest, affect:move) is already enabled\n* select small cuboidal in lower left\n* G, then X, to move along X, and attempt to snap to broad side of nearby horizontal cuboidal\n* see that the object flashes erratically, alternately assuming its original position, and alternately assuming an incorrectly snapped position\n* if instead the cuboidal we are trying to move is moved a bit first, without snapping, and then we again try to snap to the face, it doesn't flicker any more, but it still snaps incorrectly (and actually still does flicker on the overlapping face but doesn't jump around in position)\n* if the mirror modifier on the cuboidal we are trying to move is disabled, it snaps correctly\n\nStarting position: \n\nIncorrect snap result: \n\nCorrectly snapped result: \n\nDemo blend file: [snapbug.blend](snapbug.blend)", "Vertices will always merge when \"Clipping\" is checked on a mirror modifier\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3090/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 527.56\n\nBroken: version: 3.4.0\nWorked: Not sure\n\nDespite \"merge\" being unchecked, if you have clipping checked on a mirror modifier it will behave as if merge is turned on.\n\n1. Scale down the default cube to 0.1. Apply scale.\n2. Add a loop cut down the middle and delete half the mesh\n3. Add a mirror modifier and turn off merge\n4. Select the edge loop you just created and move it very slightly off center (so there's a gap)\n5. Turn on clipping (but leave merge off)\n6. In edit move move the edge loop again and you'll see it auto-merge the vertices close to the edge. If you reduce the merge distance it will not merge them, showing that when clipping is enabled it will auto merge verts despite that being unchecked.\n\nAttached example:\n[clipping demo.blend](clipping_demo.blend)\n\n", "inset and bevel edits crashing blender 3.x\nThreadripper 1950x 128 gb ram 2tb SSD (home)\nNVIDA 3080 updated to latest drivers\n\nIntel i7 64 gb ram 2 tb SSD (work)\nAMD 6900xt updated to latest drivers\n\n\nHappens on all 3.x versions of blender.\n\nIssue happens on both my home PC (the thread ripper) and my work PC.\n\nStart editing default cube add bevels, insets place some cuts... crashes within a couple of minutes. Happens in all display modes.. mesh, solid, renders. Happens with a smooth display mesh and hard edges. Super frustrating need to save just before doing any of these procedures on any modeling task. Always update video drivers which probably cant be the issue because two different families of video cards in two different machines have the same issue. Been doing this ever since I updated to Blender 3. I assume it's a plugin I installed because I can't imagine why this would happen on two completely different computers. Is there a dump file or something I can send? Here is a video of a typical crash. No dialog box... just blender vanishes in the the middle of editing.\n\n" ]

[ "Mirror modifier bug - Bisect mode results in “automerge” even if it shouldn’t. Most likely due to units setup put to cm. \nOperating system: Windows 10\nGraphics card: GeForce GTX 970\n\nBroken: All Blender versions including the latest 2.91\nWorked: Never\n\nHello guys, \n\nThe mirror modifier doesn't do it's job properly when having \"bisect\" enabled. There is a hidden threshold to which it will snap vertices to the mirror line. \nI ruled out all possible things, the mirror pivot is perfectly zeroed out and there is no merge enabled and also no auto merge. Stan from Blenderartists pointed out that switching to meters and scaling the object up 100 times “solves” the problem, so it's most likely a bug related to having switched the units to cm. \n\nIt's also not specific to the latest Blender version, it does the same error in older versions. I attached the scene so you can take a look at it. [bug.blend](bug.blend)\n\n\n\n\nLoad up the scene and simply put it to \"bisect\" and you will see how it merges the verts when it shouldn't. " ]

[ "Wrong particle tool properties displayed when using Select or Cursor tools\nOperating system: Darwin-18.2.0-x86_64-i386-64bit 64 Bits\nGraphics card: Intel(R) Iris(TM) Plus Graphics 640 Intel Inc. 4.1 INTEL-12.4.7\n\nBroken: version: 2.80 (sub 50)\n\nIn Particle Edit mode, when using the Cursor or Select tools, you can still see particle tool properties for the brushes.\n\n\nOpen this blend file:\n[particle_toolprops_bug.blend](particle_toolprops_bug.blend)\n\nIf you select the Comb tool, you just see the Comb tool properties;\n\n\nBut, if you select the Box Select or Cursor tool, the comb tool properties aren't cleared, and you see settings for both at the same time.\n", "Checker texture artifacts when feeded with object texture coordinates\nOperating system: Linux-5.4.0-90-generic-x86_64-with-glibc2.27 64 Bits\nGraphics card: GeForce GTX 1660 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 460.91.03\n\nBroken: version: 3.1.0 Alpha\nWorked: afaik never, still had this issues with all previous versions down to 2.7\n\nChecker texture has calcuation issues when feeded with object texture coordinates.\nOther textures like brick or voronoy are ok.\n\ncreate a plane\nassign the checker texture with object coordinates\n\n\n[checker_texture_problem.blend](checker_texture_problem.blend)\n\n", "Surface smoothing when sculpting does not respect mirror modifier\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1060 6GB/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.94\n\nBroken: version: 3.2.0\n\nSmoothing in sculpt mode using Surface deformation does not respect the clipping option of the mirror modifier. This is not the case in the other brushes.\n[0281-0782.mp4](0281-0782.mp4)\n\n1.Creatre an object with at least one plane of symmetry.\n2.Delete one side and add the mirror modifier to fill that same side.\n3.In sculpt mode, try smoothing over the mirror seam using Surface deformation.\n\n", "Sculpt mode: annotation and unhide conflicts with Multiresolution modifier\nOperating system: Windows-8.1-6.3.9600-SP0 64 Bits\nGraphics card: GeForce GTX 660 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 461.40\n\n**Broken: version:**\n2.93.0 Alpha.\n3.0.0 Alpha\n\nUse of annotation tool unhides the hidden face\n\n**Steps to Reproduce the error**\n- Open .blend file\n- Hide the faces with Box Hide tool\n- Erase the word \"Erase\" with annotation tool (hidden part will appear)\n- Repeat step 2.\n- Draw a line with annotation tool. (Hidden part will reappear)\n\nTest File:\n[#87821.blend](T87821.blend)\n[Blender [C__Users_Pratik_Downloads_T87821.blend] 2021-07-12 09-55-06.mp4](Blender__C__Users_Pratik_Downloads_T87821.blend__2021-07-12_09-55-06.mp4)\n", "Clay strips brush freeze when coming from outside the horizon of surface curvature\nOperating system: Linux-5.4.0-121-generic-x86_64-with-glibc2.31 64 Bits\nGraphics card: GeForce GTX 670MX/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 390.151\n\nBroken: version: 3.2.1\n\nClay strips brush freeze when reaching horizon of surface curvature\n\nOpen blender, then open a new sculpt file, select the clay strips brush, change the direction of the effect to subtract and change the space distance to scene and disable the adjust strength for spacing.\nWhen coming from outside the object blender freezes or lags creating sometimes artefacts in the horizon of the object.\n\nIn this video you can see the normal behaviour with view spacing distance and with scene spacing distance wile adjust strength for spacing is enable, and then the buggy behaviour.\n\n[vokoscreen-2022-07-12_15-29-16.mp4](vokoscreen-2022-07-12_15-29-16.mp4)\n", "Boolean modifier produces unexpected result\nOperating system: Debian bookworm / sid\nGraphics card: Intel WhiskeyLake-U GT2, but most likely doesn't matter\n\nBroken: 3.3.1 (from Debian), 3.4.1RC (Daily Build from blender.org, b87bcd3f8fe1), 2.93.12 RC (Daily Build, 1f30e…)\nWorked: unknown\n\n\nIteratively generating the difference between a mesh (in this case a plain torus) and a rotated version of its copy seems to work fine initially but results in broken meshes after a few steps. Artefacts include non-manifold bits and pieces of the original mesh, or a DIFFERENCE step that ends up adding the mesh that it just subtracted to the result.\n\nAlso discussed at 1427102 (page contains images showing said broken results).\n\n[problem.blend](problem.blend)\n\nInspect file above.", "Drag tool-tip stays when mouse leaves the window\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3070/PCIe/SSE2 NVIDIA Corporation 4.6.0 NVIDIA 536.99\n\nBroken: version: 4.0.0 Alpha\n\nTool-tips displayed on drag stays after mouse leaves window\n\nOpen any file importer and drag any file, after exit the main window tooltip will still show\n\n\n\n<video src=\"attachment\" title=\"2023-08-18 21-26-50.mp4\" controls></video>\n\n", "Alpha hashed material produces shifting artifacts when animated in EEVEE\nWhen an object with material that has alpha hashed transparency is animated in EEVEE, the noise pattern may look either unstable or shifting as shown below: \n[alpha-hashed-transparency-problem.mp4](alpha-hashed-transparency-problem.mp4)\n\n\n\nYou can observe that the artifacts on top staying in place rather than rotating with the cube while those on the sides keep flickering, thus producing shifting and unstable appearance.\n\nThe symptom is also noticeable when changing the camera position in rendered viewport mode in Blender. And it's particularly problematic in UPBGE (a fork of the old BGE based on the latest Blender source) because a game engine must animate objects in realtime with relatively small number of samples per frame.\n\nUpon an examination from the developers of UPBGE, it is found that the problem is due to the fact that the current implementation of alpha hashed transparency uses the world coordinate instead of the object space, as it is often recommended (e.g. [2017 ACM SIGGRAPH Symposium on Interactive 3D Graphics and Games](p4TYf5DDpbQ)).\n\nIt appears that a simple change from the world to object coordinate fixes the problem as shown in this [test commit (see line 80) ](e0a8638d0b255f2fc6ce778d10dd0d12a467b4d3#diff-2b8bf4f9e10009510d02c14e3659ee032c73f829de5453d6abcdf6a3f884fdfdR80), which produces the following result:\n[alpha-hashed-problem-solved.mp4](alpha-hashed-problem-solved.mp4)\n\nIt's my understanding that the [original author ]() decided to use the world matrix to avoid the difficulties with certain corner cases, like when the object is scaled by an extreme factor. However, while such a problem may be circumvented relatively easily (e.g. discard the scale information from the matrix) such a problem described above may pose more substantial issues with making an animation or rendering in realtime.\n\nSo, I hope this problem and the suggested fix may be evaluated again since we have more people interested in using Blender for such purposes now, thanks to the gaining popularity of EEVEE.", "Weight paint issue with brush Accumulate option\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 950M/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 461.40\n\nBroken: version: 3.0.0 Alpha\n\nWhen use weight paint with option Accumulate, turn on symmetry, artifacts appear on the axis of symmetry\n[test.blend](test.blend)\n\n\n[EnrLS0nY4n.mp4](EnrLS0nY4n.mp4)\n", "GP: Time offset and Armature modifier causes glitches\nOperating system:WINDOWS10\nGraphics card:GTX1060\n\nBroken: 2.93LTS\n\n\n[#89433.blend](T89433.blend)\n\n- Open file\n- Move selected bone to left until it hits constraint, and then a bit randomly\n\nGP object will keep rotating in one direction and can be reset only by changing frame.\nSame happens when scrubbing in negative frame range.\n", "Displacement map, wrong normals and shadows on triangle geometry\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: Quadro RTX 3000/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 452.69\n\nBroken: version: 2.92.0\nWorked: ?\n\nExperimental Cycles displacement introduces rendering artifacts on some geometries made of triangles. The problem seem to disappear on quad geometry.\n\nOpen the attached \"displace-problem.blend\" file.\nRender an image, or enable Cycles preview in viewport.\nThe left object is made of triangles, and appears with problems.\nThe right object is made of 1 quad, and appears fine.\n\n[displace-problem.blend](displace-problem.blend)\n", "Multiresolution sculpting broken after change in base geometry\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: AMD Radeon RX 5700 XT ATI Technologies Inc. 4.5.14802 Core Profile Context 22.1.2 30.0.14023.3004\n\nBroken: version: 3.1.0 Beta\n\nThe file was created at 3.1.0, but the model in 3.0.1 also has the same problem.\n\nafter deleting edges from a loop and recreating them in a different arrangement, some parts of the model cannot be sculpted with multiresolution on.\n\n\nIn a model with multiresolution, I deleted and recreated some edges of the topology and this broke the sculpt mode of the multiresolution: \n- In the sculpt mode, whenever I was getting close from the model, it simply dissappeared, reapearing after zooming out.\n- When hovering the mouse over some areas of the model, doesn't change the brush gizmo, and I can't sculpt on those areas neither.\n- If there is a area that you can interact/sculpting, by doing it the areas that you were unable to sculpt will become sculptable again.\n- In the areas that you can sculpt, the brush gizmo rotation doesn't match the surface sometimes.\n- baking normals from this multiresolution causes blender to crash\n\nI splited the model in some parts trying to remove the defective part, but this only changed the areas the bug was occuring, and I also tried to re-sort the vertices of the model, which had the same effect\n\nIn this video, the upper part of the leg is sculptable, but the rest of the leg is not \n[2022-02-06 14-33-33.mp4](2022-02-06_14-33-33.mp4)\n\nThis model is from a commission so I had to fair the upper part of it\n[broken multires.blend](broken_multires.blend)\n\n", "Edit mode. Wire become non-selectable dependent on resolution (regression)\nOperating system: Windows-7-6.1.7601-SP1 64 Bits\nGraphics card: GeForce GTX 660 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 430.86\n\n\nBroken: version: 2.80 (sub 75)\nWorked: 2.79\n\n\nWithout multisampling wire just disappear and with multisampling it is still visible but not selectable.\n\n[2019-09-14_23-07-53.mp4](2019-09-14_23-07-53.mp4)\n[2019-09-14_23-09-58.mp4](2019-09-14_23-09-58.mp4)\n\n\n\nTry to select loop and zoom in-out with and without viewport multisampling. Here the file, but bug will appear even with UV-sphere.\n[b.blend](b.blend)\n\n", "Viewport display of multiple Dupli-Collections when different draw-types are used : wireframe might disappear (only selected are garuanteed to draw)\nOperating system: Windows 10 / Ubuntu 20.04\nGraphics card: GTX 1080 / RTX 3080\n\nBroken: 2d9d08677e a2d59b2dac \nBroken: all versions >= 2.8\nWorked: 2.79\n\nViewport display of Dupli-Collections is inconsistant.\nDigging deeper into it, it looks to be when there are different draw types on the same dupli-collection\n\n1. Open attached blend file\n2. select the empties in the middle of the screen which do not look like they have anything attached\n3. notice how they randomly appear / disappear upon selection.\n4. note that one of instancer is set to `Textured` draw type, the other to `Wire`, if all are set to `Wire` the issue is gone\n\nThis has happened in quite a few of our production scenes.[bug.blend](bug.blend)\n\nNOTE: Might boil down to be the same as #101822 (Collection instance set to Display As Wire doesn't show in Solid mode when not selected) or #86732 (Instanced wire objects are hidden in solid mode instead wire appearance)\n\n", "sculpt mode artifacts if displace modifier is active\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce RTX 2070 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 441.28\n\nBroken: version: 2.82 (sub 7)\nWorked: (optional)\n\nIn sculpt mode if the displace modifier is active the brush behavior is unpredictable, especially with the smooth brush.\nIt looks like that the sculpting mode takes account of the mesh after being modified by the modifier and start the sculpt over it.\n'm reporting this because in other modes the modifier and the actual edit of the mesh are handled as separate thing.\n| [smooth_brush_prob.mp4](smooth_brush_prob.mp4) | [06_24_2022.mp4](06_24_2022.mp4)\n| -- | -- |\n\n- Open any of the attached files and try to smooth or grab as done in the video\n\nthanks!\nMax\n\n\n[smooth_bug.blend](smooth_bug.blend)\n[sculptbug_24062022_1422_44.blend](sculptbug_24062022_1422_44.blend)" ]

[ "Graphical glitch in dyntopo sculpt mode (ghost faces)\nOperating system: Win 10 64bit\nGraphics card: Geforce 940M\n\nBroken: 2.80, 9149e894210 win64, blender2.8, 2018-12-16\n\nUsing a constructive brush in dyntopo mode creates ghost faces, that will disappear after switching to a different mode\n\nenter sculpt mode and activate dyntopo\nchoose constructive brush (draw, snake hook...) and set it to a larger size (to recreate it more consistently)\ndetached faces will appear, that cannot be altered with sculpt tools\nchange to a different mode and they are gone\n\n(side note: I am still new to bug reporting and I don't want to spam, but another glitch as can be seen in the clip is, that the brush cursor doesn't show the first time I enter sculpt mode)\n\n[bug0001-0476.mp4](bug0001-0476.mp4)\n\n", "Dyntopo sculpt and snake hook brush artifacts\nOperating system: Windows 10\nGraphics card: Nvidia GeForce 1060 GTX\n\nBroken: branch: blender2.8, commit date: 2018-12-07 04:54, hash: 606223f6a61, type: Release\nWorked: (optional)\n\nWhen using Dynotopo and doing extreme extrusions like using Snake Hook Brush , if you want to remove those extrusions like with the flatten brush, artifacts remain behind, you can refresh if you go to edit mode and go back to sculpt by pressing the Tab key, but not always works; this didn't happen in previous versions; here's what I mean:\n\n[Dynotopo refresh issue.mp4](Dynotopo_refresh_issue.mp4)\n\nPut an object in the scene; go to sculpt mode; activate dynotopo, then do extreme deformations with snake hook brush; then try to eliminate those extrusions like using the flatten brush for example; parts and bits of the mesh stay behind;\n\nCheers!" ]

[ "Resetting singular values in the Themes disables them\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 471.68\n\nBroken: version: 2.93.5\n\nInstead of globally resetting the default themes, if I use the \"Reset to default\" for a color or a value, it simply puts it to black/0.0.\n[2021-10-27 14-12-52.mp4](2021-10-27_14-12-52.mp4)\n\nGo to themes and right click a color or value.\n[Based on the default startup]\n\n", "Rigid Body Constraints ignoring keyed rotation limits\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 531.79\n\nBroken: version: 3.5.1\n\n\nRigid Body Constraint limits seems to be only taking initial values, ignoring keyed values.\n\n\n1. Open attached blend file\n2. Verify that `Constraint.002` object has Rigid Body Constraint, Generic type, and Z upper angular limit keyed (from frame 1->40, 90->20 deg).\n3. Play animation.\n4. From frame 40 and on, that Z upper angular limit is keyed to be 20 deg, but the cube being hit will still rotate to 90 degrees (expecting limited to 20 deg).\n\nTested \"Generic\", \"Hinge\", \"Generic Spring\", none respects the keyed rotation limit.\n\n-----\n\nOriginal description:\n\n\nGeneric and Generic Spring constraints will ignore keyframed changes in their lower and upper limits.\n\n1. Create 4 Rigid body cubes and a passive ground plane under them.\n2. Line 3 of the cubes up on the X axis.\n3. Connect each of the 3 cubes with rigid body constraints set to generic.\n4. Set one of the cubes on the end to animated.\n5. Select the constraint between the animated cube you just made and the next one in line and set all limits to 0.\n6. Select the next constraint in the line and set all angular rotation limits to zero except the z angle upper limit, set that to 90.\n7. Set a keyframe on the upper limit Z axis of that same constraint on frame 1.\n8. Go to frame 40 and set the upper limit Z axis to 20 and keyframe it.\n9. Take the 4th cube you made and position it behind the cube on the end that is not set the animated.\n10. Take that same cube and make it animated and keyframe it on frame 40.\n11. Move that 4th cube to hit the none animated cube at the end of the line and keyframe it.\n12. Then hit play, the constraint will ignore the changed upper limit and rotate to 90 even though it shows as being only 20 in the properties panel.", "LOAD UI - Checkbox Status Ignored\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3060 Ti/PCIe/SSE2 NVIDIA Corporation 4.6.0 NVIDIA 536.40\n\nBroken: version: 4.0.0 Alpha\nWorked: (Unknown)\n\n\nWhen opening a file with a UI that does not match the default UI, the file's internal UI is loaded regardless of the LOAD UI checkbox being marked \"OFF\". More specifically, the MODE of a workspace is not respected and uses what ever mode is used within the file.\n\nIE, the default is \"Object\", while a scene can override it to something else - in this example, Texture Paint.\n\nAs the mode is part of the workspace setting, it should not be overriding when the scene is opened.\n\n\n\n\nSee screenshots.", "Broken material assigment after changing mesh-linking to object-linking\nWindows 11\nNvidia RTX 2060 super\n\nBroken: blender-3.2.0-stable+daily.e05e1e369187\n\n\nOn objects with multiple material slots, the currently selected slot gets bugged when switching from object-linking to mesh-linking. It allows the user to select a different material for each link type, which seems to trigger the bug and has lingering effects. For example, in geometry-nodes setting the material becomes unpredicatable, with material names seemingly swapped.\n\nFrom the provided blend file or startup scene:\n1. create a new material slot\n2. pick a material for this slot\n3. swap the material link type between object-link and mesh-link\n4. select a different material for the same slot\nThe material or slot is now bugged. Geo nodes 'set material' will become unpredictable.\n\n[bug_slots.blend](bug_slots.blend)", "Problem with parent_vertices attribute\nOperating system: Microsoft Windows 10 Professionnel\nGraphics card: Nvidia Gefroce RTX 2080 Super\n\nBlender 3.0.0, hash: f1cca3055776 , branch: master\n\n**Short description of the bug**\nThe attributes parent_vertices is by default to [0,0,0] but it cause some problem :\nWhen you select the 0th vertex of a mesh and parent an other object with the vertex_parent_set() function the list still be the same [0,0,0].\nSo when the parent mesh is deleted you cannot know if there were a parent before. \nMaybe set the default list to [None, None, None] will avoid the problem.\n", "Undo doesn't work with paint mask selection (texture paint vs. vertex/weightpaint)\nOperating system: Linux-5.2.10-arch1-1-ARCH-x86_64-with-arch 64 Bits\nGraphics card: Radeon RX 580 Series (POLARIS10, DRM 3.32.0, 5.2.10-arch1-1-ARCH, LLVM 8.0.1) X.Org 4.5 (Core Profile) Mesa 19.1.5\n\n\nBroken: version: 2.80 (sub 75)\nWorked: (optional)\n\n\nUndo does not work with paint mask selections. The expected behavior here is that undo should work as in other selection modes.\n\nClick on texture paint.\nSelect single image and create a texture.\nTurn on paint masking, and then select the \"select box\" tool.\nclick on a face on the cube to select it.\nPress ctrl-z.", "`Materials` tab not opened at the first time running the script\nOperating system: macOS 10.14.2 (18C54)\nGraphics card: Intel Iris Plus Graphics 640 1536 MB\n\nBroken: (example: 2.80, 89207DF7222A, master, 2019-05-27, as found on the splash screen)\n\n`Materials` tab not open at the first time running the script in attached .blend file\n\n\n* Open the attached .blend file\n* Click `Run Script` the first time\n* It just opens the `Object` tab\n* Click `Run Script` the second time\n* It then opens the `Materials` tab\n\nIt should open Materials tab at the first time clicking `Run Script`\n\n[bug_001.blend](bug_001.blend)", "Nodes: Group input subtype \"Angle\" max values are in radians, but the default implies degrees.\nOperating system: Linux-5.4.0-148-generic-x86_64-with-glibc2.31 64 Bits, X11 UI\nGraphics card: GeForce GT 630/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 390.157\n\nBroken: version: 3.6.0 Alpha\n\nThe subtype \"angle\" in the socket editor provides a default value in °, and a min and a max value, which seem related to the default setting in degrees, so the user assumes the slider in the modifier frontend UI to stop at the set values. Which it doesn't. Set max to say 1, and the slider will stop at 57,3° (180/pi), not at 1°.\n\n\n", "When changing the node type in material, sockets of different node types are reused\nOperating system: MacOS\nGraphics card: Apple M1\n\nBroken: main\nWorked: don't know\n\nShader nodes that have the same input sockets can have different default values.\nBut if you switch between shader nodes in the material interface, then the default values will first of all be copied from pre-existing socket of a node of a different type. Because of this, default values may not be used and may look strange.\n\nIn the default scene, assign Glass BSDF material to the default cube, and check the \"Roughness\" field. Now assign another material without \"Roughness\", such as Hair BSDF, then assign Glass BSDF material again and check the \"Roughness\" field.\n\n", "Cycles does not respect the texture alpha mode\nOperating system: Gentoo Linux \nGraphics card: AMD 290x\n\nBroken:\n2.79b\n2.80 f15205b93f56e454b76d5a222fdb9a9665b99907 2018-01-02\n\nEevee respect the alpha mode in the texture settings, however cycles does not.\n\n[alpha_test_pack.blend](alpha_test_pack.blend)\n\n1. Load the attached .blend file\n2. Change the texture settings for the `Alpha:` setting for the image texture node (with use alpha on).\n3. Notice that in cycles the `Alpha:` setting has not effect, the texture alpha is always premultilied.", "Outliner: Data API view can expand non expandable properties\nThe Data API view in the outliner let you click on the (invisble) expand arrow on non expandable properties.\n\nThis result in the property value to become invisible. You can click to bring it back, so I think it's only a display issue.\nNot all properties seems to be affected.\n\nBefore:\n\nAfter:\n", "bpy.props.IntProperty fails to register if given float values\nOperating system: Windows-10-10.0.19042-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1650 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 496.76\n\nBroken: version: 3.1.0 Alpha\n\nWhen an addon tries to register an IntProperty, whilst not giving one of the numerical values an explicit int value, the property will return an error during registration.\n\n```\nprop = bpy.props.IntProperty(default=1.0) # prop is created okay\nbpy.types.Object.intprop = prop\nTypeError: integer argument expected, got float\nTraceback (most recent call last):\n File \"<blender_console>\", line 1, in <module>\nValueError: bpy_struct \"Object\" registration error: 'intprop' IntProperty could not register (see previous error)\n```\n", "Cloth physics cache gets reset by changing object materials\nOperating system: Darwin-19.4.0-x86_64-i386-64bit 64 Bits\nGraphics card: AMD Radeon Pro 5500M OpenGL Engine ATI Technologies Inc. 4.1 ATI-3.8.24\n\nBroken: version: 2.83 (sub 11)\nWorked: unknown\n\nWhen using a cloth physics simulation, if you add or remove a material from an object, the cloth system cache is reset (\"cache is outdated\"). I don't think the material affects the cloth physics in any way (as it seems you can change any parameter within the material without the cache resetting), and therefore this should probably not happen. Please correct me if I am wrong and the material does affect the physics somehow.\n\nIn the default scene, enable cloth physics for the cube. Let the playback run for a few frames, building up the cache. Add a material to the cube; now the cache gets reset.\n\n", "Erratic behaviour of materials/material slots for text objects\nOperating system: Linux-5.19.0-45-generic-x86_64-with-glibc2.35 64 Bits, X11 UI\nGraphics card: NVIDIA GeForce GTX 980M/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 510.108.03\n\nBroken: version: 3.6.0\n\nMaterial assignment for text objects doesn't behave as expected. Removing an unused material slot will cause changes to the materials used by the object. The same happens if you rearrange the material slots.\n\n1. Add a new text object. Leave the default string \"Text\".\n2. Add 3 new material slots with different materials to the object. Use different viewport display colors to be able to see changes.\n(You will now have Material.001, Material.002 and Material.003)\n3. In edit mode, select the last two letters and assign Material.002.\n(You should now have \"Te\" using Material.001 and \"xt\" using Material.002)\n4. In object mode, remove the last material slot containing Material.003.\n(This material was not assigned to any character, and removing it should not cause any change.)\n5. All 4 letters are now using Material.001. \"xt\" should still be using Material.002.\n\nAlternatively, execute step 1-3 above and then.\n4. Move the last material slot up to the first place.\n(This should not cause any change. Normally in blender, the materials used by the geometry will not change if you rearrange the material slots.)\n5. You will now have \"Te\" using Material.004 and \"xt\" using Material.002. \"Te\" should still be using Material.001.\n\nI have attached a blend file where you just need to remove the last (unused) material slot to see the problem, or move the last material slot up to first position.\n\n", "Sequencer: Driver on any strip property does not invalidate its cache [prefetch frames]\nOperating system: Darwin-18.6.0-x86_64-i386-64bit 64 Bits\nGraphics card: Intel Iris OpenGL Engine Intel Inc. 4.1 INTEL-12.9.22\n\nBroken: version: 2.83.9\n\nSequencer: Driver on any strip property does not invalidate its cache [prefetch frames]\nThis can be any driver on any driven strip property:\nDoes not matter which property is the driving property (could be x Location of default cube as well)\n\n\n[#82776.blend](T82776.blend)\n- Open file\n- `driven` strips opacity is driven by Cube 's X location\n- change Cube 's X location --> driven strips opacity value changes, but the cache is not invalidated\n\n---\n\nOriginal report:\n\n Driver on `wipe.effect_fader` doesn't invalidate cache for driven wipe sequences. It's only invalidates the cache of the currently displayed wipe. This leads to display errors.\n\nNB: The issue is the same for any other parameter (e.g angle, blur...)\n\n- Open this file: [Bug_Wipe_Driver.blend](Bug_Wipe_Driver.blend)\n- There are two wipes, the first is the driver, the second one has the `effect_fader` value driven. Prefetch is enabled.\n- Move the `effect_fader` value of the driving wipe, its cache is reloaded, but not the cache of the driven one.\n\n" ]

[ "Blender Preferences > Themes > 'Reset to Default Value' always return the color black\nOperating system: Windows-10-10.0.17763 64 Bits\n\nBroken: version: 2.81 (sub 1)\n\nBlender Preferences > Themes > 'Reset to Default Value' will always return the color black and not the default color of the current selected theme, as I would expect it to do). \n\nI don't know if this is a bug or a paper cut? Or if it is supposed to zero all color values, maybe the entry in the Menu should be renamed accordingly?", "Re-evaluate \"Reset to Default Value\"\n# Re-evaluate \"Reset to Default Value\"\n\n## Motivation\n\nUsers have the option to reset a property to its default vaule (in UI: {nav RMB Menu > Reset to Default Value}). However, this barely works, in the majority of cases it just sets it to its minimum value. So as it is, it can definitely be considered broken.\n\n## Keep it or wipe it out?\n\nJust wiping it out would safe us some work, but IMHO this feature would be very useful - if it worked. It gives users some safety, since they don't need to be afraid of breaking something by playing around with values. (I personally wanted to use it all the time.)\n\n## How to solve?\n\nFirst of, we need to ask what we define as the default value. AFAICS there are three potential default values:\n\n* RNA default value\n* Startup file value\n* User defined value\n\n**RNA default value:**\nRNA already supports default values, assigning one to a property is a one line fix. The issue is that it had to be done for all properties that miss it (quickly checked, more than 2000, maybe 3000 properties).\nTo solve this, I'd propose we add a maniphest task that lists all RNA files and ask people to submit patches that add default values for theses files. Think operator RNA properties could be neglected a bit.\nNote that these default values should match the ones from the startup.blend, meaning they had to be kept in sync too.\n\n**Startup file value:**\nAFAIK it's possible to get the property value of a .blend using Python, so we could do this with the startup.blend.\nUsing this solution we wouldn't have to worry about keeping startup.blend and RNA defaults in sync.\n\n**User defined value:**\nThis should work 'on top' of one of the other two solutions, and it's obviously the solution that would be the most configurable. It could be set in a popup that is invoked from the context menu (this popup could also allow resetting to non-user defined default)." ]

[ "Bake fluid domain simulation crashes when using a depsgraph poll\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce RTX 2070 with Max-Q Design/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 462.59\n\nBroken: version: 2.92.0\n\nWhen using bpy.context.evaluated_depsgraph_get() in a poll and then baking a fluid blender crashes\n\n1. use the provided file with a fluid prepared to bake data or create a fluid domain and a fluid source\n[mantaflow.blend](mantaflow.blend)\n\n2. make sure the chace type is set to modular\n\n\n3. run the simple operator script (should come with the blend file)\n[operator_simple.py](operator_simple.py)\n\n4. press the bake Data in the Fluid domain settings and blender should crash\n\n\n\n", "Cycles dupli object baking issues\nnvidia g880m\n\nBroken: blender-2.73-d56c9fb-win64\nWorked: (optional) probably never\n\nwhen assigning object as duplis to some mesh, and then trying to bake these duplis to another mesh, duplis appear black. But in normals mode, it works, so they are there, just without material or something.\nAlso, you can see that the duplis are there actually twice, one offset quite strongly. In some other file, I got them also four times...\n\n\n[bakebug.blend](bakebug.blend)\nopen the file, hit bake. \nthe group of objects in the ceter bakes to the cube.\n\nalso, you can see another group of objects, where duplis have been made \"real\" and it works well, it's group of objects called \"fine\", baking to the cube works well with seemingly same meshes.\n", "Some Cycles nodes are active when muted (RGB To BW, Math, vector Math, Separate RGB)\nWin10 64bit\n\nBroken: 2.79\n\nRGB To BW node is active when muted\n\nOpen file and render\n[RGB_BW.blend](RGB_BW.blend)\n\n", "Normal map node worked is not correct\nOperating system: win10 x64\nGraphics card: gtx 1650\n\nBroken: 3.5.0, 1be25cfff18b, 2023-03-29\nBroken: 2.93\nWorked: 2.92\n\nCaused by 1c22b551d0\n\n\nI do not think that this is an error or a bug, rather it is not a refinement. \nIn some cases, the normal map node does not display them as you would like. \n\n\nI created a Hight poly model and bake it into a plane. Then I applied the resulting texture to the Low poly base mesh. It didn't display correctly there, only part of it smoothed out, but the other part showed sharp edges, very similar to the Fresnel effect\n\n", "Blender 3.0 Denoising on Transparency Creates Artifacting (Blend included)\nOperating system: Win 10 (latest update)\nGraphics card: RTX 3090\n\nBroken: (3.0.0 and 3.0.1 dc2d18018171 Branch: Master)\nWorked: (2.9.0)\n\nWhen enabling transparency under \"Film\" settings, artifacts occurs in the transparency channel\n\nOpen attached .blend in 2.9.0 and 3.0.0 and hit render to see. I've set-up a compositor node setup to highlight and visualize the artificating found in 3.0.0. Try rendering with both Denoising methods to see variations in artifacts.\n\n**Other notes**\nI've played with as many render settings as possible and found that the denoiser is responsible. OptiX is worst, OpenImage is not as bad in artifacting.\n\n[alpha-transparency-denoising-issue-sample.blend](alpha-transparency-denoising-issue-sample.blend)\n\nNotice the clean colors in the greys (background composited in Compositor to help see issue)\n\n\nNotice the vertical and horizontal line noise\n\n\n[system-info.txt](system-info.txt)", "Mix Node\n# Mix Node\n\nThis task outlines a design proposal for a `Mix` function node.\n\nSimilar to `Mix RGB`, the `Mix` node performs a linear interpolation between two values. The current way to perform this for floating-point values is to either use `Mix RGB` (which implicitly converts the values to colors) or to create a custom node setup or node group comprised of low-level math nodes. Either approach is not straightforward to users.\n\n\n\n\n - Factor: Controls the amount of influence the node exerts on the output value.\n\nThis node will likely live under the *Utilities* category.\n\n## Use cases\n\nThis functionality is used often to blend between two values.\n\n## Potential Issues\n\nMaybe lack of immediate foresight to see other mix possibilities in other \"Everything Nodes\" projects?\n\n## Additional Questions\n\n- Do want to include additional interpolation types?\n- Do we want to include different data types?\n - 3D Vector\n - 2D Vector (when it is added)\n - Float\n - Integer (the output would also be an integer) (Truncated?)\n - Other questionable types: Euler (perhaps separate node for Slerp?), Image (Blend node instead?)\n- Should *Factor* not be automatically clamped to allow for extrapolation?\n- ???\n", "Edge Sliding with subdivision surface is extremely laggy (caused by \"bad\" UVs and UV Smoothing in the modifier)\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 527.56\n\nBroken: version: 3.4.0\n\nSo I've been modelling a raptor the last week and I noticed that viewport performance in edit mode was extremely laggy with subdivision surface modifier enabled. I checked the face count and it was just 150k faces which is not a lot at all. It's definitely a modifier problem because when I apply the subd modifier the lag goes away completely even when sliding twice the amount of edges around. I thought that was maybe a hardware issue so I opened task manager and while sliding edges with subd on I get like 4 fps with severe stuttering to top it off. All that while the cpu was at 10% usage and 5% gpu usage with memory barely touched (6gb used out of 64gb). I have nothing else running in the background like an ativirus or music app, I have the latest version of windows 10, latest gpu and chipset drivers, latest bios etc... I even have gpu subdivision enabled without autosmooth to make sure it actually works properly. I even run an sfc/scannow to check for other kinds of issues but nothing changed. Subd in edit mode however is not only slow when sliding edges and vertices but also in adding loop cuts or insetting faces. It is extremely frustrating so it would be great if you could fix this. I know edit mode was not made to handle massive polycounts but 150k is again not a whole lot and definitely comfortable to run on my PC. Here is the file to see for yourselves.\n\n- Open attached .blend\n- Execute Edge slide/loop cuts/inset faces\n[Broken Subd.blend](Broken_Subd.blend)\n\n", "Crash when baking Dynamic Paint Image Sequence multiple times in loop\nBroken: 2.90.0 Alpha\nWorked: blender-2.80-windows64\n\nBlender crashes when baking dynamic paint image sequence in for loop. Depending on wait time crash will have differens stack traces.\n\n[#74052.blend](T74052.blend)\n\n- Open file\n- Run script", "Cycles GPU CUDA rendering high number of points seems broken in LTS (main is fine, request for backport)\nOperating system: Linux-5.15.0-58-generic-x86_64-with-glibc2.31 64 Bits\nGraphics card: NVIDIA GeForce RTX 3070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 515.86.01\n\nBroken: version: 3.3.3\nWorked: latest 3.5 alpha\n\nRendering point \"clouds\" with Cycles - CUDA seems to make some chunks of points disapear randomly\nThis looks like this has been fixed in 3.5, but i didn't find when. Can this be fix in LTS ?\n\nDelete the default cube and add a plane\nAdd a Geonode Modifier on that plane\nDistribute Points on Faces > Density : 250'000\nSet Point Radius > 0.0025 (for convenience)\n \nSwitch to Cycles, GPU\nSwitch the render device to CUDA\n\nResults with :\n\n|  CPU |  CUDA |  OPTIX\n| -- | -- | -- |\n\n\n\nBug is visible in both viewport shading and Render.\n\n", "Geometry Nodes: Color mix nodes don't effect the alpha\nOperating system:\nGraphics card:\n\nBroken: 3.4.1\n\nwhenever I try doing any math with colors, the alpha isn't affected when I use the color mix node in any blend mode. \n\n1. Open blender geometry nodes.\n2. Add 2 combine color, color or whatever node with different different values for the alpha's.\n3. Use a mix color node with any operation like add, subtract etc.\n4. See that the alpha didn't change (use the spreadsheet or the hover over).", "Eevee - visual artifacts during irradiance volume baking\nWindows-10-10.0.16299 64 Bits\nGeForce GTX 1080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 419.35\nBlender 2.80 (sub 51)\n\n[irradiance_volume_bake_artifacts.blend](irradiance_volume_bake_artifacts.blend)\n - Open attached file\n - Switch to rendered mode (Eevee)\n - Press \"Bake Indirect Lighting\"\n\nYou should see squares flickering:\n\nVideo:\n[Desktop 2019.03.27 - 09.18.07.01.avi](Desktop_2019.03.27_-_09.18.07.01.avi)", "Subdivision Surface modifier performance issues\nOperating system: Xubuntu 18.04\nGraphics card: Nvidia Titan X (Pascal)\n\nBroken: 2.80 26d5a3625ed\nWorked: Before the new \"Quality\" setting was introduced in the Subdivision Surface modifier\n\nThe Subdivision surface modifier has slowed down significantly, especially noticable with multiple Subdivision Surface modifiers on the same object.\nCreating the same setup in the builds before the new setting is as fast as 2.79.\n\n- Create Monkey object\n- Add Subdivision Surface modifier\n- Add Wireframe Modifier\n- Add second Subdivision Surface modifer\n[slow-subsurf.blend](slow-subsurf.blend)", "Motion Blur weird artifacts when rendering MeshSequenceCache USD\nOperating system: Windows 11\nGraphics card: RTX 2080 Ti\n\nBroken: 3.6 LTS, but it also seem to happen in older versions\n\n\nI'm getting some weird artifacts when rendering a USD mesh cache with motion blur enabled.\n\nI've exported the mesh from maya 2023 using the Arnold USD export method. \n\nWhat i've tried:\n\n- Disable Auto Smooth Normals and removing every Subdivision Surface Modifier.\n- Trying to change the position of the motion blur from Center to Start or End\n\nThe problem persist.\n\nExport a high res mesh animation from Maya using Arnold USD as export option.\n\nImport it in Blender and try to render the sequence with Motion Blur activated.\n\nBlend file Link : view?usp=sharing \n", "EEVEE-Next Irradiance Baking\nDue to time constraints it isn't possible to start implementing a solution that leverage raytracing. So the implementation will try to minimize the amount of changes to the current system while cleaning up the implementation and changing the baking algorithm.\n\nHere are the reference papers the new solution will be based on:\nSIGGRAPH2022-Advances-Enemies-Ciardi%20et%20al.pdf (Probe-based lighting section)\nHobson_Josh_The_Indirect_Lighting.pdf\n\nOne remaining unknown is whether or not we keep the light-grids behavior as it was in the original EEVEE or if we go for a single large light-grid structure.\n\nHere is a breakdown of the tasks:\n- [x] #105802 Add debug visuals for surfels.\n- [x] Port Irradiance Cache display.\n- [x] Port Spherical Harmonics lib & BSDF lib and add unit tests.\n\n#### Milestone: Dummy irradiance grid displaying.\n- [x] Rewrite cache structure load / store.\n- [x] Rewrite light-cache baking in C++ with new draw-manager and unit-tests.\n\n#### Milestone: Baking, Saving to and Loading from disk should work.\n- [x] Make shader variation to output surfel.\n- [x] Bin surfels to ray tiles.\n- [x] Sort surfels in ray tiles, dead simple sorting for now.\n- [x] Traverse the rays and add contribution to nearest surfels.\n- [x] Insert irradiance samples to ray list.\n- [x] Project result to irradiance grid.\n\n#### Milestone: Irradiance baking fully works.\n- [x] Add indirect diffuse contribution to BSDFs\n- [x] #110386 Add dilation for invalid lightprobes\n- [x] #110312 Add sampling biases\n- [x] #110851 Add leak prevention\n- [ ] #106449 Finalize the design and implement the behavior of multiple irradiance grid.\n\n#### Bugs / Cleanups\n- [x] Add memory budget option\n- [ ] Add safety check for out of memory cases\n- [x] #110858 Add back intensity option\n- [x] #110858 Add back clamp option to avoid fireflies\n- [ ] Add smooth transition between irradiance volumes\n- [ ] Fix append / link of baked data\n- [ ] Add back capture bounds parameters\n- [ ] Add option on light and object to exclude from irradiance cache\n- [ ] Add option on material to be considered double sided by irradiance bake\n\n#### Milestone: Irradiance bake minimum-viable-product for 4.0 release.\n\n#### Extended targets\n- [ ] Support transparency\n- [ ] Support translucency\n- [ ] Support volumes during baking\n- [ ] Sky visibility baking\n- [ ] Auto Sample Resolution based on density\n- [ ] Sparse grid\n- [ ] Light injection\n- [ ] Multi-LOD grid\n\n#### Dropped features (not considered regressions)\n- Capture Clip Near / Far distances: Capture is done for the whole scene.\n- Bounce Count: Light transport is done using an infinite number of bounce and converge to the correct solution with more added samples.\n", "Cycles Bake Selected to Active with transparency gives undesirable results\nSystem Information:\nOperating system: Windows-8.1-6.3.9600 64 Bits\nGraphics card: GeForce GT 635M/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 391.35\nBroken: version: 2.81 (sub 16)\n\n\nBug or unintended behavior:\nBaking using \"Selected to Active\" does not take into account transparency \n\nHow to reproduce:\nCreate two objects with materials and textures assigned to each;\n```\n One texture to be baked to\n One texture with an alpha channel (optionally a normal map)\n```\n\nIn the shader editor, modify the material of the object with a transparency channel\n```\n Join the transparent textures' alpha into the alpha of the shader that is being used\n (optionally join the color of a normal map)\n```\n\nIn render properties, enable Selected to Active and use an appropriate ray distance\n```\n Finally bake.\n```\n\nResult:\nWhen baking combined or most channels, the transparent areas will assume the color as if it were completely opaque,\nWhen baking normals with transparency, the transparent areas will mix with the normal of the geometry that is being baked to the active selection.\nBoth outcomes are unfavorable when working with transparent cards/ribbons.\n\n[bugreport.blend](bugreport.blend)\n\nI packed the textures edited the project to make the issues that I am having more clear.\n\n{[F8286028](image.png), size=full}\nThere are now 3 overlapping objects with transparency.\nAfter baking normals or color, none of the colors or normals behind the transparent objects appear because the object appears to draw over top without transparency.\nThis example also shows that with full normal transparency, the normal will then be the geometry normal of the object rather than a flat normal or anything that is below the object." ]

[ "Cycles Baking gives different results than Cycles Render\nWindows 7 x64, NVIDIA GTX970\n\n2.73\n\nCycles Baking gives different results than Cycles Render, which is particularly visible on procedurally generated materials and it makes them unusable for baking. This issue is probably caused by lack of anti-aliasing or some kind of sampling, which is enabled for Render, but not enabled for Baking. I have found that after enabling certain parts of code in kernel_bake.h, I can get more consistent results from Cycles-Baking (not the same as from Render, but really usable).\n\nI created 2 procedural materials, which I used to test rendering and baking (everything using CPU, 20 samples render, result images 512x512).\n\nThe baked images \"No AA\" come from official release 2.73.\n\nThe baked images \"AA Enabled\" come from my compiled version of Blender (64bit build, no OSL), where I enabled the following parts of code in kernel_bake.h:\n\n```\nccl_device_inline float bake_clamp_mirror_repeat(float u)\n{\n\tfloat fu = floorf(u);\n\tu = u - fu;\n\treturn (((int)fu) & 1)? 1.0f - u: u;\n}\n\nuint rng_state = cmj_hash(i, kernel_data.integrator.seed);\nfloat filter_x, filter_y;\npath_rng_init(kg, &rng_state, sample, num_samples, &rng, 0, 0, &filter_x, &filter_y);\n\n/* subpixel u/v offset */\nif(sample > 0) {\n\tu = bake_clamp_mirror_repeat(u + dudx*(filter_x - 0.5f) + dudy*(filter_y - 0.5f));\n\tv = bake_clamp_mirror_repeat(v + dvdx*(filter_x - 0.5f) + dvdy*(filter_y - 0.5f));\n}\n```\n\n\n[Test_CyclesBake.blend](Test_CyclesBake.blend)\n\n" ]

[ "Subdiv hangs blender\n**System Information** DESKTOP-HR30UII\nOperating system:Windows 11 Home Single Language\nGraphics card:2GB NVIDIA GRAPHICS DRIVER\n\n\nBroken: Blender Version 3.5\nWorked: ?\n\nBLENDER 3.5 KEEPS NOT RESPONDING WHILE I USE MODIFIERS \n\nON BLENDER ADD MESH AND ADD MODIFIERS TO SUBDIVISION SURFACE THEN CHANGING THE VALUE OF LEVELS VIEWPORT VALUE ABOVE 4 .THEN ONWARDS BLENDER COMPLETELY STOPS WORKING AND I NEED TO RESTART IT AGAIN\n\n", "Grease Pencil Arrange Stroke operator does not function with Curve Editing\nOperating system: Linux 64-bit\nGraphics card: GTX 1080 ti\n\n2.93.5 and 3.0.0 alpha\n\nIn edit mode, \"Arrange Stroke\" operator does not work when use_curve_edit is enabled.\n\n[GP_arrange_bug.blend](GP_arrange_bug.blend)\n\n - Open attached file\n - Use any of the Arrange Stroke operators on the selected stroke\n # Disable Curve Editing and try again to see that it otherwise works as intended\n\n", "Regression from 3.3: Invalid button mapping on SpaceMouse Enterprise (256f:c633)\nButton mapping on this device is now completely scrambled. For example pressing \"Shift\" deletes selected object, \"Lock rotation\" is interpreted as play/pause, \"Button 6\" rotates camera 90deg CCW and \"1\" brings up the menu. Other buttons do not seem to register.\n\n\nBus 003 Device 013: ID 256f:c633 3Dconnexion SpaceMouse Enterprise\n\nTested on macOS Ventura and Ubuntu 22.04\n\nIn blender 3.3.2 on Linux and macOS, some buttons function correctly:\n - Shift\n - Ctrl\n - Alt\n - Rotate\n - Top\n - Front\n - Right\n\n Basically all other keys in 3.3.2 bring up the on-screen menu.\n\n\nTested versions: 3.4, 3.5 and 3.6.\n\nOperating system: macOS-14.0-arm64-arm-64bit 64 Bits\nGraphics card: Metal API Apple M1 Pro 1.2\n\nBroken: version: 3.6.0\nWorked: 3.3.2\n", "Auto-Merging keyframes in the Driver Editor has unappropriate threshold (merges keyframes that shouldnt)\nOperating system: windows 10 & Linux ubuntu\nGraphics card: GTX 1070\n\nBroken: 3.6.2\nWorked: unknown\n\nSee att. blendfile and pictures; moving the base of the armature in the Y direction drives the shapekey of the torus.\nIf you edit a driver keyframe by just slightly moving it to the right, the keyframe next to it is removed\n\nTry editing keyframes in the driver f-curve editor\n\n\n**Note**\nDrivers editor has the driver value as the x-axis (this can be arbitrary low, e.g. in the report, this was the x location [in meters] so handling quite small values where \"keyframes\" where ~0.01 apart\n\n`BKE_fcurve_merge_duplicate_keys` acts with a threshold of `BEZT_BINARYSEARCH_THRESH 0.01f`", "depsgraph.view_layer.name driver expression doesn't work correctly\nOperating system: Windows 11\nGraphics card: Geforce RTX 3090\n\nBroken: 3.3.5 LTS\nBroken: 3.5\n\nI use a driver expression to control modifiers visibility or shader values depending on a view layer name. If the needed view layer is active, then values change. All works well when switching viewlayers manually inside the viewport of my scene. But as soon as I render the scene, the expression is ignored and nothing happen.\n\n* Open attached file\n* Change view layer few times\n\nDriver will change only once. This also doesn't work when rendering.\n\n", "Not recognizing numpad page up /down keys .\nOperating system: Windows-10-10.0.17134 64 Bits\nGraphics card: GeForce GTX 1050 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 431.60\n\n\nBroken: version: 2.80 (sub 75)\nWorked: (optional)\n\n\nBlender is failing to recognize the numpad keys !!!\n\nTrying to use the change workspace shortcut ctrl+PageUp /down and blender is not recognizing the keys in numpad and keep assuming the key 9 and 3 . Other software is fine and i can use Page up and Down, and the home end keys. \n\n", "Menu label does not update after \"Remove Shortcut\" from context menu\nOperating system: Darwin-19.5.0-x86_64-i386-64bit 64 Bits\nGraphics card: AMD Radeon Pro 455 OpenGL Engine ATI Technologies Inc. 4.1 ATI-3.9.15\n\nBroken: version: all\n\nAfter \"Remove Shortcut\", the menu label is not updated properly.\nJust for a moment, but it's noticeable.\n\n**Only in the case when the \"Remove Shortcut\" menu item (mouse cursor actually)\nis located above the deleted item or above separator.**\n\n", "Adding new hook while in Edit mode breaks when trying to undo/redo the operator\nWin7 (32bit)\n\nBroken: 2.78a\n\n\n- Add Path from Curve menu, add new Armature\n- Assign Bone to control point as Hook\n- Blank Hook modifier appears along with real one\n\nScreencast: [screencast0001-0250.mp4](screencast0001-0250.mp4)\n", "Average: Custom Normal's functions probably doesn't work\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.94\n\nBroken: version: 3.3.1\n\nMesh > Normals > Averame > Custom Normal functions (F9) probably doesn't work.\n\n1) Add any mesh. \n2) Switch mode to Edit Mode.\n3) Select all components.\n4) Call menu Mesh > Normals > Average > Custom Normals. At this point normals changed.\n5) Open bottom menu or press F9 and try change any sliders.\nChanges will not occur. I assumed that this was a mistake, because. these sliders are repeated in the modifier Weighted Normal, but there they work fine!\n", "Transform has hard-coded keys to the modal keymap\nCurrently some keys used during transform are hard coded, a few aren't available in the modal keymap either (`O`, `C`, `N`), see `transformEvent`.\n\nIt's not useful to be able to edit the modal keymap when there is also functionality only accessible from hard coded keys.\n\nAll functions should be moved to the modal keymap and hard coded keys removed.", "Can't create a working shortcut for Extract Face Set\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 980/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 471.96\n\nBroken: version: 3.0.0 Alpha\n\n\nCan't create a working shortcut for Face Set Extract. Creating one does appear in the Keymap, but not in the Face Sets menu. Can only manually use the operator in the menu to make it work.\n\n[Desktop 2021.09.05 - 05.28.09.02.mp4](Desktop_2021.09.05_-_05.28.09.02.mp4)\n\n1. Create a shortcut for Extract Face Set.\n2. Check that it is in the Keymap to be sure it's there.\n3. Pressing the keys to perform the shortcut will not work. The shortcut text will not appear in the menu as other shortcuts do in Blender.\n", "Autokey keys disabled driver values\nOperating system: Darwin-17.2.0-x86_64-i386-64bit 64 Bits\nGraphics card: NVIDIA GeForce GTX 1050 Ti OpenGL Engine NVIDIA Corporation 4.1 NVIDIA-10.27.6 378.10.10.10.20.109\n\nBroken: version: 2.83 (sub 2)\nWorked: (optional)\n\nWhen you have autokeying turned on for LocRotScale, then disable a driver (for say a shape key) and then manually tweak the value. Blender will key this change in the driver editor.\n\n- turn on autokey (keying set LocRotScale will do)\n- create a property on the default cube\n- copy as new driver\n- create a shape key\n- paste this driver to this shape key\n- then disable the curve for this driver in the driver editor (check box)\n- manually tweak the driver value and this will be keyed in the driver editor.\n\n[autokey_disabled_drivers.blend](autokey_disabled_drivers.blend)\n\n[auto_keying_disabled_drivers.mp4](auto_keying_disabled_drivers.mp4)\n\n", "Non-keyboard shortcuts allowed to be set via context menu, but cannot be unset without going to user-preferences\nBroken: 2.79 or latest 2.80 (fa59346c13).\n\nAlthough I can easily (and often accidentally) assign non-keyboard shortcuts to menu items, I cannot easily undo it.\n\nFrom the viewport: View > Frame All [RMB] > Change Shortcut.\n\nAlthough the UI says \"Press a Key\", if you mouse click it (e.g., LBM) will register as the new shortcut.\nNow if you try to change the shortcut again it says it is a Non-Keyboard shortcut and cannot be changed.\n\nTooltip: *\"Only keyboard shortcuts can be changed that way (...)\"*", "Python: Large memory leak if an operator modifies certain properties during the \"load_post\" handler\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070 with Max-Q Design/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 511.09\n\nBroken: version: 3.3.3\nBroken: version: 3.4.x through 3.6 as well\nWorked: n/a\n\nThere's a large memory leak if an Operator modifies certain properties during the `load_post` handler\n\n\n- Install the provided `leak.py` addon\n- Enable it and save preferences\n- Quit Blender\n- From the command line execute `blender --debug-memory`\n- Once loaded, quit blender again\n- Notice the large amount of leaks\n\n`Error: Not freed memory blocks: 543, total unfreed memory 0.040207 MB`\n\nThe property being modified -- see the 2 \"NOTE\" comments in provided addon for clues:\n```python\n def execute(self, context):\n \n # NOTE: This statement causes a large leak\n context.space_data.show_region_tool_header = False\n```\n\nThe leaks looks like:\n```\nError: Not freed memory blocks: 552, total unfreed memory 0.040894 MB\nWM_event_add_keymap_handler len: 80 000002128B026558\nWM_event_add_dropbox_handler len: 40 00000212EF01CE58\nWM_event_add_ui_handler len: 80 000002128B0264B8\nWM_event_add_keymap_handler len: 80 000002128B026418\nWM_event_add_keymap_handler len: 80 000002128B026378\nWM_event_add_keymap_handler len: 80 000002128B0262D8\nWM_event_add_keymap_handler len: 80 000002128B026238\nWM_gizmomap_add_handlers len: 40 00000212EF01CDE8\nWM_event_add_keymap_handler_dynamic len: 80 000002128B025AB8\n...\n...\n```\n\n", "Freeze on performing any operation when compositor tab is open\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.40\n\nBroken: version: 3.3.1\n\nFreeze on performing any operation when compositor tab is open.\nSubsurf modifier seems important for redoing the issue \n\nThis is the continuation of #99695 \n\nThe Bug in #99695 was fixed, but there seems to be more conditions which make the Compositor slow.\n\nDorian from Scatter5 was so nice and made again a simplified version of a secene with the bug.\n\nThe slowdown caused by an unkown combination of geometry nodes, which makes the Compositor unusable slow.\n\n1. Open the attached file\n2. Go to the Compositor\n3. Enjoy the freeze\n4. Try to disconnect a node in the Compositor\n5. Again, enjoy the freeze\n\n[COMPOSITOR SLOW.blend](COMPOSITOR_SLOW.blend)\n" ]

[ "Changing the Driver while Auto Keying is on, causes new Keys on it's F-Curve\nOperating system: Windows-10-10.0.19043-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 472.39\n\nBroken: version: 3.0.1\n\nChanging driver's expression to produce different result, while Auto Keying is on causes creation of Keyframes on Driver's F-Curve.\nWhile thechnically explainable, unfortunately, a scenario where this wouldn't be harmful (and quite possibly unnoticed) is unlikely.\n\n[driver-fcurve-keyframe1.blend](driver-fcurve-keyframe1.blend)\n- Click on the purple/driven number field of \"Curve Out X\" in the Properties Editor Bendy Bones panel (dont do this in the Drivers editor, needs to be on the field directly)\n- change \"max(x,0)*0.05\" to \"max(x,0)*0.05+2\"\n - > The changed value of the Driver generated a new Keyframe in the driver's F-Curve (seen in Drivers Editor below)\n\n(works fine [without adding a keyframe] when changing the expression in the Drivers Editor)" ]

[ "`user_remap` fails silently when object is in edit mode\nBlender 3.6\n\nWay to reproduce:\n1) Create new empty blender project\n2) Select cube\n3) Enter edit mode\n4) Run the code below\n\nResults - mesh wasn't replaced and no errors occured. Expected - to either mesh to change or get some warning/error.\n\n```python\nimport bmesh\nimport bpy\n\nobj = bpy.context.active_object\n\ndata = bpy.data.meshes.new('test')\nbm = bmesh.new()\nbmesh.ops.create_monkey(bm)\nbm.to_mesh(data)\nbm.free()\n\nobj.data.user_remap(data)\nassert obj.data == data\n```\n\n", "Remesh modifier causes splitting wrongly\nWindows 7 Professional Service Pack 1 64-bit\nIntel Core i7-2600 3.40GHz\n8GB RAM\nNVIDIA GTX560\n\nBroken: 2.77 22a2853\n\nRemesh modifier causes splitting wrongly.\nIt happens when remesh modifier's scale value is 3/4, 1/2, 3/8, 1/4, 3/16 etc.\n\n[RemeshGlitch.blend](RemeshGlitch.blend)", "UI change for the Armature Modifier\nThis task proposes a small update of the Armature modifier's UI.\n\nThe issue the UI currently suffers from is that it is unclear that checking \"Preserve Volume\" actually changes the deformation method, instead of just enabling an additional option in the modifier.\nThis is addressed by replacing the checkbox with a dropdown, containing the two available deformation methods: \"Linear Blending\" and \"Dual Quaternion\".\n\nOn the contrary, a similar method cannot be used for the \"binding\" checkboxes, since those two options are apparently not mutually exclusive.\n\nMockup:\n\n\nTooltips:\n- Linear Blending: Deforms the mesh without preserving volume. Prone to candy-wrapping artifacts, but does not suffer from bulging issues.\n- Dual Quaternions: Attempts to preserve the volume when deforming the mesh. Generates bulging issues around joints, but does not have any candiwrapping issues.", "Keymap: custom key modifiers work inconsistently\nOperating system: Arch Linux, no DE, i3wm\nGraphics card: GTX 960\n\nBroken: 2.80\nWorked: 2.79\n\nI'm left-handed on the tablet, so I bind my sculpt hotkeys to the numberpad. Binding NumpadEnter+LMB to sculpt inverted (same as holding Ctrl) no longer works in 2.80. However, binding NumpadEnter+MMB to pan the view does work. Not sure where the conflict is. This setup did work in Blender 2.79.\n\nLoad factory settings\nImport .py keymap provided (or bind manually; sculpt.brush_stroke and view3d.move)\nEnter sculpt mode\nAttempt to sculpt inverted by holding NumpadEnter+LMB\n\n[system-info.txt](system-info.txt)\n[custom_modifier_bug.py](custom_modifier_bug.py)\n\nSimilar bugs from ages ago:\nT44349\nT34822", "Multires: Duplicating objects sets the viewport level to sculpt level\nOperating system: Windows-10-10.0.19043-SP0 64 Bits\nGraphics card: GeForce GTX 1060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 445.87\n\nBroken: version: 3.1.0 Alpha\n\nWhen sculpting on an object with a multiresolution modifier there are 3 sliders to define the level of subdivision.\nWhen duplicating the object right now, it will set the viewport level to the same value as the sculpt level.\n\n- Add a multires modifier\n- Subdidive the object\n- Set the viewport level to soemthing other than the viewport value\n- Duplicate the object\n\n", "Mask brush stops working in tiling sculpting mode after changing its size through \"F\" hotkey\nOperating system: Windows-10-10.0.19043-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2070 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.59\n\nBroken: version: 2.93, 3.3.0\nWorked: -\n\nWhen an object has the Multires modifier, mask brush stops working in tiling sculpting mode after. (See video below)\n[Desktop 2022.09.17 - 21.18.10.07.mp4](Desktop_2022.09.17_-_21.18.10.07.mp4)\n\n- Open attached file or\n - Add a grid object\n - Add a multires modifier\n - Subdivide Simple a few times\n - Select mask brush in sculting mode\n - set tiling on x and y axis to 1 or 2m \n- try paint mask (sometimes the tiling doesn't work at this stage, you need to paint again and undo)\n- change brush size through F hotkey\n- try paint mask (tiling stops working)\n- press undo\n- try paint mask (tiling works again)\n\n[tiling_sculpt_bug.blend](tiling_sculpt_bug.blend)\n\n", "Sculpt mode: Set Pivot to Mask Border doesn't work with the initial mask generated by the Box/Lasso mask tools\nWin 10\n\nBroken: blender-2.81-382e7ce9ecfb-windows64\nWorked: ?\n\nThe Set Pivot to Mask Border function doesn't work with the initial mask generated by the Box/Lasso mask tools.\n\nLaunch Blender\nOpen the Sculpting template (File ➞ New ➞ Sculpting)\nChoose the Box or Lasso mask tool\nMask a part of the mesh\nSelect one of the Transform tools (just to see the pivot)\nRun the Set Pivot to Mask Border (Sculpt menu ➞ Set Pivot ➞ Pivot to Mask Border)\n*And nothing happens* \n\n[2019-10-18_18-09-13.mp4](2019-10-18_18-09-13.mp4)\n\n**Note**, if I \"blur\" the mask **(Mask menu ➞ Smooth Mask)** then the \"Pivot to Mask Border\" will start working.\n\nAlso, masks generated with the regular mask brush don't have this problem...", "Quad remesher produces exotic self-intersecting mesh despite apparently simple source mesh\nOperating system: Ubuntu 20.04.1 LTS\nGraphics card: RTX 2060\n\nBroken: 2.93.1\nWorked:\n\n\nDespite an apparently simple source mesh (manifold and normal conditions are met and mesh is not self-intersecting), *Quad* remesh produces an exotic self-intersecting mesh where an area of faces flow together to form a point that pierces through the resulting mesh.\n\n\n[Meshroom](https:*alicevision.org/) produces meshes with a huge number of vertices that I then clean up in Blender - the final step being to remesh with the*Quad// remesher.\n\nHowever, the results often contain unexpected issues - in particular instead of nice clean quads, there are often points where the quads flow together into a point that then pierces through the mesh.\n\nE.g. here we can see quads flowing together on the right and then piercing through the mesh on the left:\n\n\n\n\nThe corresponding area of the original mesh doesn't seem to have any obvious issues:\n\n\n\nI often encounter similar issues with meshes that I work on - despite having applied serious effort to cleaning up the meshes (using the *Mesh / Clean Up* functionality, a lot of manual cleanup and breaking up any intersecting faces in the original mesh with *Face / Intersect (Knife)* with the *Self Intersect* option).\n\nI've attached a 6MB `.blend` file that contains the cleaned-up original mesh as \"low-poly\" and the remeshed version as \"low-poly-remesh\":\n\n[bust-low-poly-bug-report.blend](bust-low-poly-bug-report.blend)\n\nThe original object looks like this:\n\n\n\nAfter remeshing the problem shown up above can be seen on the right:\n\n\n\nI've also highlighted on the left a weird spike that doesn't appear to correspond to anything in the original mesh but that's probably something for a separate bug report.\n\nThe remeshed object was created by duplicating \"low-poly\", renaming it \"low-poly-remesh\", going to *Object Data Properties*, expanding *Remesh*, selecting *Quad* and setting 40000 as the *Number of Faces* value.\n\nI left all the other settings, i.e. *Use Mesh Symmetry* etc., as they were (but separate experiments with them produced different but still problematic results).", "Some transform modes don't apply absolute grid snaping \nOperating system: Windows 10 \nGraphics card: GTX 1080 ti \n\nBroken:\nBlender 2.8 January 31, 01:48:40 - 8c8979067490 Beta\nBlender 2.79 (It worked)\nBlender 2.77 (It worked)\n\nAbsolut grid snaping dont work with \"Face Inset\" and neither with loop cuts. \n\nTurn on the snaping (increment) -> turn on absolut grid snaping -> spawn a plane -> try to snap while loop cutting or inseting faces.", "Multires Displacement Smear brush + Dyntopo Error\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1050 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 451.67\n\nBroken: version: 2.92.0\n\n[When you want to use the multires displacement smear brush with dynamic topology, blender closes.]\n\n\n1. Open Blender\n2. Enter sculpted mode\n3. Select the Multires Displacement Smear brush\n4. Activate Dynamic Topology dyntopo\n5. Pass the brush through the Bucket\n6. Blender closes\n\n[untitled.blend](untitled.blend)\n\n", "[Sculpt Mode] Remesh - Fix Poles problem.\nOperating system: Windows-10-10.0.18362 64 Bits\nGraphics card: GeForce GTX 650/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 432.00\n\nBroken: version: 2.82 (sub 1)\nBroken: version: 2.81 (sub 16)\n\n\nIf i apply a remesh to an objecto with \"Fix Poles\" activated, this generate extra vertex's, This happen more with extracted objects with \"Mask Extract\" i think, this is very random and i don't have idea if's normal.\n\n\nAdd an sphere >> Apply a remesh to the sphere for more details >> Create a mask for extract >> Extract mask >> to the new object extracted clear the mask >> Apply another remesh with \"Fix Poles\" On >> select the Smooth brush with the strength 1.000 >> Smooth all. \n\nVideo Example 1: [Fix Poles - ON (Problem)](u3xa8YrCcQo)\nVideo Example 2: [Fix Poles - OFF (No problem)](agZ6T4l92lA)\nVideo Example 3: [Test with an sphere](HPw87j2bews)\n\n", "Inconsistent pressure behavior between the size and the strength in the sculpt mode\nOperating system: Win 11, Wacom Intuos Pro\n\nBroken: c1beaea80f97\n\n\n\nThis bug report was requested by @JulienKaspar.t.\n\nThe strength pressure feels totally flat once the size pressure is disabled. This behavior is really inconsistent across brushes. I am using the Clay brush below but most brushes have this issue, so it is not limited to the Clay brush.\n\nHere I am demonstrating it with the Clay brush. The brush feels ok when both the pressure and the strength is enabled, there is enough pressure variety and consistency.\n\n[blender_CXuVDMlgac.mp4](blender_CXuVDMlgac.mp4)\n\nI am disabling the size pressure in the video below and the strength feels like there is hardly any pressure control. It almost feels like it is not even responding to the pen pressure. I am using the exact same pressure as in above video with my Wacom pen.\n\n[blender_sd6OpRGCHX.mp4](blender_sd6OpRGCHX.mp4)\n\n\n\nAdd a dense mesh\n\nEnter into the ssculpt mode\n\nSelect the Clay brush\n\nMake sure the brush utilizes the size and the strength pressure.\n\nUse pressure to draw over the surface.\n\nDisable the size pressure control\n\nObserve how the strength behaves \n\nIncrease the strength beyond 1 to see the issue better.", "Resize a incoming fluid cause fluid to stop flowing in\nOperating system: windows 11\nGraphics card: RTX\n\nBlender 3.6.1\n\n\nGood morning\n\nFirst thank you for the work done on Blender.\n\nI think I found a bug. In fluid simulations when we resize an incoming flow, it often happens that the flow is no longer able to generate a fluid.\nI have seen this with eco-spheres\n\nThanks in advance .\n\nBased on the default startup or an attached .blend file (as simple as possible).\n\n", "Apply Base (Multiresolution Modifier) on mesh with shape keys distorts the mesh\nBroken: version: 2.83 (sub 8)\nWorked: Doesn't Work even in Blender 2.79b official release. I haven't tested versions earlier than that.\n\nPressing Apply base using the multires modifier on a mesh with a stored Basis shape key causes artifacting and mesh corruption. Undo still works though.\n\n1. Delete Default Cube\n2. Add Plane mesh (Shift+A > Mesh > Plane)\n3. Add Basis Shape Key.\n4. Add Multiresolution modifier\n5. Subdivide multiple times. (Either Catmull-Clark or Simple type)\n6. (optional) Apply brush strokes in Sculpting mode. \n7. Press Apply Base in Modifier Properties.\n\nor\n\n1. Open attached file\n2. Press Apply Base in the Modifier Properties.\n\n[Shape key Multires Bug.blend](Shape_key_Multires_Bug.blend)\n\n", "Multiresolution displacement baking do not work properly\nOperating system: Windows-8.1-6.3.9600-SP0 64 Bits\nGraphics card: GeForce GTX 660 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 461.40\n\nBroken: version: 2.93.0 Beta\n\nOriginal on the left produces map, thad displace surface like on the right.\nI suppose, there is wrong gamut/ color conversion or baking ray shoots from Hipoly to lowpoly instead measuring height difference from lowpoly to hipoly. Displacement baking is not equal to normal baking!\n\n{[F10039179](изображение.png) size=full}\n\n[mr bake.blend](mr_bake.blend)\n\n[2021-04-26_15-44-07.mp4](2021-04-26_15-44-07.mp4)" ]

[ "Voxel Remesher eventually creates lots of holes in a base mesh at small voxel size (~0.02m)\nOperating system: Windows-10-10.0.18362 64 Bits\nGraphics card: GeForce GTX 1060 6GB/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 436.48\n\nBroken: version: 2.81 (sub 15)\nWorked: (optional)\n\nVoxel Remesher eventually creates lots of holes at small voxel size in meshes like this while anything slightly more than around 0.015 to 0.02m doesn't provide enough expected detail to indicate if I'm going too small and collapsing the mesh by going below it, so it's like a drastic sudden breakdown at around that voxel size seen in the video provided. I tried scaling the mesh up and down and the issue persists. The normals are fine and the LocRotScale is applied. The problem is not seen with the head but happens often with a human body like this.\n\n[2019-10-18 21-04-51.mp4](2019-10-18_21-04-51.mp4)\n[basemesh bug report.blend](basemesh_bug_report.blend)\n" ]

[ "LibOverride: Sanitize handling of library pointer in resync process.\n`BKE_lib_override_library_create_from_tag` takes an 'owner library' data, which is the library owning the currently resync overrides. However, `lib_override_library_resync` always calls it with nullptr library, and later reassign proper lib pointer...\n\nThis needs to be checked, there is likely better/more sensible and straightforward way to handle this. Also somewhat related to #107847.\n\n", "Mitigate the risk of wrong shape key selection in sculpt and edit mode.\n# The Issue\n\nA major weakness of any layered editing system is that it matters which layer is selected for editing operations, but this selection is not apparent in the actual editing space, because layer effects are by definition combined into a single result for display.\n\nFor instance, unless the single key mode is turned on, the effects of all relative shape keys are summed up into a final shape, which doesn't change no matter which one is selected.\n\nThis means that it is easy to apply editing operations to the wrong selected layer, including a possibility that the undo stack will be exhausted before the mistake is noticed.\n\nAn additional risk factor is that shape influence sliders are located in the same list, and a hurried misclick on a slider can change selection.\n\n# Proposed Solution\n\nThe industry standard way to mitigate this risk is layer locking: layers that are not currently being worked on can be locked by the user, in which case all editing operations that would affect that layer will instead abort with an error.\n\nBlender already supports this with Vertex Groups and weight painting - even though with groups it is more obvious which one is selected, and locking has additional effects related to normalization.\n\nI propose adding this feature for shape keys (implemented in #104463):\n\n\n\nSince it is a standard approach that is used by any advanced image editor, most users would likely be already familiar with the concept. However there are some details in its application that need discussing.\n\n## Scope Issue\n\nThe primary design question is which operations should react to layers being locked, balancing mainly implementation complexity and the effectiveness and convenience of the feature.\n\nIn implementing the prototype I decided that the main criteria should be:\n\n* Operations that affect vertex positions, and thus shape keys.\n\n* Operations where the shape key selection matters:\n\n For example, applying object mode transformation updates all shape keys. Thus the selection doesn't really matter, and respecting locks would inconvenience the user that is working on the scene layout, rather than individual object details.\n\n* Deform-only operations:\n\n Even though selection may affect the final vertex positions, operations that change topology necessarily change all shape keys. For that reason, topology manipulation also belongs in an earlier stage of the design pipeline than shape key creation, even though it may be necessary to do it later sometimes for various reasons. Finally, topology change is seldom not accompanied by using transformation operators that respect the locks.\n\n### Current Scope\n\nThis is the current scope of the checks in the prototype:\n\n* All Edit Mode <u>transformation</u> operations.\n\n Specifically, everything invoking the generic transform system initialization code.\n\n* All <u>Sculpt brushes</u> that affect vertices, and not colors or masks.\n\n Specifically, everything invoking the generic brush stroke initialization code.\n\n* <u>Smooth</u> and <u>Laplacian Smooth</u> operators in Edit Mode.\n* <u>Blend From Shape</u> operator in Edit Mode.\n* <u>Propagate To All Shapes</u> operator in Edit Mode.\n\n This one is a bit of a problem, because technically shape selection doesn't matter here: it snaps all shape keys to the current position. However currently I implemented a check to abort if any shape key is locked on the principle that it is a shape key operation. Maybe it should only check the main Basis, or work in some other way.\n\n* <u>Mirror Shape</u> and deleting shape keys.\n\n* A few other operators...\n\nThis [spreadsheet](edit?usp=sharing) containts an exact listing of which operators should and actually do check for locks.\n\n# Discussion\n\nIt is necessary to discuss whether the main criteria, or this specific list needs changing. For instance, I'm not confident that I remembered all operations that match the criteria for the checks.\n\nAnother minor design point is whether attempting to edit a locked layer should be an error (shows a message under the cursor) or a warning (appears in the status line).\n\nSince the checks are done by operators rather than low-level code, add-ons would have to implement them too where relevant. I wonder how many would be affected, and whether some API additions to help them do that easily would be necessary.\n\nThere have also been some questions raised in #104463, like:\n\n> - Are shape keys a special case? (is there other data we might want to lock).\n> - Is the current implementation disallowing editing in all anticipated situations? There seems to be quite a few ways to transform vertices that aren't included here, applying transform, moving object origins, ... to name a couple, probably other mesh tools too. It may feel like an incomplete feature if these are left out, yet, this means we need to add check in many more places.\n> - Are there ways this can be implemented in a way that doesn't frustrate users. For example, you load someone elses file, select objects and want to apply transform. If they have locked a shepe-key it could mean having to manually select and disable the the locks. Users may request a ways to disable locks on all selected objects, maybe going along in this direction is fine, but we can anticipate some of these issues and consider functionality that doesn't add unnecessary overhead. \n", "macOS todo list\nNOTE: this is a direct dump of Install-OS\n\n- [ ] Users have to manually change macOS defaults:\n * Map middle mouse to regular \"Mouse button 2\"\n * Make the Function keys use modifiers (so Blender sees these)\n * Change the Shift+NumPad options to not use Shift (change spaces)\n * The latter two could become part of a new standard macOS compliant map in the future?\n- [ ] ~~Finder aliases or~~(1f223b9a1f afb1a64ccb) hidden files are not being recognized by Blender\n- [ ] macOS swallows some key events, like e.g. emulated numpad - on Norvegian keyboard (see #38273).\n- [ ] Blender prevents macOS from sleep #38702", "Pie Menu Hotkey Customization\n# Summary\n\nAs a blender user, I am interested in customizing the hotkeys for pie menus. If properties with key press events can be added to a layout, then it follows that operators with key press events should be possible to be add to a layout as well. This applies to pie menus, but also to the items in any layout.\n\n## Current\n\n```python\n# 4 - LEFT\npie.operator(...)\n# 6 - RIGHT\npie.operator(...)\n# 2 - BOTTOM\npie.operator(...)\n# 8 - TOP\npie.operator(...)\n...\n```\n\n\n## Desired\n\n```python\npie.operator(..., key = 'l')\npie.operator(..., key = 'r')\npie.operator(..., key = 't')\npie.operator(..., key = 'd')\n...\n```\n\n\n# Intro\n\nNew to blender source code, been trying to interact on blender discord and blender developer chat for a few days. Trying here now.\n\n# Motivation\n\nIn principle, pie menus are like interactive modal hotkeys, similar to vim. New users can open a pie menu to see what functionality is relevant in a given context, and experienced users can press a key to open the pie and a second key to choose the operator. Modal pies are to modeling what tab completion is to coding.\n\n# Current State\n\nThe numpad keys 2,4,6,8 activate the items in a pie menu, and I'm struggling to figure out how I might be able to customize those hotkeys for each pie menu. If there is a way, I missed it.\n\n# Desired State\n\nEach operator added to a pie layout can be configured to have a specific hotkey.\n\n# So Far\n\nThe [UILayout class](bpy.types.UILayout.html) has a `.operator()` method which could be modified to accept a hotkey to trigger the operator while the pie is active.\n\nThere is also a `.prop()` method which accepts a param `event (boolean, (optional)) – Use button to input key events`. If properties with key press events can be added to a layout, then it should be straightforward that operators with key events can be added to a layout.\n\nI haven't been able to find the source code for the UILayout python class (the UILayout_Fake class inherits UILayout):\n\n, but there is a section of `source/blender/makesrna/internal/rna_ui.cc` that might be defining the class jointly for C++ and python.\n\nI also can't find the `.menu_pie()` implementation anywhere:\n\n\n# To Do\n\n1. Find the UILayout class implementation\n2. Find the menu_pie function implementation\n\n# Relevant Links\n\nHere is an example of typical pie menu code: ui.py\nHere is some discussion of hacking modal hotkeys into blender addons: 2\n\n", "The Tool Settings UI should update when holding modifier keys\nIn Sculpt Mode, users can hold Ctrl to invert the effect of the current tool. However, both the tool settings + and - buttons, as well as the color of the brush cursor, don't update to reflect that you've inverted the effect. \n\nBoth of these should update their state when the user holds Ctrl.\n\n\n\nThe same is true for the selection tools and their modes.\n\n{[F5686801](Screenshot_2018-11-23_at_00.36.36.png), full=true}\n\nWhen holding Shift or Ctrl to add or remove, the state of these buttons should update to reflect that. ", "Animation: create a full separate copy of BoneCollections while in armature edit mode\nThis is to avoid touching the main data structures in the Armature datablock while in edit mode, which interacts badly with undo.\n\nRelevant:\n- #111965\n\n", "Visual glitches With adaptive domain enabled when object is duplicated or transformed\nOperating system: windows 10\nGraphics card: 3080ti\n\n**Blender Version** 3.3 lts\n\nVisual glitches With adaptive domain enabled when object is duplicated or transformed\n\n- Open default scene\n- Add quick smoke on cube\n- enable `adaptive domain`\n- play animation\n- move domain around or duplicate it (snaps back to original place)\n- play animation again (will move to transformed position)", "Splash Screen: \"Updated Preferences\" Infobox\nFor Blender 4.0 there will be various default keymap updates (#105298) that users should be aware of when loading their 3.6 preferences.\nThis task will become necessary once those patches are merged.\n\nThe idea is to add a popup to make users aware and point them towards resources to inform themselves. This will avoid frustration and many invalid bug reports.\n\n### Mockups\n\nThe addition of an info icon to the operator button that loads the previous preferences (As an indication that this will open a popup with more info).\n\n\n\nA possible layout for the main popup.\nThis info might be very important for some users, so the popup can only be closed with an explicit button press.\n\nThis also includes a an operator button \"Read more...\" that links to the relevant release notes page without closing the popup itself. An added border would make this more recognisable as a button.\n\n\n\nThis popup could then be used for future releases and other important changes (only if absolutely needed).\n\n", "Dynamic Overrides Use Case - Giant object-based 'rig' with a few key controlers. \nDynamic Overrides are a good candidate to support extreme relationships nightmare involving thousands of objects, like the one in [Machine.zip](Machine.zip).\n\nLibrary Overrides cannot realistically handle such a massive amount of IDs/relationships, but Dynamic Overrides could only override the relevant properties of the few key controller objects while keeping everything fully linked, in a similar way to what proxies used to do. \n\n------------------------------------------\n\nOriginal report:\n\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 511.65\n\nBroken: version: 3.0.1\nWorked: 2.9\n\nIt's now impossible to use a single control object/bone from a linked rig without creating an override for the entire hierarchy or parent collection. This is completely un-workable for professionals working with rigged assets made up of many hundreds or thousands of parts.\n\nI can't share my scene file as it contains sensitive intellectual property belonging to a client. The issue will be evident for any remotely complex rig however.\n\nIn the project I'm currently working on, I have a top level collection of bone controllers that have no parents. Previously I could create a proxy for each controller and use those to animate the linked asset whilst retaining the full functionality of the rig. Now that create proxy has been removed, I can't just override a single controller. Instead I'm forced to unnecessarily create overrides for several thousand objects! An operation which for this rig takes in excess of 30 minutes to run (if at all)! Then after saving the file (once the operation is complete), Blender outright crashes on each attempt to re-open.\n\nSo although the new override system works in theory. There's no parity between the two systems yet in terms of the ability to override a single object. And previous functionality has been lost without a viable workaround available. Which in my estimation constitutes a bug.\n\nWith this in mind, would you please kindly add the create proxy option back again, until the overrides system is fully functional? I'll personally now have to roll back to 2.9 until this issue is resolved.\n\nThanks", "VSE: Ensure all operators in the Strip menu are working on the selected and not only on active strip (+unselected) \nAs it is confusing and inconsistent to have functions in the Strip menu which only works on properties of the active strip even if it is unselected, this design list is for ensuring all functions in the strip menu to work on selection. \n\n- - [x] Swap Strip: #91191 Fix: [D12441](D12441)\n- - [x] Effect menu. Fix: [D12451](D12451)\n- - [x] Movie(Set Render Size + Deinterlace) and Image(Set Render Size + Separate Images) menu entries should be moved into the source sidebar with +alt for selection. Fix: [D12464](D12464)\n- - [ ] Add Modifier in the menu needs to work on selection and is missing +alt for selected, when executed in the sidebar.\n- - [ ] Inputs menu: Swap data should go into the swap menu and the rest into Strip Source sidebar with +alt for selected. Bug report: #91350\n- - [ ] Toggle Meta and UnMeta should go into the strip sidebar, as they are only working on the active strip. Old patch: #74130\n(A checked checkbox means a fix has been submitted, but not necessarily been committed)\n\n", "Adjust Last Operation: Other changes outside that panel are reverted back automatically\nOperating system: Windows-7-6.1.7601-SP1 64 Bits\nGraphics card: GeForce GTX 660 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 445.87\n\nBroken: version: 2.90.0 Alpha\n\nHere the demo and file\n[2020-06-23_20-30-11.mp4](2020-06-23_20-30-11.mp4)\n[Scene state reset.blend](Scene_state_reset.blend)\n\n1. Add some objects, add last one\n2. Change visibility for other objects via outliner\n3. Change property in adjustment popup\n4. Scene settings resetted to previous", "Design: Shading: Extend Clamp and Map Range nodes to work with Vectors\nI'd like to make these nodes work with vector inputs and outputs.\n\nIt's unclear to me as the the best way of doing this. \n\n\n\nSome options:\n\n - **Separate Nodes** : Create separate versions of these nodes with Vector sockets.\n\n - **Dynamic** : Use a dynamic socket. This is better from a UX perspective but how do we show this? For these two nodes, they are pretty pointless without a value input so the edit control could be hidden. The node then operates accordingly depending on the input link. \n\n\n\n- **Mode Switch** : Add vector input and output sockets and add a mode switch to hide the unwanted input socket. \n\n- **Double Sockets** : Add vector input and output sockets and depending what is connected. For vector only input, `Vector` > `Float & Vector`. For float only input, `Float` > `Float & Vector`. If both inputs are connected do `Float` > `Float` and `Vector` > `Vector`.\n\nFor the combined cases, the node would output and show both Value and Vector.\n\n\n\n\n\n\n\n", "Warnings, Restrictions, & Centralized Administration\nWith more organizations adopting Blender there is increasing need to simplify the deployment, maintenance, and administration of the program in large installations. This plan is meant to be a number of incremental steps that, once approved, I could implement myself. I have about 10 years of prior experience as a Domain Administrator on a network of approximately 600 users and 1000 computers in a Windows domain, and am familiar with the pitfalls and best practices.\n\n**Improved Warnings**\n\nThis doesn't seem related, but is. I would like us to implement something like [D10955: UI: Improved Operator Confirmations](D10955), which allow us to make warning dialogs much more easily than now that are also more informative, more customizable, and more differentiable. This way we can define many more than we have now to cover almost any situation where some users might want warnings, like when overwriting existing files or when files differ in version. We can define more because we'd also add the ability for users to easily enable/disable them.\n\n**Move Warning Configurations from Keymap to Settings**\n\nRight now the configuration of warnings is quite obscure, buried in the keymaps as an attribute of the entry. Most users will not be able to change the current warnings. I'd like to make a \"Warnings\" section in Settings with checkboxes to allow quick enabling and disabling of warnings.\n\n**Expose Factory Settings**\n\nRight now the default startup file and the factory preferences are only stored inside the executable. I'd like to optionally expose those for customization. Perhaps add a command-line argument to the executable that extracts copies of the embedded \"default.blend\" and \"prefs.blend\" to the install folder. These would be used instead, if present, in the same way that you can now override the splash image with a custom \"splash.png\".\n\nIn a large deployment you would put the blender installation folder in a shared read-only location that everyone would launch. That way you only have to install it once, not on every workstation. With this change you could, for example, change the fonts directory to be a shared network location. Change that in this one place and it becomes the default for everyone.\n\n**Add Restriction Settings**\n\nLike the \"Warnings\" section in Settings, I'd like to add \"Restrictions\". So checkboxes that, once enabled, restrict the user from doing some operations. For example to not allow the installation of addons. In a single-user installation this would not be used. But imagine an administrator setting these, along with warnings, and placing a \"prefs.blend\" file in the shared read-only network location of the Blender installation. It now means everyone on the network has the same defaults, warnings, and restrictions.\n\n**Project Settings**\n\nAs another layer of granularity, allow putting these shared defaults and preference files within the folder path of blend files. So while loading any \"blend\" file we also look in its folder and parent folders for defaults and prefs. This way you can have per-project settings.\n\n**Platform Enhancements**\n\nDuring startup we are loading settings for settings, warnings, and restrictions. Once this is done we could also call platform-specific routines if they exist, like checking settings in the Windows Registry for example. This is all that is required to support Group Policy, so an Administrator can set a path for all user, allow some groups to add addons, and only leaders to change restrictions, etc.\n\n", "Show progress-bar & cancel button while (batch-)generating previews\nThere's currently no way to skip or cancel preview generation.\n\nThis is especially a problem when batch generating a bunch of previews because \"Mark as Asset\" was executed on multiple data-blocks.", "No Vector Displacement On Instances\nWindows 10 x64\n\n2.79.4\n\nVector displacement doesnt work with instances, It mirrors the displacement on the first object, Im sure this is working as intended but there are many cases where you need to displace actual instances as well.\n\nHaving a checkbox somewhere to enable instance displacement would be amazing\n\n\n\n\n\n[instances.blend](instances.blend)\n" ]

[ "For ID blocks, move + and x buttons into popover\nWith our ID blocks in the top right of the window, the little + and x icons make the Blender UI overly busy and complex. There's also the problem that, at least on Windows and Linux, the close button is in the top right corner of the display, creating this clumsy configuration of two X's. \n\n{[F5673488](Screenshot_2018-11-22_at_01.50.52.png), size=full}\n\nWe could change these ID blocks to work as popovers instead, creating a much simpler top level interface. When opened, we can display the controls to add, remove or rename items inside the popover, similar to what we already do for the presets:\n\n\n\nThis would result in a cleaner top level UI, and avoid the double X on some OS's:\n\n{[F6622971](Screenshot_2019-02-16_at_01.19.50.png), size=full}" ]

[ "Sample texture color dialog isn't dismissed after use\nOperating system: W10\nGraphics card: GTX 970\n\n2.82a\n\nAfter use of the pipette in the Image Texture Editor, the 'F9' dialog box is not dismissed until an UNDO operation is performed.\n\n\nSample file included; Launch file. Switch to Texture Paint. In the Texture Image Editor, invoke the sample function with the S hotkey. Observe the dialog.[Sample Dialog Dismiss.blend](Sample_Dialog_Dismiss.blend)", "ColorRamp, UI problem to hold the Color Stop\nOperating system: macOS-13.1-x86_64-i386-64bit 64 Bits\nGraphics card: Intel(R) Iris(TM) Plus Graphics 655 Intel Inc. 4.1 INTEL-20.2.48\n\nBroken: version: 3.4.1\n\nIn the ColorRamp node, it is difficult to grab a Color Stop by its selection rectangle.\n\nAdd a ColorRamp node in the Shader Editor or in Geometry Node and select a Color Stop rectangle to make it move.\nOnly a small part of the rectangle (red frame on my screenshot) can be grabbed when it should be the whole rectangle (green frame on my screenshot).\nIt is disturbing to regularly move the node by mistake instead of moving the colored rectangle.\n\nI tested several versions with the same result, today's master, 3.3, 2.83, 2.8 and 2.79.\n\n[ColorRamp_UI_problem.blend](ColorRamp_UI_problem.blend)\n\n\n- - - \n\nI can see 2 other problems in the ColorRamp UI, please tell me if I need to create a second report, but maybe it's more convenient to come only once in this area with a report rather than twice.\nThere is a line of transparent (or unfilled) pixels and 2 pixels that overflow to the left of the triangle, see my screenshot with the details in red.\n\n\nNOTE: there is a separate report for this, see #111017\n\nOlivier\n<3 Blender", "User Interface Policies\nDesign task to coordinate user interface policies before including them in the [human interface guide-lines (HIG) ](Human_Interface_Guidelines).\n\nThis is a curated list of items agreed on by module owners.\n\nOrder of importance:\n\n- {icon circle color=red} **Very Important** - *These we should handle before the next release.*\n- {icon circle color=yellow} **Somewhat Important** - *These issues would be nice to do as soon as possible.*\n- {icon circle color=green} **Less Important** - *Extra polish, nice to have.*\n- `?` **Incomplete** - *Tasks needing more details before implementing.*\n\n----\n\n### Under Discussion\n\n- {icon circle color=green} #76212 (Avoid Dynamic Menu Contents)\n- {icon circle color=green} #76213 (Avoid Dynamic Menu Text)\n- {icon circle color=green} #76227 (Make All Operators Accessible from Menus)\n- {icon circle color=green} #76261 (Tooltip Guidelines)\n\n----\n\n### Accepted\n\n- <none>\n\n### Rejected\n\n- [D7515: UI: Comma as Decimal Separator](D7515)", "Allow user to change the shading parameters of the asset preview generation scene\nUsers should be able to have some control over the render settings of generated asset preview icons.\n\nPersonally, I want to be able to turn off the specular lighting because it makes my assets look ugly in the previews (note the big shiny surfaces on these bushes).\n\n", "Move Tool Todo\n## Move\n * - [ ] {icon circle color=green} Currently, distances are not well communicated while dragging, only the axis. We can make it much clearer how far you've moved your item, like so:\n\nHere, the distance moved is brighter than the rest of the axis line.\n * - [x] {icon circle color=yellow} The Redo panel for moving items should be simplified. The Constraint Axis controls should be hidden, and the Move X, Y, Z controls should always work. The current UI here is annoying to use and takes up lots of space.\n\n * - [x] {icon circle color=yellow} The tool has 3 names: 'Grab' (in the toolbar), 'Move' (inside the Redo panel), and 'Translate' (Redo panel title). It should have one name. 'Move' is the most universally understood word, that everyone will understand. The other two are ambiguous and unclear. ", "Collection holdout on grease pencil objects\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 750/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.71\n\nBroken: version: 2.93.4\nWorked: unknown\n\nHoldout option on collection doesn't work, if objects it is containing are grease pencil objects. Alpha of contained GP objects remains on 1 / solid in viewport as well as in render.\n\nPS: This behavior would help a lot with compositing linework made with \"linework modifier\" of grease pencil. In shots where characters are obscured by objects linework should obscure them as well but if in the same shot character comes in front of the object suddenly the opposite is needed.\n\nsample .blend file (steps included):\n[collection holdout grease pencil bug.blend](collection_holdout_grease_pencil_bug.blend) \n1. create GP drawing (GP Suzanne will work fine)\n2. create collection for it and move it in the colection\n3. repeate first two steps but move the new GP object in front of the first one (Default GP stroke might be mor suitable in this case)\n4. in collection settings of the collection made for second object (the stroke) check the holdout option.\n5. render image and you should see the second gp object remaining opaque.\n\n**given result**\n\n\n**expected result**\n[achieved by converting GP stroke to curve and extruding the curve]\n\n\n", "Grease Pencil: Radius property misses blue slider background to visualize the percentage between soft min and max\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1660 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.81\n\nBroken: 2.80 -> 2.92.0 Alpha\nWorked: I believe it's been like this since initial GPencil implementation\n\nIn the GPencil editor the slider number button for \"Radius\" located in the topbar does not draw a percentage fill as is done in other editors such as Vertex Paint, Sculpt, Texture Paint and Weight Paint. \n\nStart blender with default settings\nAdd->Grease Pencil ->Blank\nSelect the Grease Pencil Object if not already selected\nSwitch from Object mode to Draw mode\nNote that the \"Radius\" topbar slider does not contain a fill color as does the \"Strength\" topbar slider \n\n[GreasePencilRadiusSliders.mov](GreasePencilRadiusSliders.mov)\n\nI've created [D9948](D9948) as a potential fix", "Redo panel disappears during edits in Tweak tool\nOperating system: Win10 x64\nGraphics card: Nvidia 1070\n\n**Blender Version** \nblender-2.80.0-git.26d5a3625ed-windows64\n\nThe info panel for move disappears in edit mode if tweak mode is in action. This does not seem to happen if one uses \"g\" for grab instead.\n\n\n- Enable left click select, optionally select all toggles\n\n- In your space bar menu turn the select mode (W) on\n\n- Create a mesh and move into edit mode \n\n- Start grab and drag of verts and watch the move info panel at the left bottom\n\n- Now try the same thing with selecting a vertex then press \"g\" to move.\n\n", "Dot dash modifier not working with time offset modifier (grease pencil)\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA GeForce MX330/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 471.41\n\nBroken: version: 3.1.0 Alpha\nWorked: not sure\n\nWhen time offset modifier is in the modifier stack before or after dot dash, dot dash has no effects{[F12118899](dot-dash-time-monkey.blend)}\n\nSee attached file\n\n", "Motion Pasths: \"Clear Motion Paths\" default to \"only selected\"\nProblem: \nAs with \"calculating\" motion paths you expect the clear motion paths function to default to the one selected, In the same way you create Keyframe the selected objects. The default deletes all motion paths in scene and wanting to clear one motion path is more common than wanting to clear them all.\n\nSolution:\nChanging the default of clear motion paths to \"clear motion path\" and turn \"Only Selected\" to default.", "Grease Pencil: Add eyedropper to palette popup\nInspired by this [idea]() by gakutada.\n\n## The Issue\nThe color picker popup in Grease Pencil is missing the eyedropper button:\n\n\nThat is present in regular color pickers:\n\n\nEven though holding the `E` key with the mouse over the color swatch works fine:\n<video src=\"attachment\" title=\"gpencil_palette_colorpicker.mp4\" controls></video>\n\n## The Solution\n\nAdd the operator to the popup. Aligned to the right-side so it's similar to the regular color popup.\n\n\n", "Grease Pencil ignores collection Line Art usage settings\nversion: 3.3.0 Beta, branch: master, commit date: 2022-07-28 22:39, hash: 6ca602dd9fa1, type: release\nbuild date: 2022-07-29, 00:21:12\nplatform: 'Windows-10-10.0.22000-SP0'\n\nrenderer:\t'NVIDIA GeForce RTX 3090/PCIe/SSE2'\nvendor:\t'NVIDIA Corporation'\nversion:\t'4.5.0 NVIDIA 516.59'\n\nGrese Pencil ingnores Collections Line Art usage settings. See my attachment and screenshot.\n\nOpen the attached file and check the line art settings in the torus collection.\n\n\n\n[gpencil_outlines-test-3_3.blend](gpencil_outlines-test-3_3.blend)", "Sculpt mode -> Curve Stroke causes Undo to behave differently\nOperating system: Windows 10 x64\nGraphics card: Nvidia 980 GTX\n\nBroken: v 2.82 Alpha, 10-12-2019\n\nIf Brush stroke is set to Curve and some sculpt detail is added (Ctrl + Left Click), (Ctrl + Z) or Undo does not work. Instead of Undo for the sculpt detail, the curve control points are removed.\n\n1. Start with default cube and add Multi-Res modifier, subdivide a couple of times.\n2. Go to sculpt mode and with the SculptDraw brush set the stroke to Curve.\n3. Draw a curve using (Ctrl + Right Click) to add some control points.\n4. Add sculpt details using Ctrl + Left Click. \n5. Press Ctrl + Z. Instead of removing the details the control points of the curve will be removed first. \n6. Press Ctrl + Z repeatedly. When the last control point remains then the sculpt detail will be removed.", "Hair particles: 'Use modifier Stack' option removes all hair if particles have been added or removed in particle editmode\nOperating system: Linux-3.10.0-957.10.1.el7.x86_64-x86_64-with-centos-7.6.1810-Core 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 418.56\n\nBroken: version: 2.81 (sub 4)\n\nUse modifier stack checkbox deletes all the hair particles if they were created with the add hair brush and combed (not necessary), even if there are no modifiers before the particle system. This also happens if particles have been deleted (X) in particle editmode.\nAnd there is no way to get those hairs back. It can really destroy all your work on grooming.\nBy the way, that check box becomes inactive in 2.79 after grooming, but it's active for some reason.\n\nCreate a hair particle system.\nGo into particle edit mode.\nDelete all the hairs (by the way, you won't be able to do that until you comb it, may be a second bug).\nAdd some hairs via add hair brush (you can add hair without deleting on the previous step, the bug will occur anyway).\nGo back to object mode.\nCheck the use modifier stack checkbox.\nAll the hairs disappear!\n\nThe attached file has all that done except the very last step.\n\n[hair_use_modifier_stack_bug.blend](hair_use_modifier_stack_bug.blend)", "Paint Mode: Erasing\nNOTE: This task is currently to document the requirements for erasing workflows for painting. This should be the basis for discussions on the implementation.\n\n# Goal\n\nFor an intuitive painting workflow it's necessary to have an eraser accessible at all times.\nThis has been a shortcoming in Blender for painting, since the erasing of alpha channels has only been available as a blending mode for brushes.\nThis is very hard to access via the UI and shortcuts.\n\n# Recommendations\n\nThe ideal solution is to have a very simple brush tool in the toolbar. Aside from the typical texture, stroke, falloff and cursor settings, it can only erase alpha.\n\nThis tool can then have the shortcut `E` assigned by default to make it very accessible.\n\n## Pure Emissive Colors\n\nThe motivation behind not offering a blending mode to \"Add Alpha\" in the Erase Tool is because this has been notoriously difficult to do like the user would expect.\nSee #86868 & #83342.\n\nWhen painting float or pre-multiplied images, the alpha is treated differently. Any color with an alpha 0 would have the RGB colors displayed as pure emissive colors and still visible.\nThe only way to erase the colors completely is to also erase the RGB channels to 0.\n\nAs a result the design for the brush should emphasise this: **The Erase brush is a destructive tool!**\nIf you erase the alpha directly via painting tools you should expect to likely lose the full RGBA color information.\nThe alpha 0 areas can be painted over with a new color afterwards. \n\nFor non destructive erasing of alpha values it is recommended to use a separate image as a mask. This can be done right now via nodes.\nThis non-destructive approach might not be very accessible right now but that can be addressed separately (Especially with the [layered textures design]())\n\n## Erasing Selections\n\nWe need to discuss how we want to handle other erasing operations like via Gesture Tools (Box/Lasso/Line Erase), or for example the Color Filter for gradual erasing. These generally affect the unmasked/selected areas.\nThere are different ways to enable this workflow.\n\n**Erase Operator**\n\nThe simplest way would be to add an operator to the menus and mapped to a shortcut (Like `Delete`) to erase the colors on any selected areas.\nThe user would in that case use the existing Gesture Tools and brushes to mask/select areas to define the areas to be deleted.\nThis approach would follow the Select -> Operate paradigm of Blender closest.\n\n**Erase Tools**\n\nThe existing Color Filter tool could receive an additional Filter Type to 'Erase'. This would allow gradual erasing in the case where it is needed.\nAdding Gesture Tools specifically or erasing is also possible but might just bloat the Toolbar.\n\n## Erase Mode\n\nAn alternative approach is to make the \"Erase Alpha\" blending mode in the regular painting tools more accessible.\nIn other painting application such as Krita every tools and brush has an \"Erase Mode\" button that can be easily toggled with a shortcut and applies globally to all tools and brushes.\nWe could implement this design as an alternative to the Erase Tools/Brush for users that prefer this workflow.\n\nIn that case the Erase Alpha blending mode would be a button in the header that can be toggled directly, instead of being part of the blending modes popup.\nA shortcut could also be assigned.\nThis toggle should be available for for all brushes and tools that affect colors. " ]

[ "Removing item from quick favorites has no effect until menu is closed\nOperating system: Linux-5.4.31-1-MANJARO-x86_64-with-arch-Manjaro-Linux 64 Bits\nGraphics card: Mesa DRI Intel(R) HD Graphics 3000 (SNB GT2) Intel Open Source Technology Center 3.3 (Core Profile) Mesa 20.0.4\n\nBroken: version: 2.83 (sub 13)\nWorked: /\n\nWhen trying to remove an item from the *Quick Favorites* menu, the item stays there as if nothing has happened, until the menu is closed & opened again. Right-click offers an option to *Add to Quick Favorites*; left-click executes the item as normal:\n[quickfav.webm](quickfav.webm)\nI found this behavior rather confusing, and would instead expect the item to disappear immediately after clicking *Remove*.\n\n1. Add item to *Quick Favorites*\n2. Open *Quick Favorites* - [x], right-click the item > *Remove from Quick Favorites.* Keep the cursor on the menu, to keep it from closing automatically." ]

[ "Text Object - Edit Mode - Special Characters - Circle and Promille doesn't get created\nOperating system and graphics card\n\nNivida 760 GTX, Win 7, 64 Bit, german\n\nBroken: (example: 2.69.7 4b206af, see splash screen)\nWorked: (optional)\n\nBlender 2.79\n\n\nText Object - Edit Mode - Special Characters, the Special characters Circle and Promille doesn't get created when i choose them. \n\nBased on a (as simple as possible) attached .blend file with minimum amount of steps\n\nCreate a new scene. Create a text object. Insert special character promille, insert special character circle.\n\nThey should create. But here nothing happens then. There is no warning. The Info editor tells me the characters are created. But the characters are simply missing in the text.\n\nThe list in the Info editor seems also to have some flaws at displaying what character gets created. Two characters shows an empty string in the brackets. One shows the common square for a not found ascii character.\n\nMaybe this is because of my german Windows 7 64 Bit Home , with american keyboard layout removed.\n\nExample images and a file is attached.\n\n\n[missingcharacters.blend](missingcharacters.blend)\n\n\n\n\n", "Custom Manipulators: General Design & Functionality\nThere are a number of features that should be supported but further design work is needed before they can be implemented properly. Everyone is welcome to join the discussion by proposing more ideas or solutions, by creating mockups or by just giving some own input.\n\n----\n\n# Scope of Widgets\n\n### Description\n\nWhen should a tool use a widget?\n\nExamples:\n\n* Edit Mesh Spin Tool\n* Edit Mesh Smooth tool (number of steps, smooth factor)\n* Edit Mesh Bevel/Inset\n* Edit Mesh UV Rotate\n* Edit Mesh Subdivide\n\nSpin tool is an example where widgets make a lot of sense (you rotate around an axis, so direct manipulation is useful).\n\nThe issue with Smooth & UV-Rotate is that using a 3D manipulator for these is using manipulator simply as a way to access buttons in the view-port.\n\nAn issue with bevel/inset depends on the kind of manipulator used. Having a single manipulator in the view doesn't directly relate to the edge, and again - its more a replacement for accessing buttons in the tool-bar.\n\n### Proposed Solutions\n\nSuggest not to use manipulators if they are simply an alternate access to buttons.\n\nAs for bevel/inset: we could make the bevel edges themselves into manipulators, so you could click on beveled edge and drag it.\n\n----\n\n# Toggling Visibility of Widgets\n\n### Description\n\nCurrently, only the widgets of the active object are drawn. While this works for most cases, being able to precisely control which widgets are visible and which not can be useful.\nFor example, in future, a camera could have widgets for the focal length, sensor size, depth of field distance, ... If all are visible all the time it's easy to accidentally manipulate them.\n\n### Proposed Solutions\n\n* Draw a panel for the active object in which its widgets can be enabled/disabled individually (similar to \"Display\" for cameras).\n* Add a toggle button next to every button represented by a widget (similar to how it's done in the transform panel in the properties region, where you can press the 'lock' icon to disable the widget for this axis)\n\n### Open Topics\n\n* Where would the manipulator toggles be stored? (Object / Viewport / Tool-Settings)\n* How to expose settings that aren't related directly to objects? (Group Center, View-Port Clipping Bounds)\n\n----\n\n# Numerical Input\n\n### Description\n\nIt should be possible for users to input precise values to manipulate a widget (and thus the value it represents). This way they can keep their locus of attention in the viewport, avoiding having to search for a button in the UI.\n\n### Proposed Solutions\n\n* Enabling number input by holding a modifier key while clicking on widget (as in - Ctrl+LMB on widget activates modal mode to input value)\n\n---\n\n# Snap System\n\nUsers may want to snap manipulators to other elements in the scene, eg:\n\n* Snapping the spin center-point to a vertex on a mesh.\n* Snapping bisect angle to a vertex\n* Snapping the camera's DOF to the surface of an object.\n\nNote that currently holding Ctrl sets `WM_MANIPULATOR_TWEAK_SNAP` but each manipulator needs to implement this.\n\n### Proposed Solutions\n\n* Each manipulator has a snap flag for 'TRANSLATE | ROTATE | SCALE', when set, holding Ctrl will snap to any of the elements set.\n* Snapping can also respect scene settings (in the 3D) view-port, so users can snap as with mesh edit-mode.\n\n----\n\n# Key Framing\n\nWhen the cursor is over a manipulator, do we want to support I-Key to keyframe that setting (DOF, lense for eg).\n\n----\n\n# Theme Colors\n\nManipulators now use theme colors, there are basic preset colors to use: (HIGHLIGHT, PRIMARY, SECONDARY, A, B).\n\nWhere A/B are for other controls.\n\n*Currently they're hard coded in manipulators but they can be set however we like.*", "Non-keyboard shortcuts allowed to be set via context menu, but cannot be unset without going to user-preferences\nBroken: 2.79 or latest 2.80 (fa59346c13).\n\nAlthough I can easily (and often accidentally) assign non-keyboard shortcuts to menu items, I cannot easily undo it.\n\nFrom the viewport: View > Frame All [RMB] > Change Shortcut.\n\nAlthough the UI says \"Press a Key\", if you mouse click it (e.g., LBM) will register as the new shortcut.\nNow if you try to change the shortcut again it says it is a Non-Keyboard shortcut and cannot be changed.\n\nTooltip: *\"Only keyboard shortcuts can be changed that way (...)\"*", "Keymap system refactor\nListing issues with the current key-map system.\n\nKey-Map System\n--------------\n\n\n- Timers should be removed from the key-map, since they are for internal operations.\n- Action-zones should be removed from key-maps.\n\n\n\nUser Interface\n--------------\n\n- The ability to re-order key-map items.\n\n```\nThis is important as key-map items are handled in the order listed, with some items passing on to key-map items defined afterwards. \n```\n\n- Ability to select from existing operators list instead of typing in a string (similar to selecting ID's bones... etc).\n\n- Other features:\n\n - Remove duplicate key-map items.\n - Show invalid key-map items (missing operators, reference to invalid menu items... etc).\n - Disallow assigning invalid/duplicate key-map items when using \"Assign Shortcut\".\n\n\nOpen Topics\n-----------\n\nThese are larger issues which need design work, which this proposal doesn't (yet) handle, and it's not clear exactly how these would be handled.\n\n- Exporting key-maps looses the ability to access key-map preferences.\n- Mapping keys from keyboards which have keys not found on a US/English keyboard.\n- Workflow for editing and maintaining custom key-maps could be looked into & improved.", "Cut previs of Loop Cut visible after changing to Knife with a hotkey \nversion: 2.81 (sub 16), branch: master, commit date: 2019-12-04 11:32, hash: f1aa4d18d49d, type: Release\nbuild date: 2019-12-04, 14:30:40\nplatform: Windows\n\nWhen I change the tool from ***Loop Cut***to***Knife***with a hotkey, I can still see the previs of the Loop Cut.\n\nISv0j8o.png\n\n\n1. Click Loop Cut tool.\n2. Hover over the mesh to see the yellow previs\n3. Press K on the keyboard to change the tool to Knife", "Edge Slide via the GG shortcut can show incorrect selected checkboxes\n%%%--- Operating System, Graphics card ---\nWindows 8 x64, GTX560SE\n\n - Blender version with error, and version that worked ---\n2.67 r56888\n\n - Short description of error ---\nEdge Slide via the G-G shortcut can show incorrect selected checkboxes in Tools Shelf: eNwW8Kj.gif\n\n - Steps for others to reproduce the error (preferably based on attached .blend file) ---\nOpen attached .blend file. With an edge selected, use \"Edge Slide\" button in Tool Shelf. Click on checkbox \"Correct UVs\". Select another edge and now use Edge Slide via the G-G shortcut - tool behaves as if \"Correct UVs\" selected, but it's showing unchecked in Tools Shelf.%%%", "Edit Dyntopo Detail Size operator is not working with all methods.\nOperating system: Ubuntu 18.04\nGraphics card:\n\nBroken: master #e7dc3c91ad82\nWorked: 2.90\n\nEdit Dyntopo Detail Size operator was recently replaced to have a better gizmo.\nThat only works with Constant Detailing method.\n\nShift D is doing nothing when Relative or Brush detailing method are used.\nThat is a regression. It is pertinent to change the size for those methods, too.\n\nProbably, gizmo should work differently for a Brush Method. But, at least, shortcut should allow to change Detail Size like in 2.90. \n\n\n\nIssue in Master :\nEnter into Sculpt Mode.\nEnable Dyntopo mode.\nPress Shift D. Nothing happens.\nChange Detail Method from Relative to Constant. \nPress Shift D. New Gizmo is available.\nChange Detail Method from Constant to Brush.\nPress Shift D. Nothing happens. \n\nRepeat same steps In 2.90 :\nCircle gizmo is adapted to used method.\nRelative method is relative to screen. So, circle does not go above a certain limit.\nConstant/Manual method is relative to World Space. Size of Circle is not limited.\nBrush Method is relative brush size. Gizmo is showing a percentage.\n", "GPencil: Add option to jump to scene from Video Sequencer Strip\nNeed a way to jump from the strip to the scene edition directly.\n\nIt's necessary to define in some place what it's the layout to use.", "Improvement ideas for using Driver Context Properties to access scene custom properties\nThe #105132 pull request introduces access to properties of the current scene and view layer via an implicit rather explicit reference.\n\nThe existing implementation is perfectly fine for built-in properties, but I have some additional ideas aimed at making it easier and more convenient to allow scene-global custom settings via custom properties.\n\nBasically, since the properties have to be added by the user, it would be convenient to support the case when it doesn't exist. It also may be nice to allow setting overrides at several different levels of abstraction.\n\nThus:\n\n* Allow an optional default value that would be returned by the variable if the referenced property doesn't exist, instead of failing with an error.\n\n* Allow fallback from Layer to Scene, and then World: one World can be used by multiple Scenes, which can have multiple Layers.\n\nSince this all only makes sense for *custom* properties, maybe it can be a new mode specifically aimed at them, e.g. some kind of \"Layer/Scene/World Custom Property\", which has options for the custom property name (without the `[\"...\"]` quoting), channel index (for vectors), and default value.\n\nThese are things missing for the new driver things to fully cover the features of the View Layer mode of the Attribute material node.", "Move To Line Begin and Reset View operators in the console editor conflict when pressing Home key\nOperating system: Windows-10-10.0.19042-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1650 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 512.95\n\nBroken: version: 3.4.0 Alpha\n\n\nIn the python console editor, both the Move To Line Begin and the Reset View operators have the same shortcut \"Home\", so it is not possible to invoke the Move To Line Begin operator.", "\"Input Pending\" for Loop Cut and Slide when called from menu.\nOperating system: Windows 10 64 bit\nGraphics card: Nvidia GTX 1660Ti\n\nBroken: 3.3.2\nWorked: 2.93\n\n**Caused by:** da2ba402689d7ed752b0b37ad49dce1a5851cc77\n\n**Reproducing the error:**\n\n- Ctrl+R on default cube edit mode, regardless of selection mode, loop cut slide works as usual.\n- Use search menu or header->edge menu and select loop cut slide, header bar only shows \"Input Pending\", if pressed left mouse, then operator runs correctly. If pressed right mouse, then operator is cancelled.\n\n\n\n", "Float slider in the redo panel of the \"add object\" operator does not hide the mouse cursor\nBroken: 2.79b, 2.83\nWorked: (None)\n\nThe problem is not seen in editing mode, but in object mode when adding an object\nand changing the properties in the redo panel, the mouse cursor does not disappear.\n\n\n", "The Shortcut keys of Play Aniamtion not work in Spreatsheet Editor\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce RTX 2060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.71\n\nBroken: version: 3.0.0 Beta\n\nThe Shortcut keys of Play Aniamtion({key Space}) not work in Spreatsheet Editor\n\n[play animation bug.mp4](play_animation_bug.mp4)\n\n", "UV Editor lacks Annotation 'Placement' option\nBroken: 2.80rc1\nWorked: 2.79b\n\nIn Blender 2.79, you had the 'View' and 'Cursor' Stroke Placement option in the left panel, under 'Grease Pencil'.\nIn 2.8 this is meant to be in the Top Bar or 'Active Tool and Workspace Settings', as it is for the 3d Viewport's Annotate tool, but this is missing for the UV Editor, meaning all annotation defaults to 'View'.\n\n - Open the blender and choose the `UV Editing` workspace\n - Compare the 3d View Annotate Tool with that of the UV Editor\n - Notice the lack of Placement options", "\"Open Custom Preview Scene\" operator for asset previews\n(Better name ideas welcome.)\n\n**Motivation**\nWhile generally quite useful, the automatic preview isn't always working too well. That's why there is the option to load a custom preview image from a file. But that tends to be cumbersome too: You typically have to open a new Blender file, import the asset, set up the scene (camera, lighting, render settings, etc), render, save the render result to disk and back in the original file, load the rendered image into the asset preview.\n\n**Idea**\nIt would be nice to have a button, say *Open Custom Preview Scene* that opens a new Blender instance with a scene set up for object preview rendering (e.g. the preview.blend, or a specific collection from it). After saving and closing it the original Blender instance could render the preview using it.\nThis is similar to the *Edit Asset* button Sybren implemented recently, which also opens a new Blender instance and reloads the asset library after closing it.\n\n**Possible Issues**\n\nYou may want to set up different scenes for different assets, but you don't want the scene to get lost every time you close the instance. So maybe the .blend(s) with the edited preview setup should be stored somehow (they are small).\nWe could store it in `/tmp/` by default and let users choose a different location in {nav Preferences > File Paths}. For each custom preview scene we store a .blend with a name like `[source_file_name]__[object_name].blend`. \nIf the file or object name changes now, that mapping will be broken. But there could be ways to restore that (in worst case, manually updating the preview file name).\n\n\nNote that I think this should be done in a script, possibly even in an add-on.\n\nPlus, I'd suggest showing a warning when you save something in the temp directory, indicating that the data might get lost." ]

[ "Keymap presets are not saved.\nLinux 64bit, apparently Windows\n\nBroken: 2.72b\n\n\nChanges in custom keymap presets are not saved with \"Save User Settings\".\n\n\n1. Run Blender and open User Preferences, Input section.\n2. Create a new keymap using the \"+\" button.\n3. Make a change (for example, disable the 3D View -> Object Mode -> Duplicate Objects hotkey).\n4. Save User Settings and quit Blender.\n5. Run Blender again and open User Preferences, Input section: the new keymap itself is present, but the changes to hotkeys are not.\n\nAt a glance, the changes to hotkeys are not written into the keymap file in user dir/scripts/presets/keyconfig.\n", "Maya ShortCuts issue : Select a bone in weight paint mode\nWin 7 64 bit\n\nBlender 2.72\n\nIn order to select a bone during weight paint mode, the \"maya\" shortcut is SHIFT+Click on the bone.\nBut in order for this to work, I have to first unchecked the :\n*File/user preferences/input/3D view/Weight Paint/WeightPaintSampleGroup*\n(Because it uses the same short cut)\n\nAnd then, I have to add the following shortcut (like for blender shortcut):\n*- File/user preferences/input/3D view/3D view (global)/Add new => view3d.select + case shift + Click left + center + object*\n\nBecause this initial shortcut does'nt seem to work :\n*File/user preferences/input/3D view/3D view (global)/Select or Deselect All (Shift Select Mouse)*\n\nIs it possible to correct this issue ?\n\nThanks in advance !!!\n\nSeb" ]

[ "New curves object type\n**Status:** Under development by the geometry nodes team. Accessible though the \"Empty Hair\" primitive. Grooming workflow solid, other areas still WIP\n\n**Overview**\n\n- [ ] New curves object data type\n - [x] Add new datablock and object type\n - [x] Custom data storage, custom data masks\n - [ ] Default curve radius and shaping parameters\n - [x] Solve problem with not enough `FILTER_ID` bits\n- [x] Geometry nodes integration (#95355)\n - [x] Convert existing nodes to new data structure (#95443)\n - [x] New building block nodes\n - [x] Geometry nodes simulation\n - [x] Node group assets for higher level nodes\n- [x] Mesh attachment (to deform curves along with mesh surface) (#95776)\n - [x] Curves datablock has pointer to surface mesh object\n - [x] Add polygon index and UV / barycentric coordinates attributes to indicate where curve is attached\n - [x] Node to deform curves based on surface (#96676)\n - [x] Subdivision surface attachment\n- [ ] Full render engine support for curves data-block (#96455)\n- [ ] Attribute support\n - [x] Generic attributes\n - [ ] Generated texture coordinates\n - [x] Batch cache generation\n- [ ] Python API\n - [x] Positions and radius\n - [x] Attributes\n - [ ] Add/remove curves and points\n- [ ] Viewport drawing and selection\n - [x] Selection engine integration (#95933)\n - [ ] Display options (display random % of curves, ?)\n - [ ] Move \"strand/strip\" display option to curves data-block, rename\n- [x] Initial sculpt mode\n - [x] Brushes (#96259)\n - [x] Sculpting deformed curves (crazyspace)\n- [ ] Edit mode\n - [ ] Node group operators (#101778)\n- [ ] Dependency graph copy-on-write integration\n - [x] Refactor mesh runtime storage to be reusable by other objects types\n - [ ] Automatic instancing for objects without modifiers, like meshes\n - [ ] Copy-on-write for attribute arrays (#95845)\n- [x] Conversion of old particle curves to new curves (#96675)\n- [ ] I/O\n - [ ] Alembic import ([D11592](D11592))\n - [ ] Alembic export\n - [ ] USD import ([D11592](D11592))\n - [x] USD export (#102376)\n\n---\n\n**Relevant links:**\n* [Nodes Object Types ](Object_Types)\n* #78515 (Hair object - Node types design)\n* #78606 (Hair object - design proposal)\n\n---\n\n### Links\n\n* Unreal's [Alembic groom specification](index.html).\n* Pixars [PxHair schema ](Siggraph2019_Pipeline%20Case%20Studies.pdf) for storing rest pose, guides and scalp", "Allow menus to be added referenced to another menu \n## Motivation\nAddon developers cannot add menus on specified place. \nThey are always added at the end (append), or at begin (prepend)\n\n## Analysis\nAfter a look at current code, i think a good approach should be allow to define a reference label on append function.\n\nWhen drawing the menu layout, a new RNA call should be done on a new function class, that should then execute the referenced drawing calls on py scope.\n\nIssue with this development is that for any but menu drawn two python calls are performed to allow addons draw before and after itself. Maybe this could be solved checking in C scope if it has post/pre functions added.\n\n## Development steps\n* {icon check-square-o color=green} Draw after label\n* {icon check-square-o color=green} Draw before label\n* {icon check-square-o color=green} Give warning if not used (label not found)\n\n## Know Issues\n* Issues with enums, if shown as submenu it's not working. Is shown on self menu, the label is the operator name, and considered a block\n\nFeedback is welcome!\n\n\n\n\n", "Design discussion: Add object instancing to support transforming linked objects?\nThis is an initial task to start discussion on this topic. More details would have to be looked into for an actual design.\n\n### Motivation\n\nLinking objects in Blender has the inherent design limitation that it also links the object's transform, which means you can't transform an object that was linked. This is a very basic and important operation. With library overrides, that issue is mitigated; however IMO we should look at this as a feature for production use-cases, not as something for more informal, low-overhead usages. This would be a complex solution for the user to manage for a simple problem. They need rather deep understanding of how Blender data-management works to manage library overrides.\n\nEspecially for the Asset Browser, it would be good to have a simple solution that behaves as users expect. Apparently what they do expect is that they can simply link objects, while being able to transform them still. Not being able to edit other object properties seems fine.\n\n### Possible Solutions\n\nFor collections we have actually addressed this issue by supporting linked collection instances. That means, while the collection and its contents are linked, the collection is placed relative to an emtpy that is local to this file. So by transforming the empty the entire collection is transformed. The idea of instancing can be brought to objects so transforming linked objects becomes possible in effect.\n\nSome ideas on how object instancing could be supported on linking:\n* **Object is made local, the mesh, materials, etc stay linked** Simple to implement - downside is that object properties, modifiers, constraints, physics and the like would not be linked anymore.\n* **Instancing via empty.** This would be very similar to collections. An empty can instance an object which means it acts as a parent of the object.\n* **Wrap into collection instance.** Effectively the same as instancing via an empty, but the user would see an intermediate collection. It's simple to add this, but kind of forces an implementation detail onto users.\n\n### Other Approaches\n\nInstead of trying to support object instancing, other things could be done on linking:\n* **Automatically create library overrides for transforms**. Another simple solution technically and at least saves the user from having to know about or setup the library overrides when linking. But obviously pushes users towards using library overrides, which they sooner or later will still have to deal with.", "LibOverride - Update - Fix handling of temp ID\nLiboverride 'update' process (the part applying overrides again over a copy of the linked data, `BKE_lib_override_library_main_update` & co) needs some specific control over its usage of `BKE_id_copy_ex` to create the temp copy of the linked ID: It needs to prevent any potential modification of any other ID in Main.\n\nThis happens e.g. with regular copy of Collections, since the `parents` lists of their children need to be updated with a new pointer to the new copy. There may be a few other similar cases, this needs to be investigated.\n\nThis default behavior prevents from safely using `BKE_id_free_ex` on that same temp ID later on, even if the temp ID is not explicitly linked to any other ID.\n\nWhether this change requires a new specific tag, or could be part of one of the 'no main' cases, remains to be investigated as well. \n\n--------------------\n\n## Original Report\n\n\nOperating system: Linux-5.15.0-10056-tuxedo-x86_64-with-glibc2.35 64 Bits\nGraphics card: NVIDIA GeForce RTX 3060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 525.60.11\n\nBroken: version: 3.5.0 Alpha, branch: master, commit date: Unknown, hash: f06a1368\nWorked: -\n\nBlender crashes when appending a collection (twice) from a file where this collection has been linked and overridden.\n\n\n*Note: Using a debug build with ASAN makes this reliable to reproduce.*\n\n 1. Open file `fileB.blend`.\n 2. Go to `File`>`Append`. Browse into `fileA.blend`>`Collections`. Untick the `Localize All` option. Choose `Asset`.\n 3. Blender crashes.\n\n[report_crash_link.zip](report_crash_link.zip)\n\n**ASAN report**\n```\n=================================================================\n==84954==ERROR: AddressSanitizer: heap-use-after-free on address 0x613000267ef2 at pc 0x000002b6fe58 bp 0x7ffc27cb7e40 sp 0x7ffc27cb7e30\nREAD of size 2 at 0x613000267ef2 thread T0\n - 0 0x2b6fe57 in collection_foreach_id /.../blender/source/blender/blenkernel/intern/collection.c:164\n - 1 0xd2a149 in library_foreach_ID_link /.../blender/source/blender/blenkernel/intern/lib_query.c:334\n - 2 0xd2a317 in BKE_library_foreach_ID_link /.../blender/source/blender/blenkernel/intern/lib_query.c:353\n - 3 0xcce9b3 in BKE_lib_id_clear_library_data /.../blender/source/blender/blenkernel/intern/lib_id.c:224\n - 4 0xcd1d22 in BKE_lib_id_make_local_generic /.../blender/source/blender/blenkernel/intern/lib_id.c:501\n - 5 0xcd2cef in BKE_lib_id_make_local /.../blender/source/blender/blenkernel/intern/lib_id.c:561\n - 6 0xbe3fa23 in BKE_blendfile_append /.../blender/source/blender/blenkernel/intern/blendfile_link_append.c:1111\n - 7 0x3a49cf1 in wm_link_append_exec /.../blender/source/blender/windowmanager/intern/wm_files_link.c:356\n - 8 0x39fe3ed in wm_handler_fileselect_do /.../blender/source/blender/windowmanager/intern/wm_event_system.cc:2724\n - 9 0x39ffd56 in wm_handler_fileselect_call /.../blender/source/blender/windowmanager/intern/wm_event_system.cc:2821\n - 10 0x3a04962 in wm_handlers_do_intern /.../blender/source/blender/windowmanager/intern/wm_event_system.cc:3315\n - 11 0x3a04e0b in wm_handlers_do /.../blender/source/blender/windowmanager/intern/wm_event_system.cc:3369\n - 12 0x3a0e0ed in wm_event_do_handlers /.../blender/source/blender/windowmanager/intern/wm_event_system.cc:3963\n - 13 0x39c81bd in WM_main /.../blender/source/blender/windowmanager/intern/wm.c:640\n - 14 0x801b57 in main /.../blender/source/creator/creator.c:585\n - 15 0x7fdac84b0d8f in __libc_start_call_main ../sysdeps/nptl/libc_start_call_main.h:58\n - 16 0x7fdac84b0e3f in __libc_start_main_impl ../csu/libc-start.c:392\n - 17 0x800e34 in _start (/.../build_linux_debug/bin/blender+0x800e34)\n\n0x613000267ef2 is located 114 bytes inside of 344-byte region [0x613000267e80,0x613000267fd8)\nfreed by thread T0 here:\n - 0 0x7fdac8bf0517 in __interceptor_free ../../../../src/libsanitizer/asan/asan_malloc_linux.cpp:127\n - 1 0x1dd70304 in MEM_lockfree_freeN /.../blender/intern/guardedalloc/intern/mallocn_lockfree_impl.c:117\n - 2 0xce415b in BKE_id_free_ex /.../blender/source/blender/blenkernel/intern/lib_id_delete.c:168\n - 3 0xd254b5 in BKE_lib_override_library_update /.../blender/source/blender/blenkernel/intern/lib_override.cc:3793\n - 4 0xd26297 in BKE_lib_override_library_main_update /.../blender/source/blender/blenkernel/intern/lib_override.cc:3840\n - 5 0x3b96f2b in library_link_end /.../blender/source/blender/blenloader/intern/readfile.cc:4526\n - 6 0x3b972ab in BLO_library_link_end /.../blender/source/blender/blenloader/intern/readfile.cc:4545\n - 7 0xbe42109 in BKE_blendfile_link /.../blender/source/blender/blenkernel/intern/blendfile_link_append.c:1283\n - 8 0x3a49c39 in wm_link_append_exec /.../blender/source/blender/windowmanager/intern/wm_files_link.c:346\n - 9 0x39fe3ed in wm_handler_fileselect_do /.../blender/source/blender/windowmanager/intern/wm_event_system.cc:2724\n - 10 0x39ffd56 in wm_handler_fileselect_call /.../blender/source/blender/windowmanager/intern/wm_event_system.cc:2821\n - 11 0x3a04962 in wm_handlers_do_intern /.../blender/source/blender/windowmanager/intern/wm_event_system.cc:3315\n - 12 0x3a04e0b in wm_handlers_do /.../blender/source/blender/windowmanager/intern/wm_event_system.cc:3369\n - 13 0x3a0e0ed in wm_event_do_handlers /.../blender/source/blender/windowmanager/intern/wm_event_system.cc:3963\n - 14 0x39c81bd in WM_main /.../blender/source/blender/windowmanager/intern/wm.c:640\n - 15 0x801b57 in main /.../blender/source/creator/creator.c:585\n #16 0x7fdac84b0d8f in __libc_start_call_main ../sysdeps/nptl/libc_start_call_main.h:58\n\npreviously allocated by thread T0 here:\n - 0 0x7fdac8bf0a37 in __interceptor_calloc ../../../../src/libsanitizer/asan/asan_malloc_linux.cpp:154\n - 1 0x1dd70a49 in MEM_lockfree_callocN /.../blender/intern/guardedalloc/intern/mallocn_lockfree_impl.c:223\n - 2 0xcd81aa in BKE_libblock_alloc_notest /.../blender/source/blender/blenkernel/intern/lib_id.c:1101\n - 3 0xcd8341 in BKE_libblock_alloc /.../blender/source/blender/blenkernel/intern/lib_id.c:1113\n - 4 0xcd9ac6 in BKE_libblock_copy_ex /.../blender/source/blender/blenkernel/intern/lib_id.c:1274\n - 5 0xcd388a in BKE_id_copy_ex /.../blender/source/blender/blenkernel/intern/lib_id.c:637\n - 6 0xd241f3 in BKE_lib_override_library_update /.../blender/source/blender/blenkernel/intern/lib_override.cc:3725\n - 7 0xd26297 in BKE_lib_override_library_main_update /.../blender/source/blender/blenkernel/intern/lib_override.cc:3840\n - 8 0x3b96f2b in library_link_end /.../blender/source/blender/blenloader/intern/readfile.cc:4526\n - 9 0x3b972ab in BLO_library_link_end /.../blender/source/blender/blenloader/intern/readfile.cc:4545\n - 10 0xbe42109 in BKE_blendfile_link /.../blender/source/blender/blenkernel/intern/blendfile_link_append.c:1283\n - 11 0x3a49c39 in wm_link_append_exec /.../blender/source/blender/windowmanager/intern/wm_files_link.c:346\n - 12 0x39fe3ed in wm_handler_fileselect_do /.../blender/source/blender/windowmanager/intern/wm_event_system.cc:2724\n - 13 0x39ffd56 in wm_handler_fileselect_call /.../blender/source/blender/windowmanager/intern/wm_event_system.cc:2821\n - 14 0x3a04962 in wm_handlers_do_intern /.../blender/source/blender/windowmanager/intern/wm_event_system.cc:3315\n - 15 0x3a04e0b in wm_handlers_do /.../blender/source/blender/windowmanager/intern/wm_event_system.cc:3369\n - 16 0x3a0e0ed in wm_event_do_handlers /.../blender/source/blender/windowmanager/intern/wm_event_system.cc:3963\n - 17 0x39c81bd in WM_main /.../blender/source/blender/windowmanager/intern/wm.c:640\n - 18 0x801b57 in main /.../blender/source/creator/creator.c:585\n - 19 0x7fdac84b0d8f in __libc_start_call_main ../sysdeps/nptl/libc_start_call_main.h:58\n\nSUMMARY: AddressSanitizer: heap-use-after-free /.../blender/source/blender/blenkernel/intern/collection.c:164 in collection_foreach_id\nShadow bytes around the buggy address:\n 0x0c2680044f80: 00 00 00 00 00 00 00 00 00 00 00 fa fa fa fa fa\n 0x0c2680044f90: fa fa fa fa fa fa fa fa 00 00 00 00 00 00 00 00\n 0x0c2680044fa0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00\n 0x0c2680044fb0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00\n 0x0c2680044fc0: 00 00 00 fa fa fa fa fa fa fa fa fa fa fa fa fa\n=>0x0c2680044fd0: fd fd fd fd fd fd fd fd fd fd fd fd fd fd[fd]fd\n 0x0c2680044fe0: fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd\n 0x0c2680044ff0: fd fd fd fd fd fd fd fd fd fd fd fa fa fa fa fa\n 0x0c2680045000: fa fa fa fa fa fa fa fa fd fd fd fd fd fd fd fd\n 0x0c2680045010: fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd\n 0x0c2680045020: fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd\nShadow byte legend (one shadow byte represents 8 application bytes):\n Addressable: 00\n Partially addressable: 01 02 03 04 05 06 07 \n Heap left redzone: fa\n Freed heap region: fd\n Stack left redzone: f1\n Stack mid redzone: f2\n Stack right redzone: f3\n Stack after return: f5\n Stack use after scope: f8\n Global redzone: f9\n Global init order: f6\n Poisoned by user: f7\n Container overflow: fc\n Array cookie: ac\n Intra object redzone: bb\n ASan internal: fe\n Left alloca redzone: ca\n Right alloca redzone: cb\n Shadow gap: cc\n==84954==ABORTING\n```\n", "Grid in rendered orthogonal views does not adapt to displacement\nOperating system: Darwin-19.6.0-x86_64-i386-64bit 64 Bits\nGraphics card: AMD Radeon Pro 5500M OpenGL Engine ATI Technologies Inc. 4.1 ATI-3.10.16\n\nBroken: version: 2.92.0 Alpha\nWorked: unknown\n\nWhen in rendered view, in an orthogonal view, the grid does not adapt to displacement, giving some strange looking results.\n\nAdd displacement to an object. Go to rendered view, orthogonal. Grid does not adapt to the new object shape, see screenshot.\n\n", "Node Tools: popover menus for options\nWhen creating a new Node Tool the defaults are immediatelly set. But users can still change them from the menus\n\n* **Object Types**: Mesh, ...\n* **Modes**: Object, Edit, Sculpt\n\n\n\nNote: This is only for the Tool context. But once assets are also supported in the Modifiers Add Menu we will need the `Object Types` popover in the Modifier context as well.\n\n", "Library Overrides: Making an override (for a linked object in the scene collection) creates a duplicate object\nOperating system: Linux-6.1.18-200.fc37.x86_64-x86_64-with-glibc2.36 64 Bits, WAYLAND UI\nGraphics card: AMD Radeon Graphics (renoir, LLVM 15.0.7, DRM 3.49, 6.1.18-200.fc37.x86_64) AMD 4.6 (Core Profile) Mesa 22.3.7\n\nBroken: version: 4.0.0 Alpha\nWorked: --\n\nLibrary Overrides: Making an override (for a linked object in the scene collection) creates a duplicate object\n\n- open `main.blend`\n- (there is a linked object in the scene collection)\n- `Object` > `Library Override` > `Make`\n- A separate (overridden) Cube object is created in the scene collection (along with the linked object still there)\n\nNOTE: doing the same from the Outliner (RMB > `Library Override` > `Make` > `Selected`) seems to work as expected\n\nas a comparison:\n- open `main_with collection.blend`\n- (there is a linked object in a collection [not the scene collection])\n- `Object` > `Library Override` > `Make`\n- The overriden Cube object is just there without the linked object Cube being present\n\n", "Python API : OBJECT_OT_convert can't be fully abstracted from context with override\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1660 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 472.39\n\nBroken: version: 2.93.5\nIt didn't work in latest Beta build either.\nWorked: I don't think it ever worked otherwise ?\n\nOverriding the context of an operator with a dictionary enables us to call them from pretty much anywhere in the code without having to make sure the context is correct (selected objects, active object, called from a particular editor, etc.)\n\nHowever the `OBJECT_OT_convert` called with `bpy.ops.object.convert` operator still requires an active object to be selected to function with an override, which defeats the purpose of using an override.\n\nIt is explained quite thoroughly in this BSE Q&A 86891\n\nScript to reproduce in an empty file :\n\n```\nimport bpy\n\nbpy.ops.curve.primitive_bezier_curve_add()\n\n# Commenting these 2 lines removes the RuntimeError (context incorrect) on last line\nbpy.context.active_object.select_set(False)\nbpy.context.view_layer.objects.active = None\n\nbpy.ops.object.convert({\"selected_objects\": bpy.data.objects}, target=\"MESH\")\n```\n\n", "Overlay-Next\nUpdate Overlay engine to leverage the new capabilities of the new draw manager.\n\nRequires #102177 (Draw: Selection Next)\n\n#### Complex:\n- [ ] Armature\n- [ ] Edit Mesh\n- [ ] Edit Curve\n- [ ] Edit Uv\n- [ ] Gpencil\n- [ ] Selection Outlines\n- [ ] Wireframe\n\n#### Medium:\n- [ ] Line Anti-Aliasing\n- [ ] Particles\n- [ ] Sculpt\n- [ ] Volume\n\n#### Easy:\n- [ ] Attribute Viewer\n- [x] Background\n- [ ] Bounds\n- [ ] Curves\n- [ ] Camera\n- [ ] Edit Text\n- [x] Empties\n- [ ] Facing\n- [ ] Force Fields\n- [x] Grid\n- [ ] Lattice\n- [ ] Lights\n- [ ] Light Probes\n- [ ] Relation Lines\n- [ ] Speaker", "Non animatable properties\nBroken: version: 3.3.0 Alpha\n\nWorking on refreshing the in-built AnimAll addon by adding the latest properties supported in Blender.\n@ideasman42 asked me to file a task\n\nCertain properties are missing animation capabilities:\n\n - Lattice Vertex Groups\n - Mesh Vertex Creases", "GPencil: Add multiframe support to Move to Layer\n", "Replace object-linked materials feature\nMaterials linked to objects have some issues:\n- They cause problems for geometry nodes, which can arbitrarily change material slots during evaluation.\n- They tend to be confusing, partially due to relying on consistent indices between two different lists\n- They cause large performance issues for reference counting in the main data-base, since geometry-level operations also have to be aware of objects\n- Object-linking is theoretically redundant now that materials can be changed during evaluation.\n\nBecause of that, it would be best to remove the feature and replace it with something more generic.\nMost workflows are already possible with nodes like the \"Replace Material\" node, but we have to be able to replicate the behavior exactly in order to version the feature away. There are a few options:\n1. Add a modifier with a UI-list interface that sets materials at specific slots.\n2. Add the necessary features to geometry nodes (loop? lists?) necessary to make the behavior completely generic\n3. Just add a geometry node group with enough nodes to handle whatever list of object linked materials the object had.\n", "Custom Nodes customizable Header for the Node Editor\nHi\n\nThis is a follow up from my closed design task:\n 104066\n\nAs @brecht suggested the best way will be to create a dedicated function that every addon can use to customize the header. \n\nI'm going to investigate if it can be donein `release/scripts/modules/bpy_types.py`.\n\nI hope I can get it done this time. \n\nCheers, \nDavid\n\n", "Graph Editor: Stacked mode for F-curves\nDesign proposal for “Stacked mode” for F-curves in Graph Editor:\n“Stacked mode” is the mode for Graph Editor where F-curves are placed one below the other\n\nSome advantages are described on Rightclickselect - 4jbbbc/ \n\nI am not UI designer so examples may look ugly but I hope the idea is clear enough. \n\n\nPROS:\n1. This mode allow you to show and manage your F-curves more effectively in some cases\nFrom this  to this \n2. It allows you to see the “flow of energy” in your F-curves in a more comfortable way\nHere is a vertical movement on Z axis \"flows\" into a squash\n\nAs you can see there are some mess with F-curves in default mode \n\n\n3. It allows you to control you F-curves more precisely e.g. If you move handles of different curves at once when these F-curves are at the same place\n\n\nHOW IT SHOULD WORKS:\n- This mode is not default\n\n- I suppose menu -View- in Graph Editor is the best place for it, where you could toggle it on\\off\n\n- Also toggle button next to the normalize button is a good idea\n\n- Stacked mode replace the non-stacked view when it toggled on.\n\nWorkflow and behavior are almost the same as in the non-stacked view. I see only two difference. \n1. F-curves in stacked mode must be normalized only.\n2. when you zoom in\\out e.g. by mousewheel it must work on vertical axis only in the channel at where your cursor is. And when you press e.g. ctrl+mousewheel all channels should zoom in\\out on horizontal axis at once.\n\n- When you press Home button all your F-curves at once should be fitted in the graph editor area\n- If there are too many channels in stacked mode you supposed to scroll them up\\down by e.g alt+mousewheel or by dragging scroll bar.\n\nShould be select channels, go stacked mode, those channels appear, unselect one of them it disappears from the stack, the stack rearranges itself to fit the graph editor space, you select an other channel (add) and it will again refresh to accommodate the channels into the stacked view. You deselect a channel in the stacked mode and it should disappear from the stack.\n\n \n- If none of the channels are selected and you toggle stacked mode on it should show nothing (same as none of objects or bones are selected in the non-stacked view). First of all you have to select channels that you need in non-stacked view mode (both for bones and objects) then toggle stacked mode on.\n\n\n- You should be able to rearrange channels in the stack manually (e.g. by mouse drag) \n- You should be able to change vertical size of a channel string (field? place?) by dragging border of the channel up or down (to give more room for the channel you want). For example Maya does not allow you to change it (this features and the previous one), but Pixar’s Presto does.\n\n\n\n\n", "Curves edit mode Bezier handle drawing\n- [ ] Draw points at handle positions for Bezier curves\n\t- [ ] Use handle selection attributes for point highlight\n- [ ] Draw Bezier handle segments\n\t- [ ] Use different colors depending on handle type\n- [ ] Support `View3DOverlay.display_handle` property\n\nRelated: #105038, #96455" ]

[ "Dynamic Overrides\nWIP\n\n------------------\n\nThis task will be used as general design, and umbrella task for the Dynamic Overrides project.\n\n## General Idea\n\nDynamic Overrides (or dynoverrides for short) aim at providing a simple and efficient way to override property values of any data-block, linked or local.\n\nWhile generally based on the same idea as Library Overrides (aka liboverrides), dynoverrides are a fairly different approach. In a nutshell, they will be simpler to maintain and manage, but way less powerful and not suitable for complex things like a whole character animation.\n\nHere is a list of the key differences:\n\n| Library Overrides | Dynamic Overrides |\n| ----------------- | ----------------- |\n| Work only on linked data | Work on any data, linked or local |\n| Allows modifying ID relationships | No ID relationships changes |\n| Allows adding complex entities like Modifiers or Constraints | Can only modify simple ‘value’ properties |\n| Are handled on file save/load | Are handled as part of evaluation |\n| Are stored in the ID they affect | Are stored in a different ID |\n| Directly linkable | Linkable only if the ID owning them is linked |\n\nUse cases include:\n* Tweaking some shaders or lamp settings in the lighting stage of a production file.\n* Tweaking the position of a few props in the shot at the animation stage of a production file. *(Note that this would imply a way to export dynoverrides through ID linking too.)*\n* Dealing with giant hierarchies of IDs (see e.g. #96092 ).\n\nConceptually, dynoverrides are also quite similar to animation data, the main difference being that they are stored in a different, not directly related data-block than the one they affect.\n\nOne big change/challenge implied by this project is that linked data become (partially) editable. This will require a lot of changes all around the Editor codebase (to properties drawing code, operators etc.). Support for that will need to be added gradually, so a restrictive list of overridable properties is most likely needed (similar to what is done with liboverridable properties already).\n\n## UI/UX\n\nThe main, basic interaction would happen at properties level, similar to how animation of a property is handled:\n\n- Define a property as dynoverridden (*and where - active scene or active viewlayer).\n- From there on, any editing of the property affects the dynoverride value, and not the actual RNA/DNA value.\n- From there on, creation of an animation would animate the value of the dynoverride.\n\n*Later some similar feature to the `I` shortcut (inserting groups of animation fcurves) could exists for similar properties for dynoverrides.*\n\nA list of dynoverrides (grouped by IDs) is shown in Scene and ViewLayer properties, with additional bulk-editing features. If a dynoverride defined at scene level is superseded by a same one defined in the active ViewLayer, it should be visible (grayed out in the Scene list?).\n\nThere could also be an overview of dynoverrides in the Outliner?\n\nSome questions:\n* Re-use 'teal blue' color of liboverridden properties, or have yet another color for dynoverridden properties?\n* Need a dedicated icon (probably reusing part of the liboverride one for consistency?).\n\n## Technical Design\n> NOTE: Using existing animation system here also needs to be investigated; dynoverrides could simply be animation (fcurves) defined at scene and/or view layers level.\n\n### Data Model\n\nDynamic Overrides can be stored in Scene data-blocks, at two levels:\n\n* A global level, applying the dynoverrides to all view layers in the scene.\n* A per-view layer level, replacing the global one when defined.\n\nEach of these contain a list of dynoverrides, together with some runtime data (mappings) to search efficiently whether a property is dynoverridden or not.\n\nA Dynamic Override contains a pointer to the data-block it affects, an RNA path leading to the affected property in that data-block, and the value of the override.\n\nPotential animation of these dynoverride values: could be stored either as part of the Scene animation data (*simplest solution, not sure about the draw-backs yet*), or into their own animation data at the scene and view layer levels? Latest could make it easier to somehow 'export'/link dynoverrides into other files?\n\n### Evaluation\n\nDynamic Overrides will be evaluated as part of the depsgraph evaluation, after the handling of animation. This implies that they:\n\n* Can be animated, but…\n* Will supersede any animations or drivers defined on the properties they affect.\n\n*How exactly this fits in current and planned design for the Depsgraph still needs to be evaluated.*\n\n## Ideas for Later\n\n### Linking of Dynamic Overrides\n\nFinding out how critical it is to be able to link or export/import those dynoverrides, and how this could be done, needs to be further investigated. This would probably not be part of the initial deliverables though.\n\nAn idea could be e.g. to 'compile' the current set (from current scene and view layer) of dynamic overrides applying to a given ID, and store it with that ID. Then on linking that ID, code could re-use that 'compiled overrides' data and apply it into the active scene, with a specific status flag indicating that they are 'from linked data' (and therefore not editable, not deletable, and not saved in blend file, but rather re-created from linked data)?\n\n### Overrides From User PoV\n\nFind out how much Library Overrides and Dynamic Overrides could be presented as a single override interface to the user, at least at a high level of usability.\n\n### Dynamic Overrides as Variants\n\nOne of the ideas from original design back from 2.80 area was that it would be possible to define different overrides for different users of a same ID, and that the depsgraph would detect this and generate different evaluated IDs then.\n\nE.g. two different objects using the same material could define their own dynoverrides of the material color, and evaluated objects would get different evaluated materials IDs assigned then." ]

[ "One of four edited hair systems won't simulate unless edit is deleted\nOperating system: Win10\nGraphics card: GTC 1060\n\nBroken: **2.92.0 master 2021-02-24 16:25 02948a2cab44** and **2.93.0, master 2021-06-02 11:21 84da05a8b806**\nWorked: Didn't find a working version\n\n4 particle systems, one (\"Back\") won't sim unless the edit is deleted. It stays in place and stretches towards the animated head cap.\n\n**Attempted things to fix it:**\nDelete all bakes and resim - failed\nChecked the weight painting - doesn't seem to be the source of the issue\nAppend into new files and resim - failed\nDisabled dynamics - now everything moves with the haircap, but obviously no sim\nRemoved hair edit - now it works but I loose the edit\n\nOpen the file and press space to run the sim and animation. I would not know how to make a sim from scratch which won't work though so I'm sending the file.[hair_bug.blend](hair_bug.blend)", "Enhance ways of accessing active modifier for poll functions of operators\nCurrently, poll function of operators can only get the modifier they are supposed to work on from the context. This is often missing (when called from shortcuts e.g., also with drag-n-drop of modifier panels...).\n\nThis is a sever problem for liboverrides, as some modifiers should then be edited (the locally added ones), and others not.\n\nI can see two ways to solve that: \n \n- Add current 'active' modifier to context (not sure how, or even if, this is always possible).\n- Work around the issue by adding some kind of second 'polling' helper function to be used from invoke/exec functions (since when not found in context, those use event to get the modifier to work on, but events are not passed to poll functions).\n", "Tweak tool deselects previously selected items.\nOperating system: Windows\nGraphics card: 1060\n\nBroken: All 2.8 versions\n\nTweak tool deselects previously selected items.\n\n\n", "Cycles: Bump node doesn't generate anything in certain conditions\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3090/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 522.25\n\nBroken: version: 3.3.0\nWorked: all versions from 2.79 to 3.4.0 alpha are affected, there is no working build\n\nI'm trying to generate a normal map using Bump node and color output of the env map as height map, it doesn't work on Cycles while it works on Eevee. Seems like the \"reflection\" output of Texture coordinate node is broken, but it works fine when I'm using another outputs though\n\nOpen the project, then click on viewport Rendered mode button - it'll start Cycles which for some reason can't generate bump map, then switch to Material preview mode - it'll start eevee which works as expected\n[bug_report.blend](bug_report.blend)\n\nEdit: typos and bad english grammar fixed, I believe", "Grease Pencil Arrange Stroke operator does not function with Curve Editing\nOperating system: Linux 64-bit\nGraphics card: GTX 1080 ti\n\n2.93.5 and 3.0.0 alpha\n\nIn edit mode, \"Arrange Stroke\" operator does not work when use_curve_edit is enabled.\n\n[GP_arrange_bug.blend](GP_arrange_bug.blend)\n\n - Open attached file\n - Use any of the Arrange Stroke operators on the selected stroke\n # Disable Curve Editing and try again to see that it otherwise works as intended\n\n", "Blender 2.90.1 crashes after rendering the first few frames of animation.\n**System Information** \nOperating system: windows 10 64 bit\nGraphics card: Nvidia GeForce GTX 1650\n\nBroken: (2.90.1 3e85bb34d0d7; master)\nWorked(didn't): (only tried on 2.8.3 and did not work at all)\n\nWhen I try to render the animation, after about 50 or so frames(varies randomly), blender crashes with no alerts or anything that tells me what the error is, it simply closes.\nI've rendered another project, and that did indeed render, so I'm left confused why this short animation would cause a crash.\n\nEdit: I disabled motion blur, and it rendered fine. However, I would really like to know why blender won't render if motion blur is enabled. The renderer is Eevee and while my laptop isn't a super computer, it should be able to handle it. Thank you all in advance!\n\n- Open attached file\n- render the animation\n[cube-loop-tut-V1.0.blend](cube-loop-tut-V1.0.blend)", "Armature modifier slows down any transforms on an object\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 466.27\n\nBroken: version: 2.92.0\n\nHaving the armature modifer on an object with a slightly high polycount slows down any transformations (move, scale, rotation) done to it in the viewport.\nEven with the modifier disabled in view still slows down its movement.\nThis slowdown is very impactful to the smooth playback framerate of projects.\n\nHere's a video of the steps (I started out in the default scene and deleted all the starting objects)\n0Bgx_1JEouc\n\n- Make a high poly mesh (Eg. Suzanne subdivided 5 times)\n- Add Modifier: Armature\n- Transform the high poly object (Mode, rotate ...) and note the delay in fps", "Crash on applying boolean modifier\nOperating system: Linux Fedora V35\nGraphics card: Nvidia 3070TI\n\nBlender 3.1.2 and Blender 3.2.0 both crash\n\nI created a mesh object, then used boolean modifier to create a hollow center. I applied the modifier. When I go into edit to clean it up, it crashes.\n\napply the boolean modifier to the mesh \"Bottom\" with object \"Plug\" in the included blend file using all defaults. Enter edit mode and see the mesh is badly corrupted.\n\n[part7.blend](part7.blend)\n[printReady1.blend](printReady1.blend)", "Displacement distorts textures\nOperating system: Windows 10\nGraphics card: RTX 3060Ti \n\nBroken: 3.6.0\nWorked: Never\n\nAll textures are distorted where displacement texture is greater than 0.\n\nWhen you plugin displacement as 'bump only' the box projected textures glitch out and start to stretch.\nSee attached file.\n", "Microsoft store: Blender cannot be used in terminal\nWindows 11\n\nBroken: 3.4\nWorked: 3.3\n\n**Short Description**\nEver since blender 3.4 released, I have been unable to use blender in the command line. Running just the command \"blender\" gives error 0xc0000022 and whenever I try to run any actual command like \"blender -b untitled.blend -a\" nothing happens. I have never had any issues like this in versions prior to 3.4. I am running the Microsoft store version.\n\nIssue only happens with blender from Microsoft store\n\n**Steps to Reproduce**\n * Assuming that blender is already in your system path, which it should be by default,\n * open a terminal in absolutely any folder;\n * run absolutely any blender command;\n * for example, run \"blender -E help\"\n\n- - -\n!!For now, workaround is to type \"blender-launcher\" instead of \"blender\" to launch blender from command line (but that won't accept any arguments)!!", "Crash while baking the Line Art modifier\nOperating system: Windows 11\nGraphics card: RTX 2070\n\nBroken: 7c2dc5183d03\n\n\n\nI get regular crashes when baking Line Art modifier in relatively complicated geometries. \n\nI can't provide my project file. I am hoping that the crash log will give insight into this bug.\n\n", "Open Blender\nHello\n\nI can't open the app, i dowload all versión windows, windows store and portable.\n\nsxstrace.exe\n\nI have windows 10, but i can open the app,\n\nI hope you can help me\nThanks\n\n", "Blender 3.5 crash on render with Intel 4600 integrated GPU\nOperating system: WIndows 10 x64\nGraphics card: Intel 4600\n\nBroken: 3.5 \nWorked: 3.4.1\n\n\nBlender crashes when attempting to render in Evee. Cycles seems to work.\n\n", "Optix fails when geometry nodes instance has animated translation and scale 0 and motion blur is on\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 531.18\n\nBroken: version: 3.6.0 Alpha\nAlso broken in 3.4.1\n\nRendering will not start. It stops at \"Building OptiX acceleration structure\"\n\n\n\nThe culprit seems to be a combination of animated Translation and Scale 0 together with animated Rotation of the object and motion blur.\n\n\n\nDoesn't work in OptiX. Cuda and CPU both work.\n\nOpen the .blend file and render the frame.\n\n", "Most sculpt tools do not work while model is animated\nLinux 64-bit\n2.93.2\nMost sculpt tools do not work as expected when the object is spinning like it's on a pottery wheel. Some tools like the clay tool work okay, but other tools like smooth and scrape do not. I think it has to do with the stroke method; the stroke accounts for screen space but not the local transform of the object if it's moving during a stroke. Video of my attempt here:\n[pottery.mp4](pottery.mp4)\nI made this as a joke and it looks like garbage, but unfortunately it went viral and everybody thinks it's a great idea so now more people are gonna try it and complain that it doesn't work. It's a valid workflow that you have to account for now apparently. I'm really sorry.\n - Open the attached file\n - Play the animation\n # Try to sculpt the thing while it spins\n[bowl.blend](bowl.blend)\n" ]

[ "smooth modifier don't work on vertex group after a geo nodes\nOperating system: windows 10\nGraphics card: rtx 3080ti\n\n**Blender Version** it works fine with 3.1,but don't work in later versions\nBroken: (example: 2.80, edbf15d3c044, master, 2018-11-28, as found on the splash screen)\n\ni add some vertex groups to use as selection in some opetions in geo nodes, afterwords.i add a smooth modifier to smooth specific area with vertex groups,but it doesn't work. \nBased on the default startup or an attached .blend file (as simple as possible)." ]

[ "Crashing while viewing animation in Cycles\nOperating system: macOS-13.0.1-arm64-arm-64bit 64 Bits\nGraphics card: Apple M1 Max Apple 4.1 Metal - 83\n\nBroken: version: 3.4.0\n\nWill crash seconds after playing animation in viewport when Cycles is selected and in render view. Both CPU and GPU crash. Does not crash when removing the Subdivision Surface Modifier so polycount may be a factor? This is a brand new Apple M1 Max computer from BestBuy with nothing else installed. Also tried initializing a render to video file and crashed as well. (Note: This is an error strictly with Apple M1 computers as I have tested it on an Apple intel MacBook Pro and it works just fine.) Thank you and happy holidays!\n\n1. Load file. It is a mesh plane with default cloth and subdivision modifiers acting on a wind force.\n2. Make sure Cycles is selected as Render Engine.\n3. Select Viewport Shading to Rendered (to view cycles render)\n4. Push play. Will crash or freeze seconds after play.\n\n[Cloth_Cycles_Crash.blend](Cloth_Cycles_Crash.blend)\n\n[Blender 3.4 Crash Report 12.13.22.rtf](Blender_3.4_Crash_Report_12.13.22.rtf)\n[Blender 3.5 Crash Report 12.14.22.rtf](Blender_3.5_Crash_Report_12.14.22.rtf)", "Animation Scene Strip rendering slower than Animation Scene rendering\nOperating system: Windows 10\nGraphics card: NVidia RTX 2060\n\nBroken: 3.3\nWorked: N/A\n\nRendering an animation using Scene Strip resync / reevaluate every mesh each frame, resulting in slow rendering.\nRendering the same animation directly from the same Scene only does this process for the 1st frame, hence a quicker rendering.\n(It doesn't seem to be related to the render engine, Eevee or Cycles same issue).\n\nAdditionally during Scene Strip rendering, on Windows, the memory doesn't seem to be cleaned correctly between each frame resulting in a crash at some point (depending on your memory size).\n\nAdd 2 subdivision modifiers to the cube, with 6 and 4 render levels.\nCreate a new scene, go to the VSE, add a Scene Strip pointing to the 1st scene.\nRender animation from VSE scene => slow rendering\nRender animation from the scene => fast rendering\n\n[scene_strip.blend](scene_strip.blend)", "Remesh Modifier in Voxel mode produces surface artifacts when Adaptivity enabled\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce RTX 2080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.71\n\nBroken: \n - 2.83.9\n - 2.91.0 Beta\n - 2.92.0 \nWorked: never\n\nRemesh modifier produces artifacts on mesh which look like a result of bad triangulation.\n\n1. Open attached blender file\n2. Open modifier panel and select testMesh\n3. Change adaptivity value from 1 to 0 cm and back\n4. Observe black shading artifacts \n[remesh_bug.blend](remesh_bug.blend)\n", "Using \"Rebuild Subdivisions\" in Multires modifier on hires mesh returns an error: Not valid subdivisions found to rebuild lower levels\nOperating system: Windows-10-10.0.19043-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 466.47\n\nBroken: version: 3.0.0\nWorked: IDK, first time user\n\nUsing \"Rebuild Subdivisions\" in Multires modifier on hires mesh returns an error: Not valid subdivisions found to rebuild lower levels\n[Head_Subdiv_Error.blend](Head_Subdiv_Error.blend)\n1. In the attached file add Multires modifier to the Head mesh\n2. Click Rebuild Subdivisions\n3. An error should be reproduced.\n\n", "Multires modifier data (CD_MDISPS) can remain on a (once shared) mesh, bloats filesize, adding a new multires modifer will already have its levels then\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 960/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 457.09\n\nBroken: version: 2.93.0 Beta\n\nWhen adding multires and press subdivide, a single press adds 3 levels at once and gives the result in the picture\n\nThe left part is obtained after applying the mirror modifier\n\nAdd multiresolution and press subdivide [bug.blend](bug.blend)\nIf the modifier is removed and re-added, the bug disappears\n\nFound a way to reproduce this:\n\n- - [x] add Multires to the Default Cube object, subdivide twice\n- - [x] add a Monkey\n- - [x] change the mesh of the Monkey to the Cube mesh under `Object Data Properties`\n- - [x] (you can delete the default cube object now if you like)\n- - [x] add Multires modifier to the Monkey object\n - > it already has 2 multires levels (without subdividing even once), if you now subdivide, it will jump right to level 3\n - > saving the file at step - [x] will bloat the filesize (even though there is no multires modifier anymore :/)\n\nnote: this behavior does not occur if one removes the Multires modifier from the Default Cube object prior to deletion in step 4\n\nSo opposed to any other modifier (afaict), Multires seems to store info in mesh that is kept around.\nIt does get removed from the mesh when the modifier gets removed, but it does not get removed from the (possibly shared) mesh when the object gets deleted.\n\nUpon further inspection, these are stored in a `CD_MDISPS` custom data layer on the mesh.\nNow when `ob->data` is changed (step - [x]), or when a new modifer is added, the levels of the modifer will be made matching (see `BKE_mesh_assign_object` & `BKE_modifiers_test_object` & `multiresModifier_set_levels_from_disps`)\nThis can also be observed when for example a second multires modifier is added on top of an existing mutires modifier (levels will also be made matching).\n\nNow I assume that in the end this has its reasons (and it might very well turn out to not be a bug), but will leave up to module devs to decide.\n(I think it would be reasonable to remove CD_MDISPS initially when adding a **new** multires modifier?)", "Artifacts appear in texture bakes [stripe pattern]\nOperating system: Windows 10 Pro 64-bit\nCPU: i7 4790\nGraphics card: GTX 1050\n\nBroken: Every Blender Build on this PC, Baking.\nWorked: never\n\nWeird artifacts appear in bakes, especially visible in the AO bakes (see attached images)\n\n\n\n\n\n[AOArtifacts_Test.blend](AOArtifacts_Test.blend)\n- Open blend\n- hit `Bake`", "Applying mirror modifier doesn't work with mesh-deform data\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce RTX 2070 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.38\n\nBroken: version: 2.91.0 Beta\n\napply the mirror modifier while is used with the multiresolution produces artifacts.\n\nopen the attached file and apply the mirror modifier.\n\nThank you\n\n\n\n[multires_err.blend](multires_err.blend)\n\n", "Compositor artifacts from image wrapping\nOperating system: Linux-5.15.13-arch1-1-x86_64-with-glibc2.33 64 Bits\nGraphics card: NVIDIA GeForce GTX 1060 6GB/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 495.46\n\nBroken: version: 3.1.0 Alpha\nBroken: 2.80 upwards\nWorked: 2.79\n\nIn the compositor, when using a Scale node with a value of < 1.0 after a Translate node with wrapping enabled, a strange pattern can be sometimes seen in the final render (sort of like Z-fighting). Also, re-rendering with the same settings doesn't always yield the same results with this setup.\n\nChanging the chunksize to anything other than 256 seems to fix this for me.\n\n\n\nI've created and attached a .blend file that has the compositor setup in the right way to display this bug. So all that's required is to render the scene - possibly refreshing the compositor a few times as the bug doesn't always occur.\n\n[blender-bug-report.blend](blender-bug-report.blend)", "Sharp normals along base mesh wires in multires sculpting with smooth shading\nOperating system: Linux-5.4.0-47-generic-x86_64-with-debian-bullseye-sid 64 Bits\nGraphics card: Quadro GP100/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 440.100\n\nBroken: version: 2.91.0 Alpha\nWorked: As far as I know in 2.79 before the inclusion of OpenSubdiv\n\nThis has been a long standing issue in multires sculpting with shading artefacts in smooth shading.\nThese will only start appearing when sculpting on the surface and may flicker between different spots when sculpting directly over them.\n\n- Create a suzanne head\n- Add a multires modifier with 4 subdivisions\n- Change the object to smooth shading\n- Enter sculpt mode and smooth on the surface of the object\n\nOn these spots the sharp shading should consistently appear among others. A high contrast Matcap can help with spotting them.\nIn this example they are very subtle but they can be extremely noticeable on very curved surfaces.\n\n\n\n\n\n", "Black Artifacts While Using Adaptive Subdivision\n{[F7718097](bug.png)}Operating system: Windows-10-10.0.18362 64 Bits\nGraphics card: GeForce GTX 1080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 436.15\n\n\nBroken: version: 2.80 (sub 75)\n\n\nBlack artifacts on edges of geometry while using adaptive subdivision. I was trying to duplicate my workflow from other 3d software. Extruding geometry from 2d curves creates extra edges on the model. And extra edges create weird shadows on displacement.\n\nImport curve from dxf file. Use extrude Object Data > Geometry. Fill mode: Both. Use subdivision and displacement.\n\n[1.blend](1.blend)\n", "Multires Unsubdivide freezes Blender with hidden geoemtry\nOperating system: Linux-5.15.0-41-generic-x86_64-with-glibc2.35 64 Bits\nGraphics card: Quadro GP100/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 510.73.05\n\nBroken: version: 3.3.0 Alpha\n\nIn the Multiresolution modifier the buttons \"Unsubdivide\" and \"Rebuild Subdivisions\" do a similar operation.\nWhen having any faces in edit/sculpt mode hidden though, using any of these operators will freeze Blender.\n\n- Subdivide Cube a couple times\n- Apply modifier\n- In edit mode, hide some faces\n- Add multires modifier (Don't add any more subdivisions)\n- Use \"Unsubdivide\" or \"Rebuild Subdivisions\" \n\n[#99887.blend](T99887.blend)", "Regression: Damaged normals in sculpting with Multires modifier\nBroken: version: 3.4.0 Alpha\nWorked: version: 3.4.0 Alpha\n\nWhen sculpting, sometimes breaks normals. Switching between modes fixes this.\n| Edited sculpture | Object mode |\n| -- | -- |\n|  |  |\n\n1. Open attached file.\n2. Go to Sculpt mode.\n3. Do something.\nLook at the fact that in some places the normals are damaged.", "QuadriFlow with the new symmetry option creates holes at the center\nOperating system: Windows-10-10.0.18362 64 Bits\nGraphics card: GeForce GTX 980/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 436.30\n\nBroken: version: 2.81 (sub 12)\n\n\nQuadriFlow creates hole artifacts like these when remeshing. Usually occurs after using Voxel Remesher and/or Dyntopo which can break symmetry at the center very often.\n\n\n1) Sculpt something in Sculpt Mode and use Voxel Remesher/Dyntopo.\n2) Start QuadriFlow. It doesn't always produce these holes in the mesh, but keep repeating part one and two and you will eventually get these holes where each side hasn't been properly merged together.", "Broken textures and shading\nOperating system: Linux-5.15.0-40-generic-x86_64-with-glibc2.35 64 Bits\nGraphics card: NVIDIA GeForce MX130/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 510.73.05\n\nBroken: version: 3.2.0\n\nBroken textures when starting up blender and loading object textures and shading issues\n\n[Blender Texture Bug.mp4](Blender_Texture_Bug.mp4)\n\nPossible causes:\n * Not loading the parent asset blend file before the asset in another scene (opening it first seems to fix the issue in my testing) \n * 4k textures (i've heavily used 4k textures in the asset)\n * Viewport Texture size limit setting (Changed it to 2k)\n\n\n[untitled.blend](untitled.blend)\n", "Limited dissolve ruins UV\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 531.79\n\nBroken: version: 3.5.1\n\nI made 3 or 4 times cycles of \"Limited dissolve - triangulate\" and get ruined UV after 3-rd or 4-th iteration. \nthe goal is to rid of exceeding geometry like on a screen.\n\n\n" ]

[ "Multires spikes when pressing Subdivide\nBlender 2.8 Beta Bug\n\nI Bug aprea when I increas the edges, sculpt in the first MultiRes Layer and after the third\nI produce some artefacts.\n\n[Knight.blend1](Knight.blend1)\n\n\n\n[Knight.blend](Knight.blend)" ]

[ "Cycles: overlapping VDBs cause rendering artifacts\nOperating system: Windows-10-10.0.19043-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 511.79\n\nBroken: version: 3.1.2\nWorked: version: 3.0.1\n\nFirst reported here: #97062\nOverlapping VDBs cause black rendering artifacts.\nThis is likely a known issue or related to a known issue.\nFrom the release notes for 3.1: [Ray Tracing Precision](Cycles#Ray_Tracing_Precision)\n>Many artifacts from rendering small, large and far away objects have been eliminated. There will always be object scales where numerical precision becomes a problem, but it's further out now.\n>\n>There can still be artifacts with rendering overlapping geometry, in some cases more severe than before. Such overlapping geometry should be removed, or have a small distance added in between.\n\n\n|3.0.1|3.1.2 CPU|3.1.2 GPU|\n| -- | -- | -- |\n||||\n\n1) Import a VDB volume.\n2) Duplicate the volume without moving it.\n3) Render with Cycles.\n\nHere is the test file:\n[#97094.zip](T97094.zip)", "Display issue in model corner when loop cutting\nBroken: 2.79, 2.81\nWorked: -\n\nWhen creating and edge loop around de outside faces it creates in one of the cornes an extrange artifact\n\nCreate an edge loop across the outside faces of the model.\n\nBlender file:\n[loop cut.blend](loop_cut.blend)\n\nArtifact:\n\n", "Ambiant Occlusion cycle node gives artecacts when the mesh has Custom Split normals\nOperating system: win 10 \nGraphics card: Geforce RTX 2070\n\n**Blender Version** \nBroken: 2.90.0 alpha\nWorked: N/a\n\nOn my mesh , with custom normals, (i applied a withged normal modifier) \nI had to clear the custom split normal date to get rid of artefacts given with the Ambiant Occlusion node in Cycles\n\n\nAlso seen with smooth shader:\n\n\nAnother way to have this bug (without Custom Split normals): \n- Create a suzanne.\n- Smooth the normals \n- Use Cycles render. \n- Add an Ambient occlusion node and plug it on an Emit shader\n[cycles_ao_smooth_normals_bug.blend](cycles_ao_smooth_normals_bug.blend)", "Multires Unsubdivide Deletes Hidden Geometry\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 527.56\n\nBroken: version: 3.6.0 Beta\nWorked: ??? also a bug in 3.5\n\n**Description**\nIf you use the unsubdivide tool on the multires modifier while geometry is hidden in edit mode, the geometry that was hidden gets 'deleted' (or more like dissolved) from the mesh. It's easy to work around, but it definitely isn't the expected behavior.\n\nSelect the default cube and enter edit mode, subdivide it several times, hide part of the mesh, back to object mode and add a multires modifier, click unsubdivide, the hidden geometry is dissolved.\n\n", "MultiRes+Solidify while trying to use Sculpt Mode cause Blender to crash\nOperating system: Windows-10-10.0.18362 64 Bits\nGraphics card: GeForce GTX 980/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 436.30\n\nBroken: version: 2.81 (sub 12)\nWorked: (optional)\n\nI tried deactivating one modifier while keeping the other one active to see if Blender would crash, but this was not the case. Only when I used both simultaneously did the program reliably crash every time, regardless of brush.\n\nAdd the modifiers that are active in the video and follow along. \n[Desktop 2019.09.23 - 00.49.39.02.mp4](Desktop_2019.09.23_-_00.49.39.02.mp4)\n", "Sharp normals along base mesh wires in multires sculpting with smooth shading\nOperating system: Linux-5.4.0-47-generic-x86_64-with-debian-bullseye-sid 64 Bits\nGraphics card: Quadro GP100/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 440.100\n\nBroken: version: 2.91.0 Alpha\nWorked: As far as I know in 2.79 before the inclusion of OpenSubdiv\n\nThis has been a long standing issue in multires sculpting with shading artefacts in smooth shading.\nThese will only start appearing when sculpting on the surface and may flicker between different spots when sculpting directly over them.\n\n- Create a suzanne head\n- Add a multires modifier with 4 subdivisions\n- Change the object to smooth shading\n- Enter sculpt mode and smooth on the surface of the object\n\nOn these spots the sharp shading should consistently appear among others. A high contrast Matcap can help with spotting them.\nIn this example they are very subtle but they can be extremely noticeable on very curved surfaces.\n\n\n\n\n\n", "Performance test and notes regarding multires\nDaniel Bystedt and Dalai Felinto had a catch up meeting 11:th of May. During that meeting we talked a bit about sculpting. Dalai asked me to create this task and to add a test file and some more info which is detailed below.\n\n**Computer and Blender specs**\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 465.89\n\nversion: 2.92.0\n\n\n**Blend file**\n[orc_head_5_subd_v001.blend](orc_head_5_subd_v001.blend)\n\n**Switching symmetry on/off**\nSwitching symmetry on/off is something a user does quite often in order to not make human faces (for example) look totally mirrored when sculpting. It’s a bit surprising that it is so slow when the vertex count is high. \n\n@PabloDobarro explained via chat: *The lag from turning on/off symmetry is because of a BVH rebuild. Symmetry is a property of meshes, so each time a property of the mesh changes, it considers it as a change in geometry and rebuilds the BVH again. It should be possible to do a more proper tagging of these changes to avoid this, which will fix this multires and any other future system we can make.*\n\n|Vertex count| Time to switch symmetry on/off\n| -- | -- |\n|10 million| 3 s 7ms|\n\n**Subdivide performance**\nSubdividing when vertex count will be above 1-3 million is usually rather slow. The user only have to subdivide the model once when sculpting a model so it’s not a frequent thing that you do. If it could be made faster it would be a nice improvement though\n\n|Vertex count| Time subdivide with multires|\n| -- | -- |\n|subdiv 2.6 million > 10 million| 16 s 5 ms|\n\n**Switching sculpt levels performance**\nSwitching sculpt subdivision levels is something the user does quite often, so if it is fast it would be great. The performance is not terrible though.\n\n|Vertex count| Time subdivide with multires|\n| -- | -- |\n|2.6 million > 10 million| 1 s 8 ms|\n\n\n**Performance investigation starting point**\n- Run a profiler to see if we do not have obvious bottlenecks and see that all threads are occupied\n- Measure exact memory footprint of OpenSubdiv topology. If it is not terribly big, can cache those on different levels while being in the sculpt mode\n- Look into sparse patch tables. This seems to be a solution to a specific of OpenSubdiv which is related on a single-threaded nature of topology construction\n- Run quick and dirty check whether using GPU evaluation brings 10x speedup (quick and dirty is because doing proper integration is not trivial and you don't want to do it if the speedup is not that great)\n- Check whether all tools are properly multi-threaded\n- Make ID_RECALC_PARAMETERS to work for non-geometry properties of meshes (to mitigate the symmetry lag)", "unsubdivide of Multires bug\n**Blender Version**2.91\n\nI try to use unsubdivide of Multires function in blender ,but the result is error.So I test in Zbrush with same modle, it work rightly.\n\nThe thing is,I want sculpt with tank top with no thickness,and add thick ness at last.So I first sculpt with Multires (3 level subdivide),and apply it,and I loop cut the edge with 7 loop cut.After that I add Multires,and use unsubdivide,the result go wrong.\n\nI use the same modle in Zbrush,it work rightly.[unsubdivide error.zip](unsubdivide_error.zip)", "Sculpt Mode Stroke Perfomance\nThis task intention is to plan how to tackle the stroke performance issue that sculpt mode has. This is the issue we are trying to solve:\n\n[2020-08-08 13-07-14.mp4](2020-08-08_13-07-14.mp4)\n\nThe plan is to avoid the slowdown that freeze the UI during the stroke. In any case, if the computer hardware is not enough to keep up with the mesh deformation, the stroke should lag consistently instead of freezing for several seconds.\n\nThis should not focus on:\n - Dyntopo performance (should be handled in a separate task as it has some specific performance issues).\n - Making large brush strokes and mesh filters faster.\n\nThese are the attempt the were made. There were some successful experiments, but none of them can provide the expected performance sculpt mode should have in a high end computer: \n\n### Control paint operator events rate\n\n- Dynamic Stroke Spacing [D8508](D8508): This is the solution with the most noticeable effect so far. In order to make this work, high frequency pen tablet events should be available in all platforms. This patch can be beneficial to merge also for not performance reasons as it will also allow to reduce the spacing of some brushes that need it. \n- Paint stroke step queue [D5676](D5676): This is not related to improve performance (it may even make performance worse), but to improve UX avoiding locking the UI. The idea was to use it for 2D painting were it works fine, but it 3D projection painting and Sculpt Mode it still has some issues.\n- Ignore INVETWEEN_MOUSE_EVENTS in grab brushes (in master since 2.81) [D5429](D5429): This also had a noticeable effect in the performance of these brushes.\n\n\n### Modify PBVH scheduler settings\n\n- Change Multires to limit the PBVH leaf size using vertices [D8454](D8454): This reduces the leaf limit for Multires. It improved performance in high end computers (Julien workstation). Before this patch, Multires was lagging with 80k vertices with 3 subdivisions levels.\n- Change PBVH leaf limit size [D8442](D8442): This has a noticeable effect in performance. Stroke performance usually improves when making the limit size lower, but this makes tools that modify all nodes slower (like elastic deform or the mesh filters). When increasing the limit size, stroke performance is worse, but tools that modify the whole mesh are faster.\n- PBVH Scheduler performance design task: #72943\n\n\n### Other info\n\nI tested sculpt mode in the following computers, all of them running Ubuntu:\n\n- Mini-PC media center: i3, 8GB integrated graphics.\n- Gaming Laptop: i7 9750H, GTX 1650, 16GB\n- Workstation: Threadripper 3990X, Radeon Pro VII, 128GB\n\nThis is my experience so far\n\n- Performance does not scale: The workstation and the mini-pc have the same stroke performance and lag problem. In the case of the mini-pc, performance is clearly limited by the GPU as the viewport lags with high poly meshes without doing any strokes.\n- Removing the normal recalculation completely before drawing the PBVH does not have any noticeable effect. If we can find a fix that only works for flat shading (so we can move these updates to somewhere else), it should be acceptable, \n- I did a refactor during the 2.82 development to unify all the loops for the Clay Strips brush as suggested in #68873. It didn't have any noticeable performance effect.\n- In 2.81 the scheduler was changed to TBB. Before that change, Sculpt Mode was noticeable faster disabling multithreaded sculpting. This option is no longer available.\n- This bug fix removed a heavy computation per vertex in clay strips. Even after the fix, the stroke lag is still there\n- Stroke lag is generally higher in Multires than in meshes, when we should expect Multires to be faster than meshes (less cache misses).\n- In the gaming laptop with the EEVEE PBVH drawing enabled, stroke performance is almost the same (if not better) in EEVEE than in workbench. I did't test this with the workstation because I can't guarantee that the new GPU is working correctly. This makes me think that this problem is not related to a viewport drawing issue.\n- The default Sculpt Vertex Colors paint brush does not need normal updates, sampling the area normal or updating the nodes bounding boxes, but it is also affected by the stroke lag issue. \n- The Brush cursor is using a PBVH raycast and a multithreaded task to sample the area normal, but it is not affected by the lag, no matter how fast you move the cursor over the mesh. \n\n### Performance reference benchmark\n\nEven this is not an easy to define thing, we should have a fixed performance reference benchmark for sculpting for supported devices. This should allow people to report performance issues with different setups instead of just assuming that performance is bad because Blender is not a especialized software, so we can have more data to debug this issue (the same way if we notice that moving elements in edit mode on a 20 vertices mesh lags, we know it is a bug).\n\nAs and initial proposal, I would expect that with a 500k - 750k vertices mesh, Sculpt Mode should never lag regardless of the stroke size or stroke speed, in any computer that meets the minimum requirements to run Blender.\n\n\n\n\n\n\n\n", "Render issues when two VBD overlap if rendered with motion blur\nBroken: 3.2, 3.6\nWorked: 3.1\n\nRender issues when two VBD overlap if rendered with motion blur\n\nRender the attached file with and without motion blur. If you render them separatly they both render OK. Only when together it doesn't work.\n\n", "Small Custom Bone Shapes have some Wireframes Artifact.\nWindows 10 64 Bits.\nGraphics card GeForce GTX 1080 Ti \n\nBroken: version: 2.82.6, hash: a854840e76ae\nWorked: (optional)\n\nSmall custom bone shape has wireframe glitching, the Scale is based on real world but if you resize it in/out u can see it happen.\n\n{[F8257584](Shape.gif) size = full}\n\n\n\n[Bug.blend](Bug.blend)\n", "Unsubdivide and rebuild subdivisions from multires.\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3060 Laptop GPU/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 531.41\n\nBroken: version: 3.5.1\n\nModel with multires applied doesnt works with unsubdivide option or rebuild subdivisions.\nI´ve applied the modifier, cleaned the data blocks but persist...\nThere is something strange in the mesh because it works with the default cube...\n\n- Open attached .blend file\n- In the Multires modifier, click in Subdivide\n- Apply modifier\n- Add multiresolution modifier again\n- Click in Unsubdivide\n\n> No valid subdivisions found to rebuild a lower level\n\n[UnsubdivideMultires.blend](attachment)\n\n", "Multires disp bake doesn't smooth UV\nOperating system: Windows 10\nGraphics card: RTX 3060ti\n\nBroken: 2.92, 02948a2cab44\nWorked: Don't know\n\nWhen baking a multires sculpt it will bake the disp towards a subdivided mesh but not a smoothed UV.\n\nGuess the best way is just to download my attached .blend, make sure the image texture is selected, that the Level Viewport on the multires modifier is at 0 and use the bake under Render properties. The resulting texture should not have this clear square in the middle as in my example picture below, but it do because the UV ain't getting smoothed. I did make sure that I had UV smooth enabled when I first subdivided the mesh.\nAlso added an example from what happens when I do this on a real project. This is baked and then applied to a subdivded mesh, look at the wrinkles below the eye.\n[multires_bug.blend](multires_bug.blend)\n\n", "Inconsistent behavior of the Subdivide Edge Ring and Subdivide actions on meshes with shape keys\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 511.23\n\nBroken:\n3.2.0 Alpha 2022-05-01 a5644f9a28ea\n3.1.2 2022-03-31 cc66d1020c3b\n3.0.1 2022-01-25 dc2d18018171\n2.93.9 2022-04-19 31712ce77a6a\n2.83.19 (can't find build details)\n\nWorked: -\n\nThere's an inconsistent behavior of the Subdivide Edge Ring and Subdivide actions on meshes with shape keys.\n\nWith Subdivide Edge Ring, no matter what kind of interpolation and profile factor you use, the one applied to all other shape keys will use \"Linear\" and \"0\" respectively.\n\nWith Subdivide, if you use it on the Basis shape key, all other shape keys will inherit half of the transform of the \"smoothness\" setting, and they will be transformed in some undefined manner by the \"fractal\" setting as well. If you use it on non-Basis shape key, in addition to the described behavior with remaining shape keys, the Basis shape key will always use the default settings (smoothness and fractal = 0).\n\n**Exact steps for others to reproduce the error / sort-of-transcript of the video**\nwatch?v=s9dPcdHOlpo\nIf you use the Subdivide Edge Ring, leave the default settings, and go through the shape keys, you can see that the positions of vertices are slightly off between the current/modified shapekey and others.\nHowever, that changes when you change interpolation to \"Linear\" - then they start being the same.\nWhen you go through various options it seems that the only thing applied properly across all shape keys is the number of cuts.\nAll shape keys other-than-the-active-one seem to always be using the \"Linear\" interpolation, and profile factor = \"0\"\nFrom my tests it doesn't matter which shape key you use this action on - it will always have the same effect between the current shape key and all other.\n\nNow, there's something even weirder goin on with the regular Subdivide.\nIf we Subdivide the mesh with default settings, it will look the same in all shape keys - that's great.\nHowever, if we change something, then the mesh across shapekeys will look different.\nAs a general rule, if we Subdivide a shapekey like Key 2, then the Basis will look like it was Subdivided with the default settings instead of the chosen ones (besides the number of cuts), but all other shape keys will also be modified, just in a different way than the active shapekey.\nFor example, if we change the smoothness of our Subdivide, then other shapekeys will look just \"less modified\". If we check the coordinates of the vertices, then we can see that in Key 1 they are transformed exactly half-way to their position in Key 2.\nHowever, if we use the Fractal setting, then the results are different: The Basis is still default-Subdivided, but the vertices of Key 1 are not just halfway there to the positions of Key 1 - they're completely different.\nOn a final note, with regular Subdivide, if, instead of Subdividing Key 2 you did that to the Basis, then both Key 1 and 2 would look like Key 1 in my example. There would be no shape key that stays with default Subdivide settings.", "Mesh to Volume intersection artifacts\nOperating system: Ubuntu18x64, Ubuntu20x64, Win8x64.\nGraphics card: Mesa DRI Intel(R) HD Graphics (BYT) Intel Open Source Technology Center 4.2 (Core Profile) Mesa 20.0.8\n\nBroken: version: 3.0, 3.1\n\nIntersecting Mesh to Volume produces artifacts (on CPU. GPU not tested).\n\n[volume-intersection.blend](volume-intersection.blend)\n - Open,\n - Run viewport Render Preview on the left side." ]

[ "multiresolution modifier artifacts at 4 subdivision level\nOperating system: Windows-7-6.1.7601-SP1 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 431.60\n\nBroken: version: 2.81 (sub 15)\n\n\ngoing up to 4 subdivision using the multiresolution modifier break the mesh\n\nopen the file, assign multiresolution modifier and subdivide 4 times\n\nthank you.\n\n[multires.blend](multires.blend)\n\n" ]

[ "EEVEE crash while rendering animation, cycles works.\nOperating system: Win10\nGraphics card: GTX1070MAX-Q\nCPU : i7-6700HQ\nRAM : 32GB DDR4\n\nBroken: 2.90\nWorked: -\n\nBlender crashes w/o any errors when I started to rendering animation in Eevee. \nIt happens on 2 laptops and it always happen in 154 or 155 frame while rendering.\nIn attachment you can download the project which gets this error. \nWhen I use cycles rendering it works perfect.\n\n[#80740.blend](T80740.blend)\n\n- Open file\n- Render frames 154 and 155\n", "Crashes when Switching to Material Viewport / Rendered Viewport / Rendering\nOperating system: Windows 10\nGraphics card: 2 Nvidia GTX 1660 Super's\n\n[blender_system_info.txt](blender_system_info.txt)\n\nBroken: blender 2.8 / 3.4\n\n\nBlender runs completely fine when editing and building a model but once rendering is involved, even in preview, it immediately closes with no explanation. This crash will also happen with the default cube. This had not happened before and have tried Updating blender, reinstalling, simplifying, etc but no change and can reproduce the issue on my computer. i can have it in material mode if i have 0 assets in the scene but once i add something it crashes, i am suspecting a bad pathing issue as i had recently refreshed my computer but do not know how to fix it or give blender a hard reset.\n\n{[F14170509](blender.crash.txt)}[blender_debug_output.txt](https://archive.blender.org/developer/F14170500/blender_debug_output.txt)\nOpen Blender, have any Physical asset inside the scene, switch to rendered / Material Viewport, Crash\n\nI have included the project I was working on and the default cube as they create the same issue.\nPlease note that I have installed the most recent Blender program onto my laptop and opened the files attached and did not have any crash happen so it seems the issue is solely between my computer and the blender program.\n\n[Default Cube.blend](Default_Cube.blend) [Skell Rig base.blend](Skell_Rig_base.blend)", "Cycles HIP rendered viewport crashes system/GPU on Linux with RDNA2 GPU\n Operating system: Fedora Linux Release 36 - Linux-5.18.16-200.fc36.x86_64-x86_64-with-glibc2.35 64 Bits\nGraphics card: AMD Radeon RX 6900 XT (sienna_cichlid, LLVM 14.0.0, DRM 3.46, 5.18.16-200.fc36.x86_64) AMD 4.6 (Core Profile) Mesa 22.1.5\nROCm Version: rocm-5.2.0\n\nBroken: version: 3.2.2\nWorked: never\n\nWhen using HIP on AMD GPU together with Cycles GPU Rendering my entire system freezes when using the rendered viewport, and I have to hard-reset the PC. I think it's the GPU /GPU-driver that crashes. \nKernel messages from journalctl suggest amdgpu (kernel module?) crashing (see attached) [journalctl.txt](journalctl.txt)\n\nThe crash is not always immediate and might instead only happen after using the rendered viewport for a while or switching back and forth between textured and rendered a few times.\n\nRendering with F12 works fine and I have observed no other crashes as long as I don't use the rendered viewport.\n\nI have only started using blender a few days ago creating procedural textures in the node editor and using the rendered viewport all the time and the crash didn't happen once.\nNow that I started working with image-texture-based assets like in the attached .blend file this problem started happening.\nMaybe it's related with this issue: T97591\n\nCan I somehow get more detailed debug output from blender? The \"-d\" switch didn't generate much more relevant output.\n\nI could reliably reproduce the issue by following these steps [test.blend](test.blend)\n1) Open up two Blender instances with the attached .blend File loaded and have the viewport running in rendered mode\n2) move the viewport in the first instance to get it to \"refresh\"/re-render\n3) while the first instance is still working - move the viewport in the second instance - for me the System/GPU freeze/crash happens right there. \n\n--\nlet me know what info I can provide to help.\n\nThank you!", "Running multiple instances of Cycles HIP ends in segfault\nOperating system: Windows 10\nGraphics card: 2x Radeon W6800 32GB\n\nBroken: 3.0.0, 3788003cdaeb, master, 2021-11-30\nWorked: Never\n\nWhen I have 2 instances of Cycles rendering using HIP at the same time on a particular production scene, one instance will crash with the following error: \n\n\n```\n...\nFra:441 Mem:8965.99M (Peak 8965.99M) | Time:00:21.57 | Mem:1469.81M, Peak:1469.81M | Scene, View Layer | Updating Images | Loading Scene2_Garage_Furniture_Roughness.png\nFra:441 Mem:8965.99M (Peak 8965.99M) | Time:00:21.70 | Mem:1533.81M, Peak:1533.81M | Scene, View Layer | Updating Images | Loading Metal02_nrm.jpg\nFra:441 Mem:8965.99M (Peak 8965.99M) | Time:00:21.73 | Mem:1569.81M, Peak:1569.81M | Scene, View Layer | Updating Images | Loading Scene2_Garage_Objects_normal.png\nFra:441 Mem:9030.99M (Peak 9030.99M) | Time:00:21.73 | Mem:1570.69M, Peak:1570.69M | Scene, View Layer | Updating Images | Loading Tires_rough.png\nMalloc returns null: len=37748736 in Cycles Aligned Alloc, total 1149156740\nError : EXCEPTION_ACCESS_VIOLATION\nAddress : 0x00007FFED98C853D\nModule : amdhip64.dll\nThread : 00000734\n\n```\nI was unable to reproduce it in a simple scene or any commonly available demo scenes such as classroom or Barbarshop. And unfortunately I cannot share the particular scene that's causing the issue. But the scale of the scene that's crashing should easily fit onto the GPU vram three times.\n\nI've attached the crash report. If more info is needed I will try to create a dummy scene that show the same behavior.\n\n[Anim-A-Garage.crash.txt](Anim-A-Garage.crash.txt)", "Blender 2.91 crashes when enabling Archipack addon\nOperating system: Linux-5.8.0-23-generic-x86_64-with-debian-bullseye-sid 64 Bits\nUbuntu 20.10 Groovy Gorilla (development branch)\nGraphics card: Mesa Intel(R) UHD Graphics 620 (KBL GT2) Intel 4.6 (Core Profile) Mesa 20.2.0\n\nBroken: version: 2.91.0 Alpha\n\nBlender crashes after enabling Archipack addon. Terminal says 'Aborted (core dumped).' Also, Blender doesn't record a crash.txt (that I can find). \n\nStart Blender>Edit>Preferences>Addons>Enable Archipack addon.\n\n[CoreDump.mp4](CoreDump.mp4)\n", "Crash choice asset library\nOperating system: Win 10\nGraphics card: RX460 2gb\n\nBroken: (3.2)\nWorked: (3.0)\n\n[Tests.zip](Tests.zip)\n\n- Extract archive and add `...\\Tests\\Test Asset\\` path to asset libraries\n- Open Asset library blend file\n- Open New file\n- Change any editor to Asset Browser\n# List contents of `Test Asset` library\n\nIt should crash.\nSometimes, omitting step 2 would prevent crash, but no assets would be shown.\nAlso omitting `tests` folder somehow prevented the crash, \n\nCrash log:\n[P3143: (An Untitled Masterwork)](P3143.txt)\n```\n# Blender 3.4.0, Commit date: 2022-08-14 18:40, Hash 659b63751d63\nbpy.context.space_data.params.asset_library_ref = 'Test Asset' # Property\n\n# backtrace\nException Record:\n\nExceptionCode : UNKNOWN EXCEPTION\nException Address : 0x00007FFF3F66D759\nException Module : KERNELBASE.dll\nException Flags : 0x00000001\nException Parameters : 0x4\n\tParameters- [x] : 0x0000000019930520\n\tParameters- [x] : 0x000000FB7C5FD460\n\tParameters- [x] : 0x00007FF678278398\n\tParameters- [x] : 0x00007FF66CA80000\n\n\nStack trace:\nKERNELBASE.dll :0x00007FFF3F66D6F0 RaiseException\nVCRUNTIME140.dll :0x00007FFF283E63F0 CxxThrowException\nblender.exe :0x00007FF672BC42D0 nlohmann::detail::serializer<nlohmann::basic_json<nlohmann::ordered_map,std::vector,std::basic_stri\nblender.exe :0x00007FF672BC39A0 nlohmann::detail::serializer<nlohmann::basic_json<nlohmann::ordered_map,std::vector,std::basic_stri\nblender.exe :0x00007FF672BC39A0 nlohmann::detail::serializer<nlohmann::basic_json<nlohmann::ordered_map,std::vector,std::basic_stri\nblender.exe :0x00007FF672BC39A0 nlohmann::detail::serializer<nlohmann::basic_json<nlohmann::ordered_map,std::vector,std::basic_stri\nblender.exe :0x00007FF672BC39A0 nlohmann::detail::serializer<nlohmann::basic_json<nlohmann::ordered_map,std::vector,std::basic_stri\nblender.exe :0x00007FF672BC36C0 nlohmann::basic_json<nlohmann::ordered_map,std::vector,std::basic_string<char,std::char_traits<char\nblender.exe :0x00007FF672BC82B0 blender::io::serialize::JsonFormatter::serialize\nblender.exe :0x00007FF66E27B970 blender::ed::asset::index::AssetIndexFile::write_contents\nblender.exe :0x00007FF66E277D40 blender::ed::asset::index::update_index\nblender.exe :0x00007FF66E653B90 filelist_readjob_list_lib\nblender.exe :0x00007FF66E654370 filelist_readjob_recursive_dir_add_items\nblender.exe :0x00007FF66E653710 filelist_readjob_asset_library\nblender.exe :0x00007FF66DB341A0 do_job_thread\nblender.exe :0x00007FF672E5A810 _ptw32_threadStart\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThreads:\nThread : 00000448\nntdll.dll :0x00007FFF41D2C620 ZwDelayExecution\nblender.exe :0x00007FF672B3A220 BLI_thread_is_main\n\n\nThread : 000069ac\nntdll.dll :0x00007FFF41D2F9F0 ZwWaitForWorkViaWorkerFactory\nntdll.dll :0x00007FFF41CA0E20 LdrAccessResource\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00003130\nntdll.dll :0x00007FFF41D2F9F0 ZwWaitForWorkViaWorkerFactory\nntdll.dll :0x00007FFF41CA0E20 LdrAccessResource\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 000069cc\nntdll.dll :0x00007FFF41D2F9F0 ZwWaitForWorkViaWorkerFactory\nntdll.dll :0x00007FFF41CA0E20 LdrAccessResource\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 000054bc\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 000026c0\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00008788\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 0000aa4c\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 0000608c\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 0000ad7c\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00009c7c\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00008d78\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 000029c4\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00009918\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 0000162c\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00001478\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00009a78\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 000067cc\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 000040b4\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 000036d0\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00003468\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00007444\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00005208\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 000059e4\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 0000a018\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 000085a8\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 0000b1ac\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 0000921c\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00005544\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 0000ad94\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 000067e0\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 0000401c\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 0000b360\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00008b38\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00005884\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00003334\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nblender.exe :0x00007FF673904E60 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF673904670 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF66CDB3A50 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00008c60\nntdll.dll :0x00007FFF41D2F9F0 ZwWaitForWorkViaWorkerFactory\nntdll.dll :0x00007FFF41CA0E20 LdrAccessResource\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 0000a2e8\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\nnvoglv64.dll :0x00007FFE81C3A540 DrvValidateVersion\nnvoglv64.dll :0x00007FFE819D057C Symbols not available\n\n\nThread : 00009be0\nntdll.dll :0x00007FFF41D2CAF0 ZwWaitForMultipleObjects\nKERNELBASE.dll :0x00007FFF3F690C50 WaitForMultipleObjectsEx\nKERNELBASE.dll :0x00007FFF3F690C30 WaitForMultipleObjects\nnvoglv64.dll :0x00007FFE81C6EF90 DrvPresentBuffers\nnvoglv64.dll :0x00007FFE81C6EF90 DrvPresentBuffers\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00003f9c\nwin32u.dll :0x00007FFF3FCCA0F0 NtUserMsgWaitForMultipleObjectsEx\nUSER32.dll :0x00007FFF411906F0 MsgWaitForMultipleObjectsEx\nnvoglv64.dll :0x00007FFE81C3A540 DrvValidateVersion\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00007978\nntdll.dll :0x00007FFF41D2F9F0 ZwWaitForWorkViaWorkerFactory\nntdll.dll :0x00007FFF41CA0E20 LdrAccessResource\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 0000ace0\nntdll.dll :0x00007FFF41D2CAF0 ZwWaitForMultipleObjects\nKERNELBASE.dll :0x00007FFF3F690C50 WaitForMultipleObjectsEx\ncombase.dll :0x00007FFF41539AA0 RoGetServerActivatableClasses\ncombase.dll :0x00007FFF41539AA0 RoGetServerActivatableClasses\ncombase.dll :0x00007FFF41539AA0 RoGetServerActivatableClasses\ncombase.dll :0x00007FFF41539AA0 RoGetServerActivatableClasses\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 0000ada4\nwin32u.dll :0x00007FFF3FCCA0F0 NtUserMsgWaitForMultipleObjectsEx\nUSER32.dll :0x00007FFF411906F0 MsgWaitForMultipleObjectsEx\ndirectmanipulation.dll:0x00007FFF3012457B Symbols not available\n\n\nThread : 00009dbc\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00005f9c\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00009dac\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 0000405c\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 0000834c\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 000099e8\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00006284\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00005ec8\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 0000a8fc\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00003240\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00000a10\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 000082dc\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 0000196c\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00004374\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00007dec\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00004294\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 0000559c\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00009a90\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00004cc8\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 000029cc\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 000055f8\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 000069a4\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00005c40\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00001a80\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00004308\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 000081c8\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 000071a0\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 0000b0d4\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00004bc0\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00009d28\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nThread : 00008010\nntdll.dll :0x00007FFF41D2C020 ZwWaitForSingleObject\nKERNELBASE.dll :0x00007FFF3F65A2C0 WaitForSingleObjectEx\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFF1D64D800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFF3F931430 configthreadlocale\nKERNEL32.DLL :0x00007FFF40637020 BaseThreadInitThunk\nntdll.dll :0x00007FFF41CDD0B0 RtlUserThreadStart\n\n\nLoaded Modules :\n0x00007FF66CA80000 3.4.0.0 blender.exe C:\\Users\\John\\Desktop\\blender-3.4.0-alpha+master.659b63751d63-windows.amd64-release\\blender.pdb \n0x00007FFF41C90000 10.0.19041.662 ntdll.dll \n0x00007FFF40620000 10.0.19041.662 KERNEL32.DLL \n0x00007FFF3F640000 10.0.19041.662 KERNELBASE.dll \n0x00007FFF40C70000 10.0.19041.546 WS2_32.dll \n0x00007FFF41310000 10.0.19041.662 RPCRT4.dll \n0x00007FFF41170000 10.0.19041.685 USER32.dll \n0x00007FFF1D640000 tbb.dll \n0x00007FFF3FCC0000 10.0.19041.662 win32u.dll \n0x00007FFF41940000 10.0.19041.685 GDI32.dll \n0x00007FFF1D600000 tbbmalloc.dll \n0x00007FFF3FAB0000 10.0.19041.685 gdi32full.dll \n0x00007FFF3FA10000 10.0.19041.546 msvcp_win.dll \n0x00007FFF3F910000 10.0.19041.546 ucrtbase.dll \n0x00007FFF417D0000 10.0.19041.610 ADVAPI32.dll \n0x00007FFF406E0000 7.0.19041.546 msvcrt.dll \n0x00007FFF41980000 10.0.19041.662 sechost.dll \n0x00007FFF3FED0000 10.0.19041.662 SHELL32.dll \n0x00007FFF283E0000 14.29.30139.0 VCRUNTIME140.dll \n0x00007FFF3AE70000 14.29.30139.0 VCRUNTIME140_1.dll \n0x00007FFF40F00000 10.0.19041.546 SHLWAPI.dll \n0x00007FFEC90D0000 14.29.30139.0 MSVCP140.dll \n0x00007FFF3FC70000 10.0.19041.546 CFGMGR32.dll \n0x00007FFF36DA0000 10.0.19041.488 dbghelp.dll \n0x00007FFF3FBC0000 10.0.19041.662 bcrypt.dll \n0x00007FFF40CE0000 10.0.19041.546 ole32.dll \n0x00007FFF41470000 10.0.19041.662 combase.dll \n0x00007FFEC8180000 10.0.19041.546 OPENGL32.dll \n0x00007FFF41440000 10.0.19041.546 PSAPI.DLL \n0x00007FFF41A20000 10.0.19041.662 shcore.dll \n0x00007FFF3FEA0000 10.0.19041.546 IMM32.dll \n0x00007FFEB74B0000 9.0.0.0 openvdb.dll \n0x00007FFEC5BE0000 59.16.100.0 avformat-59.dll \n0x00007FFEB4FB0000 59.18.100.0 avcodec-59.dll \n0x00007FFF1D5D0000 59.4.100.0 avdevice-59.dll \n0x00007FFF40FC0000 10.0.19041.546 OLEAUT32.dll \n0x00007FFEC9030000 6.4.100.0 swscale-6.dll \n0x00007FFEC5A00000 57.17.100.0 avutil-57.dll \n0x0000000070440000 1.0.28.0 libsndfile-1.dll \n0x00007FFEC5190000 1.21.1.0 OpenAL32.dll \n0x000000006ACC0000 libgmp-10.dll \n0x00007FFEC29E0000 2.0.20.0 SDL2.dll \n0x00007FFF39630000 libgmpxx.dll \n0x00007FFEC0D70000 3.10.2150.1013 python310.dll \n0x00007FFF40780000 10.0.19041.546 SETUPAPI.dll \n0x00007FFF300E0000 tbbmalloc_proxy.dll \n0x00007FFF31280000 6.10.19041.488 COMCTL32.dll \n0x00007FFF1C490000 10.0.19041.1 AVIFIL32.dll \n0x00007FFF39120000 10.0.19041.546 VERSION.dll \n0x00007FFF3D210000 10.0.19041.546 dwmapi.dll \n0x00007FFF1C460000 10.0.19041.546 GLU32.dll \n0x0000000070680000 libfftw3-3.dll \n0x00007FFF27EE0000 10.0.19041.546 Secur32.dll \n0x00007FFECB6B0000 4.3.100.0 swresample-4.dll \n0x00007FFECBF00000 10.0.19041.1 AVICAP32.dll \n0x00007FFF2EBF0000 10.0.19041.546 WINMM.dll \n0x00007FFF20660000 10.0.19041.1 MSACM32.dll \n0x00007FFECB5D0000 10.0.19041.1 MSVFW32.dll \n0x00007FFF24F20000 10.0.19041.1 winmmbase.dll \n0x00007FFF3F280000 10.0.19041.662 SSPICLI.DLL \n0x00007FFF3D240000 10.0.19041.546 kernel.appcore.dll \n0x00007FFF3D520000 10.0.19041.662 windows.storage.dll \n0x00007FFF3EE40000 10.0.19041.662 Wldp.dll \n0x00007FFF3CE80000 10.0.19041.610 uxtheme.dll \n0x0000022043090000 10.0.19041.662 bcryptPrimitives.dll \n0x00007FFF3F300000 10.0.19041.546 profapi.dll \n0x00007FFF41880000 2001.12.10941.16384 clbcatq.dll \n0x00007FFF38BE0000 10.0.19041.546 MMDevApi.dll \n0x00007FFF3F180000 10.0.19041.546 DEVOBJ.dll \n0x00007FFF38EF0000 10.0.19041.662 AUDIOSES.DLL \n0x00007FFF3E980000 10.0.19041.546 powrprof.dll \n0x00007FFF3E840000 UMPDC.dll \n0x00007FFF41AD0000 10.0.19041.662 MSCTF.dll \n0x00007FFF39850000 10.0.19041.662 AppXDeploymentClient.dll \n0x00007FFE80E80000 31.0.15.1640 nvoglv64.dll \n0x00007FFF3AE30000 10.0.19041.546 \n```\n\n\n - The conditions have been changed, but the error is still present. -----------------------------------------------------------", "Crash when rendering a specific 4k scene with Cycles Metal\nOperating system: macOS-13.3.1-arm64-arm-64bit 64 Bits\nGraphics card: Apple M2 Pro Apple 4.1 Metal - 83.1\n\nBroken: version: 3.5.1\n\nWhen I render with Blender 3.5.1 on a MacBook 16\" with an M2 Pro chip at a 4K quality, Blender closes before finishing the rendering process.\"\n\n- Open attached .blend file\n- Make sure you have enabled Metal GPU+CPU in the Preferences\n- Render the scene\n\nIt should crash about halfway through the rendering process.\n\n\n", "Crash clicking on Material Preview\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce RTX 2080 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 461.92\n\nBroken: version: 2.92.0\n\nSometime, clicking on Material Preview button (Eevee) Blender 2.92 crashes without a reason.\nSee file and log attached\nThank you\n[Rocks_Pack.crash.txt](Rocks_Pack.crash.txt)\n\n-\n\n[Rocks_Pack.blend](Rocks_Pack.blend)", "Blender Opens Window then Closes/crashes\nOperating system: Windows 10\nGraphics card: AMD FirePro W2100\n\nBroken: 3.5.1 2023-04-24 18:11 Hash:e1ccd9d4a123 Branch: Blender -v3.5-release\nWorked: first version used on this computer\n\n-First attempted to run blender, found error by running the debug, error:\n\n_EXCEPTION_ACCESS_VIOLATION_\n\n- Did specific driver updates for AMD FirePro W2100 and Intel CPU\n\n- Then ran dell command update\n\n- New error appeared after driver updates and running blender:\n\n_A graphics card and driver with support for OpenGL 3.3 or higher is required.\nInstalling the latest driver for your graphics card might resolve the issue._\n\n- Then downloaded GPU-Z to figure out OpenGL version\n- Open GL version was 1.1 when monitor DisplayPort was plugged into the motherboard\n- Open GL version was 4.5 when monitor DisplayPort was plugged into GPU\n\n- Now blender window would appear but closes immediately\n- running blender_debug_gpu_glitchworkaround.cmd in the Blender file worked to make Blender run but is NOT the overall fix\n- updated python\n- did windows updates\n\nunsure\n\n", "Scripting: undo after mesh.separate causes crash\nOperating system: Linux\n\nLatest tested: blender-3.5.0-beta+v35.f23e3c7f04f3-linux.x86_64-release\nEarliest tested: blender-3.1.2-linux-x64\n\nTrying to undo `bpy.ops.mesh.separate(type='MATERIAL')` in a script crashes Blender.\n\n1. Create a cube and two materials. Go to edit mode and assign some faces to both materials. Go back to object mode.\n\n \n\n2. Open the Text Editor and run this script\n\n ```py\n import bpy\n bpy.ops.ed.undo_push( message = \"temp\" )\n bpy.ops.mesh.separate( type = 'MATERIAL' )\n bpy.ops.ed.undo()\n ```\n\n3. Crash (segmentation fault).\n\nSomewhat mysteriously if I save just before running the script, close Blender, then reopen the .blend and then run the script, I get non-determinstic results (sometimes the undo does nothing, sometimes the white part is gone). The only consistent way I found to get a crash was to create the cube from scratch every time. So no .blend is attached to this issue.\n\n", "simulation in new scene causes crash in original scene's simulation\nOperating system: Windows-7-6.1.7601-SP1 64 Bits\nGraphics card: GeForce GTX 1080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 442.19\n\nBroken: version: 2.83 (sub 6)\nWorked: (optional)\n\n**steps**\n- make cube an active rigid body. \n- play simulation\n- then click on new scene button and on full copy.\n- in the new scene, rewind playhead to frame 0\n- run the simulation.\nBlender should crash.\n\nThere are more ways to cause crash when full scene copy is done (see comments), but I guess these will stem from single issue.", "Blender 2.8 and 2.81 (ab519b91b2c4) has high chance of crashing when using Python to create a new node-tree link.\nOperating system: Windows 10 (x64) v1903 Build 18362.356\nGraphics card: Nvidia GeForce GTX 1060 6GB\nCPU: Intel Core i5 7600K\n\nBroken: 2.81-ab519b91b2c4 **AND** 2.80 release\n\nCreating a new node-tree link via Python sometimes causes Blender to crash with the error \"EXCEPTION_ACCESS_VIOLATION\".\nI have attached a .blend file that will cause the crash, just by clicking \"Run Script\" once or twice. \nWhen I found the bug, I spent some time making the script as small as possible while still causing the crash, so the script is less than 30 lines.\n\nI also installed Blender 2.81-ab519b91b2c4 on a Windows 10 laptop that had never seen Blender before, and the bug was present on there too, so I'm confident the issue isn't specifically due to my hardware.\nI've also included blender_debug_output.txt and blender_system_info.txt, generated by the blender_debug_log.cmd file included in the Blender 2.81 download.\n\n1) Open the attached .blend file.\n2) Click \"Run Script\" in the bottom right of the text editor pane.\n3) Repeat 2 until crash (should only take 1 or 2 tries)\n\n[Crashy crash.blend](Crashy_crash.blend)\n[blender_system_info.txt](blender_system_info.txt)\n[blender_debug_output.txt](blender_debug_output.txt)", "Regression: Crash when Sculpting on a mesh from a GP Lineart Modifier\nOperating system: Windows-10-10.0.22621-SP0 64 Bits (W11)\nGraphics card: NVIDIA GeForce RTX 4090/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 531.68\n\nBroken: version: 3.2, 4.0.0 Alpha\nWorked: 3.1\n\n\nWhen trying to sculpt a mesh in an assigned collection that a grease pencil lineart modifier references and the layer it is drawing to contains Grease Pencil strokes - it will instantly crash Blender. \n\n\n1. Download and extract zip to get the test blend file\n2. Load attached file with Load UI off\n3. Select the bust, it is in a GP lineart modifier collection \n4. Go to sculpt mode\n5. Try sculpt\n6. Crash\n\nAlternative steps:\n1. Create new scene\n2. Create a sphere\n3. Create a Grease Pencil object\n4. Add sphere to a new collection\n5. Select the GP object and add a lineart modifier\n6. Reference the sphere collection in the modifier\n7. Assign layer and material to the modifier\n8. Draw in the same layer in Draw mode of the GP\n9. switch to sculpt mode\n10. Try sculpt over sphere surface\n11. Instant crash\n\n", "Blender 3.5.0 Win11 RTX3090 Fatal Crash on activating Cycles\n@deadpin thanks for replying.\n\nDriver Version: 31.0.15.3161 (NVIDA 528.24) DCH Win1164 installed\n\nCycles now working without issue thank u.\n\n_Originally posted by @Jaymi-McManus in /blender/blender/issues/106953#issuecomment-923094_\n\n>>>>>\n\nUpdate: \n\n1. Closed the ticket too soon, after adding some simple color, specular, metallic, changes in BSDF and then navigating between layout and other tabs, fatal crash occurs again\n\n2. Updated NVIDIA drivers to 31.0.15.3161 (NVIDIA 531.61) DHC / Win11 64 the issue still persists\n\nUpdate logs attached.", "crash freeze with ellie_animation.blend file\nOperating system: macOS-10.15.7-x86_64-i386-64bit 64 Bits\nGraphics card: AMD Radeon RX 580 OpenGL Engine ATI Technologies Inc. 4.1 ATI-3.10.23\n\nBroken: version: 3.2.0\nv3.0.1 also broken for this file\n\nwhen opening the demo file ellie_animation.blend the Blender crash freezes with spinning beachball.\n\nopen the demo file ellie_animation.blend \nBlender has spinning beachball\nmacOS Activity Monitor built-in utility shows Blender (Not responding). \nI have to force quit blender.\n\nother blender files I have work ok. other blender demo files work ok." ]

[ "Select object after adding armature causes immediate crash\nWindows 7 Pro 64bit, Intel HD Graphics [P3000](P3000.txt), driver version 9.17.10.3517 (current)\n\nBroken: 2.72b (both zip and installer)\nWorked: none?\n\nI noticed this on some 2.6X version which I uninstalled and then installed the latest (2.72b) and saw the same error. Uninstalled again and tried the zip instead of installer, same result. Updated the GPU driver, found it was already current.\n\nSteps:\n* Open blender (new scene)\n* Add Armature\n* Right-click new armature or the original block\n* Blender immediately closes\n\nOther notes:\n* Adding non-armature elements does not have the same effect.\n* The crash file (blender.crash.txt) contains one line - which indicates that the armature was created\n* If I go to User Preferences and change OpenGL \"Selection\" option from \"Automatic\" to \"OpenGL Occlusion Queries\" then this crash does not happen. If I change \"Automatic\" to \"OpenGL Select\" then it will crash as well." ]

[ "Huge memory usage when trying to paint on UVs far ouside the 0-1 range\nBroken: version: 2.82a (sub 7)\nWorked: Never\n\nBlender freezes when I attempt to use the paint options, after adding an image to Texture Paint. Memory will skyrocket from 150MB to 13GB and does not un-freeze.\nYoutube tutorial I'm following trying to learn to paint textures, bug happens after adding the image at 2:35, and I attempt to paint the model: watch?v=lmbAs9jE1vI\nI am very new to Blender, but pretty sure it shouldn't be doing that when attempting to use the Paint tools.\n\n\n\n- Open attached file\n- Try to paint the mesh\n\n[texture_paint_huge_memory_usage.blend](texture_paint_huge_memory_usage.blend)\n\n", "Regression: Importing of collada model with instances became a LOT slower since 2.80 (practically freezing)\nOperating system: windows 10\nGraphics card: geforce gtx 1070\n\nBroken: blender 3.4.1\nWorked: 2.80\n\nImporting of collada model makes Blender not respond anymore with high CPU usage in windows task manager (and never finishes no matter how long you wait)\n\n- import attached collada file\n- Blender not respond anymore with high CPU usage in windows task manager (and never finishes no matter how long you wait)\n\nLink to model file:\n\nCruisn-Exotica-arcade-game-Dual\n\n[doubleexotica.zip](doubleexotica.zip)", "Shader compilation freeze Blender on Intel Arc770\nOperating system: Windows 10\nGraphics card: Intel Arc770 16Gb (Resizable Bar enabled)\n\n3.6.0\nDate: 2023-06-27 08:08\nHash: c7fc78b81ecb\nBranch: blender-v3.6-release\n\n**Intel driver version**\n31.0.101.4665\nReleased 11.08.2023 (latest on moment of this post)\n\nAll modes of viewport on new scene (default, for example) works fine: solid, material preview, display render. But, if I open my scene, works only 1 and 3 option. Building a shaders freezes Blender. It's around 800+ materials on scene, and GeForce cards works fine with that. The video card works at full power, with enabled Resizable Bar function. Cycles render (oneAPI) also works fine, but i can't switch the viewport on material preview. Blender freezes when shaders start compile.\n\n", "Grease Pencil: unable to change erase size while using the Draw tool\nOperating system: Linux-5.19.0-76051900-generic-x86_64-with-glibc2.35 64 Bits\nGraphics card: AMD Radeon RX 5700 XT (navi10, LLVM 13.0.1, DRM 3.47, 5.19.0-76051900-generic) AMD 4.6 (Core Profile) Mesa 22.0.5\n\nBroken: version: 3.5.0 Alpha\nWorked: ???\n\nCan't change the size of the eraser brush while erasing while using the draw tool (`Ctrl + LMB`), I can do it with the scroll wheel on the mouse but not with the shortcuts assign for it `[]`\nAlso applies to annotation tool.\n\n- go to greasepencil draw mode\n- draw something\n- hold control to erase\n- try to change the size of the eraser brush while erasing (Either scroll wheel or brackets)", "Subdiv hangs blender\n**System Information** DESKTOP-HR30UII\nOperating system:Windows 11 Home Single Language\nGraphics card:2GB NVIDIA GRAPHICS DRIVER\n\n\nBroken: Blender Version 3.5\nWorked: ?\n\nBLENDER 3.5 KEEPS NOT RESPONDING WHILE I USE MODIFIERS \n\nON BLENDER ADD MESH AND ADD MODIFIERS TO SUBDIVISION SURFACE THEN CHANGING THE VALUE OF LEVELS VIEWPORT VALUE ABOVE 4 .THEN ONWARDS BLENDER COMPLETELY STOPS WORKING AND I NEED TO RESTART IT AGAIN\n\n", "Blender crashes when using temp_override on 'UV' area with transform.translate, transform.resize\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 531.41\n\nBroken: version: 3.3.5\nBroken: version: 3.4.1\nBroken: version: 3.5.0\nWorked: none\n\nWhen executing `transform.resize` or `transform.translate` with `context.temp_override(area=uv_area)` Blender crashes. This happens when the area.ui_type is UV, with VIEW_3D is works fine.\n```\nError : EXCEPTION_ACCESS_VIOLATION\nAddress : 0x00007FF72DEF3C80\nModule : blender.exe\nThread : 000009dc\n```\nthe complete crash file is attached\n\n1. create a new Text and paste the below code into the text (the attached file issue.blend can also be used)\n```py\nimport bpy \ncontext = bpy.context\narea = context.area\narea.ui_type = \"UV\"\nwith context.temp_override(area=area):\n bpy.ops.transform.resize(value=(0.5, 0.5, 0.5), orient_type='GLOBAL', orient_matrix=((1, 0, 0), (0, 1, 0), (0, 0, 1)), orient_matrix_type='GLOBAL', mirror=False, use_proportional_edit=False, proportional_edit_falloff='SMOOTH', proportional_size=1, use_proportional_connected=False, use_proportional_projected=False, snap=False, snap_elements={'INCREMENT'}, use_snap_project=False, snap_target='CLOSEST', use_snap_self=True, use_snap_edit=True, use_snap_nonedit=True, use_snap_selectable=False)\n bpy.ops.transform.translate(value=(0.25, -0.25, 0), orient_axis_ortho='X', orient_type='GLOBAL', orient_matrix=((1, 0, 0), (0, 1, 0), (0, 0, 1)), orient_matrix_type='GLOBAL', mirror=False, use_proportional_edit=False, proportional_edit_falloff='SMOOTH', proportional_size=1, use_proportional_connected=False, use_proportional_projected=False, snap=False, snap_elements={'INCREMENT'}, use_snap_project=False, snap_target='CLOSEST', use_snap_self=True, use_snap_edit=True, use_snap_nonedit=True, use_snap_selectable=False)\n```\n2. run this text, Blender will now crash\n\n**Note**: not using `temp_override()` and executing these operators in the current context with the ui_type set to UV works fine. Like this: \n```py\nimport bpy \ncontext = bpy.context\narea = context.area\narea.ui_type = \"UV\"\nbpy.ops.transform.resize(value=(0.5, 0.5, 0.5), orient_type='GLOBAL', orient_matrix=((1, 0, 0), (0, 1, 0), (0, 0, 1)), orient_matrix_type='GLOBAL', mirror=False, use_proportional_edit=False, proportional_edit_falloff='SMOOTH', proportional_size=1, use_proportional_connected=False, use_proportional_projected=False, snap=False, snap_elements={'INCREMENT'}, use_snap_project=False, snap_target='CLOSEST', use_snap_self=True, use_snap_edit=True, use_snap_nonedit=True, use_snap_selectable=False)\nbpy.ops.transform.translate(value=(0.25, -0.25, 0), orient_axis_ortho='X', orient_type='GLOBAL', orient_matrix=((1, 0, 0), (0, 1, 0), (0, 0, 1)), orient_matrix_type='GLOBAL', mirror=False, use_proportional_edit=False, proportional_edit_falloff='SMOOTH', proportional_size=1, use_proportional_connected=False, use_proportional_projected=False, snap=False, snap_elements={'INCREMENT'}, use_snap_project=False, snap_target='CLOSEST', use_snap_self=True, use_snap_edit=True, use_snap_nonedit=True, use_snap_selectable=False)\n```\n\n", "CPU render will look different to GPU render and fails randomly\nWin 10, GTX 880M, GTX 1080TI Founders Edition\n\nBroken: blender-2.79.0-git.2ca1f29-windows64\nWorked: na\n\nAttached you'll finde a scene what has issues with GPU rendering. The scene is using the old shaders with a bevel shader (OSL).\n\nThe preview will not start or maybe it takes a very long time. Here nothing happened.\nIn GPU mode the preview starts as expected.\n\nWhen I try to render with F12 in CPU mode it doesn't start or maybe it will start after a long time. Here nothing happened after 10 minutes.\nWhen I render a small region with CPU it takes a long time till it starts to render.\nWhen I start rendering in GPU mode it starts fast as expected ignoring OSL of course.\n\nIn the ZIP and attached you'll see a picture with a split screen. Left side is GPU. Right CPU. The results are different. Transparent parts with DOF will have extreme noise.\n\nLoad the scene and press F12. Here CPU render will not start as expected.\nSwitch to GPU and press F12. Rendering will start.\nDefine a small region where the transparent parts are. Press F12. It will take really long but start. Result has noise at the transparent parts.\nThis might be a bug in the Cycles render engine or an issue with the shader tree.\n\n[170902_Blender_279_Cycles_CPU_Issue.zip](170902_Blender_279_Cycles_CPU_Issue.zip)\n\n\n\n", "Blender Wont run.\nOperating system: Windows\nGraphics card: AMD Radeon HD 8570\n\nBroken: 2.92\nWorked: Nothing has worked. But blender worked before I did this. \n\nWhen I open blender it closes immediately.\nAnd I get this \n\nwatch?v=3rugIsYwl7s).\nSo Out internet was down, And I wanted to get blender On my other pc account, So I went to the main account And Thinking that the installer required internet, I took the system files, and copied them to a thumb drive and pasted them Into the other account when I did that this first started happening, so when The internet came back On. I just got rid of the files I pasted in and then Installed it. With the same result as last time. But this time the bug/glitch was on all accounts. and I tried uninstalling and reinstalling it from the apps and features and from the control panel. I also installed the app from the windows app store with no luck. If you guys know how to fix this issue then please email me at [email protected]\n Ps, My CPU wasn't hot.\n- - -\nAble to run blender from `blender_debug_gpu_glitchworkaround.cmd` batch file", "Crashing and freezing when working on files with assets\nOperating system: macOS-12.6.5-x86_64-i386-64bit 64 Bits\nGraphics card: Metal API AMD Radeon RX 580 1.2\n\nBroken: version: 3.5.0\n\nCrashing and freezing when working on files with assets\n\nif I open the demo file cube_diorama.blend the whole system freezes and either crashes or logs out needing to login again.\nevery blender crash a WindowsServer log is created.\n\n cube_diorama.blend never crashed in previous versions of blender for me.\n\n", "Failed to load OptiX kernel OPTIX_ERROR_DISK_CACHE_INVALID_DATA\nOperating system: Debian Testing\nGraphics card: RTX 3090 + RTX 3050 Ti (laptop)\n\n\nBroken: 3.3.2\n\nI recently bought external RTX 3090 videocard, and when i switched Cycles to RTX 3090, it says:\n```\nFailed to load OptiX kernel from '/home/scaled/git/blender-git/build_linux_release/bin/3.3/scripts/addons/cycles/lib/kernel_optix.ptx' (OPTIX_ERROR_DISK_CACHE_INVALID_DATA)\n```\n\nI hide directory /var/tmp/OptixCache_scaled and videocard started normally. But when i switched back to my laptop GPU (RTX 3050 Ti), it said same error. I returned my old cache directory back (/var/tmp/OptixCache_scaled) and it started normally.\n\nThen i decided to enable both videocards at the same time in hope it will generate cache for both videocards, so i will not have to move cache directories around every time i go outside with my laptop, but i got same error, and cache generated only for first videocard in the list (RTX 3090).\n\nBy the way, when i tested GTX 1080 in this external box, it don't created cache problems, so i think it happens when using different videocards of one generation.\n\n1. Setup Cycles to use Optix with one videocard.\n2. Turn on Cycles renderer and wait until cache is generated.\n3. Turn off Cycles and select second videocard for Cycles.\n4. Cycles will not turn on because cache conflict.", "Weight Paint gradient operators wait for mouse input\nOperating system: Linux-6.4.6-gentoo-x86_64-AMD_Ryzen_Threadripper_1950X_16-Core_Processor-with-glibc2.37 64 Bits, X11 UI\nGraphics card: AMD Radeon RX 7900 XT (gfx1100, LLVM 15.0.7, DRM 3.52, 6.4.6-gentoo) AMD 4.6 (Core Profile) Mesa 23.3.0-devel (git-50c29e1ffa)\n\nBroken: version: 4.0.0 Alpha\nWorked: --\n\nThis is based on feedback from #108791\nWhen using the gradient operators in weight paint mode, they don't start on shortcut press but require a mouse press to start and release to confirm.\n\nThis worked well with the previous default shortcut with mouse clicks but when reassigned to keyboard shortcuts this is very inconsistent with other operators in Blender and seems unintentional or outdated.\n\nA good counter example for a better behavior is \"Expand by Topology\" in sculpt mode, which only waits for mouse input if started from a menu. If executed with the keyboad shortcut `Shift A` it will start and then confirm with `LMB`.\nIf the shortcut is remapped to a mouse press, it will start on press and confirm with release. \n\n- Enter Weight Paint Mode\n- Press `Shift A` and use the operation\n- Try the same in Sculpt Mode\n\n", "2.90 Not opening on Linux\nOperating system: Linux Mint 18.1\nGraphics card: Intel HD 620\n\nBroken: \nBlender 2.90 won't open up fully, it appears to open up and immediately closes\nWorked: \n\nTried a few different ways to open, won't open.\n\nOne of the methods I tried opening with was through the Terminal and this was the message I had returned back:\n\nRead prefs: /home/john/.config/blender/2.90/config/userpref.blend\nfound bundled python: /home/john/Desktop/3D Related/1BlenderVersions/blender-2.90.0-linux64/2.90/python\nGPUShader: linking error:\nerror: Input block `ShaderStageInterface' is not an output of the previous stage\n\nGPUShader: linking error:\nerror: Input block `ShaderStageInterface' is not an output of the previous stage\n\nWriting: /tmp/blender.crash.txt\nSegmentation fault", "Entire system freezing while using NLA in Blender 2.91 and newer (Mesa only?)\nOperating system: Fedora 33\nGraphics card: Intel HD Graphics 520 (Mesa 20.2.4)\n\nBroken: 2.91\nWorked: 2.90.1\n\nBlender randomly freeze entire graphics system. After freezing, last 2 frames is flickering if you move mouse. I can even save my work with CTRL+S, but after that i need to restart my computer. Setting INTEL_DEBUG=reemit environment variable fixes the issue.\n\nMethod 1\nAdd a few keyframes.\nScroll up and down in the NLA editor.\n[blender-freeze-2.mp4](blender-freeze-2.mp4)\n\nMethod 2\n\nJust do something, even in Solid view, even in Video Sequence editor. I don't know how to reproduce, it happens after 10 minutes of work. Also i notice some glitches in VSE with image strips. Also bug with Depth of Field was fixed in this version, probably this fix relate to this bug.\n\nBug with depth of view that was fixed:\nT77828\n\nIssue with VSE\n[blender-vse-artifacts-1.webm](blender-vse-artifacts-1.webm)", "Shading change (smooth/flat) resets current simulation cache. \nOperating system: Windows-7-6.1.7601-SP1 64 Bits\nGraphics card: GeForce GTX 660 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 442.19\n\nBroken: version: 2.83 (sub 5)\nversion: 2.82 (sub 7)\nWorked: 2.79\n\n[2020-02-26_04-40-40.mp4](2020-02-26_04-40-40.mp4)\n\nSet simple simulation, play it (to the end or pause), change shading to smooth/flat. Your cache lost.\nHere file.\n[simeset.blend](simeset.blend)\n\n", "Blender just crashes for no reason, and the crash report is empty. (Segmentation fault)\nOperating system: Arch Linux x86_64\nGraphics card: AMD Radeon RX 6900 XT\n\nBroken: (example: 3.6.2, ffe93138751e, master, 2023-08-31, as found on the splash screen)\nWorked: (No idea when this error started happening.)\n\nBlender does not start up. When I start it via the command line, I get this error:\n```\nWriting: /tmp/blender.crash.txt\nSegmentation fault (core dumped)\n```\n\n`/tmp/blender.crash.txt`:\n```\n# Blender 3.6.2, Commit date: 2023-08-14 14:05, Hash ffe93138751e\n\n# backtrace\n\n# Python backtrace\n```\n\nTrying to open Blender.\n\n" ]

[ "Blender Freezes when using the 3d Scale Gizmo (Ubuntu 22.04)\nOperating system: Pop!_OS 22.04 LTS\nGraphics card: Intel Corporation Iris Pro Graphics 580\n\nBroken: 3.2, 3.3, 3.4\nWorked: 3.1\n\nWhen Gizmo is activated to manipulate an object, everything is frozen.\n\nI add the default cube to the scene.\nActivate move, rotate, scale parameter for the Gizmo.\nEverything is frozen." ]

[ "KeyingScreen Compositor node emits \"black holes\" \nThe attacged folder shows a bug in the KeyingScreen compositor node.\n\nIn the Screenshot GreenPlateBug-Screenshot.png it can be seen in the above compositor window that some areas are filled with black\ninstead of grey / green.\n\nThat appeared after creating the \"last\" track marker (the selected / larger track marker in the below window).\n\nIf you remove that larger marker, the bug disappears.\n\nTo reproduce, open the attached blend file, close render window, disconnect / reconnect the viewer node to the keying screen node to\nsee in backdrop display.\n\nThe bug appeared in a different usage of the KeyingScreen node with much fewer track markers, but since the circumstances of\nthe bug to appear is unclear, i had to add a lot marker here.\n\n\nBlender version 2.80 (snap Ubuntu version).\nOS: Ubuntu 19.04 x86\nGraphics Card: Prime Laptop (NVidia / Intel combination) Dell Inspirion 4500 Laptop in Intel mode\n", "Some hotkeys stop functioning after conforming Chinese/Japanese text input\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 980/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 471.68\n\nBroken: version: 3.0.0 Alpha\nWorked: 2.93.4\n\nSome hotkeys stop functioning after conforming Chinese/Japanese text input.In my case,the pie menu on {key ~}{key ,}{key .} are disabled, though if after input,switch to another input method, these hotkeys become available again.\nI‘m using **Microsoft Pinyin** input method on Win10.\n\n**Exact steps for others to reproduce the error==\n{icon fa-plus} OS Conditon: **WIN 10** |Blender: **3.0 Alpha**\n- Install **Microsoft Pinyin** input method:\n\n- Default setup.\n- Rename the cube in Chinese/Japanese.\n- Test the hotkey {key ~}{key ,}{key .}, they shouldn't be functioning.\n- Change the input method to something else, test {key ~}{key ,}{key .}, they should be functioning again.\n- Input some Chinese/Japanese in the text field, test {key ~}{key ,}{key .}, they shouldn't be functioning again.\n\n", "Regression: Crash when sculpting shape keys in 3.6.2 release\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 536.99\n\nBroken: version: 3.6.2\nWorked: 3.5 did not crash, I'm told.\n\n**Description**\nA user in a Discord community shared a file which was crashing for them while in Sculpt mode. I could reproduce the crash, so I simplified the file and found repro steps.\n<video src=\"attachment\" title=\"2023-09-14 00-35-34.mp4\" controls></video>\n\nAs you can see, it takes a few seconds of aggressively dragging the shape key value to get the crash to occur.\nAnother way to get the crash is to just make sculpt brush strokes. I found the shape key slider dragging easier, but I could get it to crash with sculpt strokes too, which was the original complaint from the user.\nMaybe worth noting that, as far as I can tell, if the shape key is not this extreme in its effect, then it seems even harder or impossible to produce the crash.\n\n- File: [Vessa_clean5.blend](attachment)\n- Make sure you're in sculpt mode (I don't think it can crash otherwise)\n- Drag the shape key value around for several seconds, or make a few dozen sculpt strokes\n- Should crash eventually.\n\n", "unable to use 'grid fill' or 'fill' on screwed object\nOperating system: Window 10\nGraphics card: NVIDIA GeForce RTX 2060 SUPER\n\nBroken: Blender 2.83\nWorked: Blender 2.76 \n\n\nI am trying to create a solid spring using Screw modifier. After applied grid fill, I expected a full capped end of the screwed object, but it turns out to be like figure below. The complete file also attached in this post. Hope to get your response soon. \n\n[screw.blend](screw.blend)\n\nBased on the default startup or an attached .blend file (as simple as possible).", "Blender 2.91 alpha - Debug menu, instant crash\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX TITAN X/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.55\n\nBroken: version: 2.91.0 Alpha\nWorked: 2.90.1\n\nShader compilation is causing an error when the debug menu is invoked on a blank scene or working scene.\n\nOpen the .blend attached file, press F3>search: Debug. Type 22 (eevee GPU speed benchmarking verbose),\nBlender will crash immediately. The same is true on a blank new scene in the general application template.\n\nFile to test: \n44cb6eb461af31ad1209dedb738d76c0.blend\n\n", "Many Blender's keyboard shortcuts do not work on Russian keyboard layout\nOperating system: Linux (elementary OS 5.0, Ubuntu-18.04-based) \nGraphics card: NVIDIA GTX 1080Ti\n\n2.80 (sub 60), branch: blender2.7, commit date: 2019-04-30 22:49, hash: 480a09a92f7f, type: Release build date: 2019-04-30, 23:02:55\n\nThere are two keyboard layouts on my system: English (US, primary) and Russian (RU, secondary).\nWhen Russian layout is selected, only some of Blender's keyboard shortcuts work as expected, namely:\n\nBlender 2.8 keyboard shortcuts that **work correctly** on Russian keyboard layout:\n- All F keys (F1-F12),\n- All top row Numeric keys (1, 2, 3 ...),\n- All Numpad keys,\n- All Ctrl+any single key combos, e.g. Ctrl+B, Ctrl+S...\n\nBlender 2.8 keyboard shortcuts that **do not work (=have no effect whatsoever)** on Russian keyboard layout:\n- All single-letter keys (e.g., B, A, G, S, R...)\n- All Alt+any single key combos, e.g. Alt+B, Alt+A...\n- All Shift+any single key combos, e.g. Shift+V, Shift+D...\n\n1. Add a secondary (Russian) keyboard layout to your system.\n2. Switch your keyboard layout to Russian.\n3. Launch Blender 2.8 beta.\n4. Select a default primitive object (e.g., a cube) \n5. Try to modify the object using any single-letter keyboard shortcuts. e.g., move (G), scale (S), extrude (E), inner extrude (I), etc. or Alt+any letter, or or Shift+any letter.\n\n[system-info.txt](system-info.txt)\n", "Regression from 3.3: Invalid button mapping on SpaceMouse Enterprise (256f:c633)\nButton mapping on this device is now completely scrambled. For example pressing \"Shift\" deletes selected object, \"Lock rotation\" is interpreted as play/pause, \"Button 6\" rotates camera 90deg CCW and \"1\" brings up the menu. Other buttons do not seem to register.\n\n\nBus 003 Device 013: ID 256f:c633 3Dconnexion SpaceMouse Enterprise\n\nTested on macOS Ventura and Ubuntu 22.04\n\nIn blender 3.3.2 on Linux and macOS, some buttons function correctly:\n - Shift\n - Ctrl\n - Alt\n - Rotate\n - Top\n - Front\n - Right\n\n Basically all other keys in 3.3.2 bring up the on-screen menu.\n\n\nTested versions: 3.4, 3.5 and 3.6.\n\nOperating system: macOS-14.0-arm64-arm-64bit 64 Bits\nGraphics card: Metal API Apple M1 Pro 1.2\n\nBroken: version: 3.6.0\nWorked: 3.3.2\n", "GHOST/Wayland Support\nGHOST/Wayland support has been committed to master and is enabled for daily builds, this task is an overview of the remaining TODO's. \n\nNote that these could be expanded into their own tasks, listing here for now.\n\n**Library Linking**\n\n- [x] dynamically load libwayland (e58e023e1a)\n\n**Integration**\n\n- [ ] Support display server reconnecting (new feature, not yet supported by main-line MESA), see [KWIN/Restarting](Restarting)\n- [ ] IME (text input) support (see: [text-input-unstable-v3](text-input-unstable-v3)), See: [D17030: [WIP] initial GHOST/Wayland IME support](D17030)\n- [x] Primary clipboard support b0eff51fb7\n\n\n**Input Devices**\n\n* [x] #104272 Touch-pad support for detecting inverted (natural) scroll events 98272e5bb24b865380ad580ce0354ccacb0fa52d (depends on Wayland v1.22 release).\n- [x] Tablet support (pressure sensitivity, angle), 97f894881c\n- [x] NDOF (3D mouse) support 8841bd9660 \n- [x] Detectable key repeat events (`GHOST_TEventKeyData.is_repeat`).\n- [x] #98793 (Wayland clamps cursor movement failes with gnome-shell (fixed up-stream)), up-stream bug in gnome-shell, worked around in Blender.\n- [x] #76429 (GHOST/Wayland event's don't correspond to physical keys)\n- [x] #77311 (Mouse cursor stays in-place with Ghost/Wayland for operators that grab cursor input)\n- [x] High resolution track-pad scroll events 30d9a6245f\n- [x] Touch-pad gesture support (zoom, swipe, rotate.. etc). *lower priority, not supported by Blender under X11.* 8bb211a771\n\n\n**Display**\n\n- [x] Support for DPI detection (`GHOST_GetDPIHint`) 2770010224\n- [x] Support fractional scaling (`fractional-scale-v1`) cde99075e87032d99d986182e98d9c367b22d417\n\n**Windowing**\n\n- [x] Support activating windows 3071ec486bffd0fef180f221a85f0b15f07b7950\n- [x] Support dialog windows (so they're always in-front `GHOST_CreateDialogWindow`, respecting parent windows).\n- [x] #98612 (Window Decorations Not Present on GNOME - Wayland) 29755e1df8\n\n**OpenGL**\n\n- [x] Switch from GLEW to libepoxy, to support dynamic loading of either `libOpenGL.so` or `libGL.so`, patch: [D15291](D15291) (see [D12034](D12034)#312834).\n- [x] #98462 (Save Screenshot (glReadPixels) fails under Wayland)\n- [ ] Enable EGL by default for X11 (This is trivial at this point. Just remove the `USE_EGL` ifdefs in `GHOST_SystemX11.cpp` and `GHOST_WindowX11.cpp`)\n- [ ] Support `libOpenGL.so` and EGL in OpenSubdiv and USD `glLoader`\n\n\n----\n\n**Limitations**\n\nWayland has constraints that might not be practical to work around.\n\n- [ ] #102985 (Wayland: Unable to lower windows)\n- [ ] #98928 (Wayland: Unable to access window position, finding the window under the cursor doesn't work)\n- [ ] #100276 (Wayland: Windows reset when reloaded on a multi-monitor configuration)\n- [ ] Window decorations look out of place (libdecor) see, #101176 and this [merge-request ](43).\n\n**Bugs**\n\n- [x] #100855 (GHOST/Wayland: Terminal Ctrl-C closes the window while Python is running) *Low priority, needs further investigation.*\n- [ ] #103059 (GHOST/Wayland: Huge cursor when using a graphics tablet in KDE) *Low priority, needs further investigation.*\n- [ ] #102999 ([Wayland] Incorrect UI scaling on high-resolution screen under GNOME 42.5 w/ fractional scaling) *Fixed in GNOME 43, not clear how Blender would workaround the bug.*\n\n**Bugs with LibDecor**\n\n- [ ] #109194 Window decoration size doesn't match Blender window size.\n- [ ] #108308 Crash on startup with LibDecor on Wayland.\n- [ ] #107797 Resizing window causes crash on Wayland.\n\n**Bugs with GPU Drivers**\n\n- NVidia\n\n - #110327 Only 3D View-port region is transparent on Wayland (NVidia + Nouveau)\n\n**Bugs with Other Components**\n\nHere are issues that are noticeable with Blender.\n\n- [ ] Gnome-shell doesn't scale custom cursors with Hi-DPI displays: 1215\n- [ ] Gnome-shell doesn't receive tablet input for new windows #107778 2658\n- [ ] #98927 (Cursor input \"low resolution\" with Hi-DPI in gnome-shell), reported to mutter: 2311", "Text Object: D key is used for the Annotate Tool\nOperating system: Linux-5.4.0-104-generic-x86_64-with-glibc2.31 64 Bits\nGraphics card: Quadro GP100/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 470.103.01\n\nBroken: version: 3.2.0 Alpha\n\nIn the keymap preferences there is an option to use the Active Tools for shortcuts instead of the modal operation.\nIt's called \"Tool Keys -> Immediate / Active Tools\".\n\nIf this preference is used and you go into edit mode of a text object, you are unable to type in the letter `D`.\nIt will always make the Annotate Tool active instead.\n\n- Enable the \"Active Tool\" keymap preference\n- Add a text obejct\n- Enter Edit Mode\n- Type in a word with the letter `D` in it like \"Addon\" ", "Shape keys missing from `bpy.data.meshes.new_from_object()` even when the mesh Object is not evaluated\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 531.41\n\nBroken: version: 4.0.0 Alpha\nWorked: 2.79 (sort of, the API was a bit different due to the depsgraph not existing)\n\nCreating a new Mesh from a mesh Object with `bpy.data.meshes.new_from_object` removes the shape keys (or simply does not copy the them in the first place), even when the Object is not evaluated.\n\nThe API in 2.79 is a little different to 2.80+, since there is no depsgraph; you choose with the function arguments whether or not to apply modifiers:\nWhen using `apply_modifiers=True`, the new mesh has no shape keys as expected. Providing an evaluated Object to the function in 2.80+ is equivalent to this.\nWhen using `apply_modifiers=False`, the new mesh retains its shape keys as expected. In my opinion, providing a non-evaluated Object to the function in 2.80+ should be equivalent to this.\n\n2.79 API docs: bpy.types.BlendDataMeshes.html#bpy.types.BlendDataMeshes.new_from_object\nCurrent API docs: bpy.types.BlendDataMeshes.html#bpy.types.BlendDataMeshes.new_from_object\n\nSince the API changed between 2.79 and 2.80, it's not clear whether this is a bug or intended behaviour.\n`Object.to_mesh()` and `BlendDataMeshes.new_from_object()` appear to have identical behaviour in 2.79, though this is again a bit different in 2.80+ because `Object.to_mesh()` now creates a temporary mesh owned by the Object that is not added to `bpy.data.meshes`.\n\n1. Create a mesh Object\n1. Add a shape key (other than the 'Basis')\n1. Enter Edit Mode\n1. Modify the shape key so that it noticeably moves the mesh\n1. Exit Edit Mode\n1. Change the value of the shape key to 1.0\n1. Run in the Python Console:\n 1. `new_mesh = D.meshes.new_from_object(C.object, preserve_all_data_layers=True, depsgraph=C.evaluated_depsgraph_get())`\n 1. `C.object.data = new_mesh`\n1. Observe that the new mesh has no shape keys (and is not evaluated as expected)\n\nAttached are some .blend files created in different Blender versions that have a script and three Objects that go over different cases (has modifiers/no modifiers/has modifiers but they're disabled) of creating meshes with both `BlendDataMeshes.new_from_object` and `Object.to_mesh` as well as making a copy of an evaluated mesh with `Object.copy` in 2.80+.\n\nThe 2.79 version is a bit different due to the different API, but the 2.80+ version in 4.0.0a results in this:\n(the .to_mesh().copy() behaviour of evaluated Objects has been reported separately as 109327)\n\n\n", "Cannot input non-ASCII characters with third-party software on Windows\nOperating system: Windows 10\nGraphics card: any\n\nBroken: master\nWorked: n/a\n\nBecause Blender exclusively relies on WM_INPUT for text input and ignores WM_CHAR, any characters provided by third-party software are silently ignored. Such software includes [AutoHotKey ]] or [[ http:*wincompose.info/ | WinCompose ](https:*www.autohotkey.com/).\n\n\nSteps to reproduce with AutoHotKey:\n * Install AutoHotKey\n * Create a simple hotkey in `%USERPROFILE%/Documents/AutoHotkey.ahk` containing non-ASCII characters, for instance: `^!p::Send déjà-vu`\n * Run AutoHotKey; it will automatically register the hotkey for Ctrl-Alt-P\n * Observe that Ctrl-Alt-P in *e.g.* Notepad will write `déjà-vu`.\n * Observe that Ctrl-Alt-P in a Blender text control will write `dj-vu`.\n\nSteps to reproduce with WinCompose:\n * Install WinCompose; it will automatically run and use Right Alt as the compose key.\n * Observe that Right Alt followed by A E in *e.g.* Notepad will write `Æ`.\n * Observe that Right Alt followed by A E in a Blender text control will write nothing.\n", "input() raises error in python console\nWindows 7 64bits / ASUS EAH6670\n\nOfficial 2.70a 64bits\n\nThe input function doesnt work in the python console.\n\n```\nimport bpy\nfrom bpy import data as D\nfrom bpy import context as C\nfrom mathutils import *\nfrom math import *\n\n```\n - ~ PYTHON INTERACTIVE CONSOLE 3.3.0 (default, Nov 26 2012, 15:38:36) [MSC v.1500 64 bit (AMD64)]\n - ~ \n - ~ Command History: Up/Down Arrow\n - ~ Cursor: Left/Right Home/End\n - ~ Remove: Backspace/Delete\n - ~ Execute: Enter\n - ~ Autocomplete: Ctrl-Space\n - ~ Zoom: Ctrl +/-, Ctrl-Wheel\n - ~ Builtin Modules: bpy, bpy.data, bpy.ops, bpy.props, bpy.types, bpy.context, bpy.utils, bgl, blf, mathutils\n - ~ Convenience Imports: from mathutils import *; from math import *\n - ~ Convenience Variables: C = bpy.context, D = bpy.data\n - ~ \n```\ninput()\n```\n - ! Traceback (most recent call last):\n - ! File \"<blender_console>\", line 1, in <module>\n - ! RuntimeError: input(): lost sys.stdin\n - ! \n\n\n\n\nIn python console, type\n\n```\ninput()\n```\n\n", "Current value 0 matches no enum in error with creating a new keymap from Industry Compatible or Blender 27x keymap\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 1060 6GB/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 461.09\n\nBroken: version: 3.1.0 Alpha\n\nWhen you create a new keymap from the industry compatible keymap or the Blender 2.7x keymap, then you get errors in the console about this and that operator does not match current value 0 in enum ...\n\nRelated task: T92754\n\n\n\n- Open Blender\n- Open Preferences\n- Switch to Industry compatible Keymap\n- Create a new keymap\n- Watch the console\n- Repeat with the Blender 27x keymap as the base ...\n\n\n\n\n\n\n\nEDIT, the params.action_mouse parameter could be the cause.\n\n\n\n\n\n", "GPencil: Shortcuts in Draw Mode are hardcoded\nMultiple keys in the `gpencil_draw_modal` function are hardcoded and (some) don't show up in the info bar. This makes it impossible for the user to know what certain keys do as well as impossible to change them in the keymap. All of those keys should be in the keymap with the default keys set as they are now.\n\nHere is a list of these (hardcoded) shortcuts:\n\n - {key Enter}, {key Pad_Enter}, {key Esc}, {key Space}, {key Ctrl Z} (?): Confirm\n - {key Mousewheel_Up}, {key Pad_Plus}: Increase size of eraser\n - {key Mousewheel_Down}, {key Pad_Minus}: Decrease size of eraser\n - {key O}: Set center of guide\n - {key V}: Turn off speed guide\n - {key M}: Alternate or flip guide direction\n - {key L}: Turn on line guide; {key Ctrl L}: Turn on line guide with angle = 0; {key Alt L}: Turn on line guide with last angle\n - {key C}: Turn on circular guide\n - {key K}: Turn line guide clockwise\n - {key J}: Turn line guide anti-clockwise\n", "Insert key dropdown on keymap doesn't show keying sets anymore\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2070 with Max-Q Design/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 466.77\n\nBroken: version: 2.93.2 Release Candidate\nWorked: 2.92\n\nwhen trying to assign a default keying set to the insert key menu, it doesnt show the list anymore just a dropdown that says default\n\n2.92\n\n\n\n2.93+\n\n\nopen blender, go to settings > keymap\nsearch for \"insert keyframe\"\nexpand the \"insert keyframe menu\" under pose mode\nclick the keyingset dropdown... all you get is default.\n" ]

[ "Dead Keys failing (accented letters, double strokes)\nOperating system: Windows 10 X64 Enterprise\nGraphics card: Nvidia RTX 2070 MaxQ\n\nBroken: 2.73, 2.91.0\nWorked: Never\n\nIn a text editor window, using Unicode national characters (Greek in my case), there are accented vowels that register with two keys from the keyboard (eg: ά, έ, ό … etc).\nIf I type the 2 key sequence fast, the accent is missed, slower works. Is this a delay that can be set in Blender, or an interpreter lag that cannot be adjusted?\nActually I think it happens everywhere (not only text editor), but it’s especially nagging problem in a text window, where I may type whole paragraphs and almost each word has an accent.\n\n- In the text editor, press and hold the dead key\n- without releasing the dead key, press the key to be transformed. (this will fail)\n\nIf you release the dead key before pressing the key to be changed, it will work normally." ]

[ "Exact Boolean with Incorrect Result - multiprecision/float round trips not exact\nOperating system: Windows 64 bit\nGraphics card: Nvidia GTX 750 Ti\n\nBroken: 2.91.0 Beta dca36a8ec922\n\nAnother case, where the exact boolean does not work as expected. (Sorry @howardt, I had to break it. I have to admit, that I have not run into this problem in a real scene.)\nAgain, this manifests after applying the same difference operation several times.\n\n[boolean_rotated_box3.blend](boolean_rotated_box3.blend)\nLook at the scene.\nThe first run of the modifier works as expected, the second creates overlapping faces, the third removes most of the mesh.\nThis \"bug\" is very erratic, and if the transform of the objects is slightly changed, it does not occur.\n\nThis is what I see:\n\n\nThe expected behavior is that the second and third modifier do nothing.", "Particles reset their position when using time stretching\nMinimal reproduction steps:\n- Add a particle system to the default cube and make it renderable.\n- Set the old time stretching setting to 20 and leave the new at 100.\n- Play the animation until frame 90 and render.\n- Find that the particles are sticking on the object as opposed to their correct position in the viewport.\n- Render again. They now have their correct position.\n- Play another 5 frames.\n- Render again, they are now sticking to the object again.\n\n[timeStretchingBadRender.blend](timeStretchingBadRender.blend)\n\n**Original Report**\n\nHello,\n\ni am using blender-3.1.0-linux-x64 and working on particles with time stretching but the viewport display is correct.\n\nattached is the blend file. [experiment with text.blend](experiment_with_text.blend) \n\nproblem seen in cpu/gpu/cycles/evee.\n\nregards,\nYaju", "Compositor in start up file does not show specific passes\nBroken: version: 3.1.0 Alpha\nWorked: Prior to 7e712b2d6a\n\nWhen loading the startup file, some passes will be missing from the compositor. For example the denoiser.\n\nHere is a video showing off the issue: [F12781327](Missing_Passes_in_Compositor.mp4)\n\nWatch the video above or follow the steps below:\n\n1. Start up Blender and change the render engine to Cycles.\n2. Go to the `Viewlayer Properties` tab and enable Denoising Data.\n3. Change one of the areas to the `Compositor` and tick the `Use Nodes` box.\n4. Take note of the passes that are available on the `Render layer` node.\n5. Select from the top of Blender `File -> Defaults -> Save Startup File`\n6. Close Blender\n7. Open a new instance of Blender 3.1 and notice that the `Render Layer` node is missing some of the inputs that should be there.\n\nHow it looks before you re-open Blender:\n{[F13081662](image.png), size=full}\n\nHow it looks afterwards:\n{[F13081667](image.png), size=full}\n\n\nExample of startup.blend:\n[startup.blend](startup.blend)", "VSE: Variable framerate videos cause proxies to mismatch\nOperating system: Linux-5.15.11-76051511-generic-x86_64-with-glibc2.34 64 Bits\nGraphics card: Quadro RTX 6000/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 470.86\n\nBroken: version: 3.1.0 Alpha\nWorked: unknown (never?)\n\nWith video files that have a variable framerate generated proxies end up not matching the real video in timing. This is a big flaw, as proxies are used by default and the rendered sequence will end up looking different than what is edited with the preview.\n[variable_framerate_proxy_mismatch-2022-01-20_19.31.26.mp4](variable_framerate_proxy_mismatch-2022-01-20_19.31.26.mp4)\nIn this example I am using a screen recording that was recorded with frame skipping and choppy input framerate.\n\nExample files attached:\n[variable_framerate_proxies.zip](variable_framerate_proxies.zip)\n\n\n", "Remove BKE_layer_collection_resync_forbid and BKE_layer_collection_resync_allow\nThese can be removed after each area has been checked and optimized to use the new structure (synced_ensure)\n\nRecently ceb0e7fcea was done and added a use for resync_forbid. This makes this ticket perhaps irrelevant?", "Killed particle's position is reset to the initial position when rendered\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce RTX 2080 with Max-Q Design/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 461.40\n\nBroken: version: 2.92.0\n\nI am encountering a strange particle behavior.\nWhen a particle collides with a plane where collision is enabled and \"kill particles\" is checked, the particle position seems to be reset to its initial position. This can not be observed in the viewport. This unexpected position reset is observed only when rendering the images.\nAs shown in this video:\n[0001-0050.mp4](0001-0050.mp4)\n\n[debug.blend](debug.blend)\nBake particle simulation and render images. You will get the same video as shown above.\n", "Dope sheet: seconds/frame numbers at top intersect when zooming in and out with mouse wheel, making them hard to read\nOperating system: Windows 10\nGraphics card: Nvidia GTX 1060\n\nBroken: 9c5d54bfaf48, 2019-06-23\n\nWhen I use the mouse-wheel to zoom in and out of the dope sheet, the timeline numbers (seconds, frames) at the top of the dope sheet cross over each other and become difficult to read.\n\nExpected behavior would be that they would cleanly arrange themselves as I used mouse-wheel to zoom in and out without the numbers crossing each other, so everything would still remain legible. See attached image for example:\n\n", "VSE: Cycles displayed as Rendered in Scene strip makes Blender hang - and other Scene strip bugs\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce RTX 2060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 462.59\n\nBroken: version: 2.93.0\n\nThe Render modes of Scene Strips are either missing or broken:\n- When Cycles is set as render and Rendered set as display, it'll make Blender hang.\n- Wireframe and Rendered doesn't work with any Render. \n- The only thing working is Eevee with either solid or Material Preview. \n- Workbench and Cycles settings seems to just use the Eevee render, since these are producing the same results in Solid and Material Preview.\n\n{[F10177251](scenestrip.gif),size=full}\n\nTest file:\nCycles is set up as render in the 3D Footage scene. In the Master Edit scene, switch display mode to Rendered. Blender will hang.\nTry the various other display modes and renders(by changing the render in the 3D Footage scene), to see wireframe and rendered doesn't work with any render. \n[multicam_3d_c.blend](multicam_3d_c.blend)\n\n\n", "Pose Mode X-Axis Mirror is not working properly with IK when skeleton is moved\nIntel Core i7, 16 Gb RAM, NVidia GeForce 2070 RTX\nWindows 10 x64\n\nBroken: 2.83 cfdb5d9a8b07 from 2020-02-19\nWorked: 2.83 5df494d75c9a from 2020-01-27 although it seems that this feature was introduced recently so technically it wasn't working before because symmetry didn't work with IK at all.\n\nPose Mode X-Axis Mirror is not working properly with IK when skeleton is moved or rotated from zero in Pose Mode.\n\n1. Open provided test file and try moving selected bone - opposite side will twitch weirdly instead of moving symmetrically.\n2. Now reset the skeleton to its default position and move the same bone - symmetry will work properly.\n3. Rotate first bone and try again - weird twitching again.\n[IK_Symmetry_Error.blend](IK_Symmetry_Error.blend)", "VDB sequences are rendered blocky here and there ( issues with Volume Displacement Modifier )\nOperating system: Windows 10 \nGraphics card: RTX 2070 CUDA\n\nBroken: blender-3.0.0-alpha+master.ef5a362a5b24-windows.amd64-release.zip\n\n\nBlender renders these blocky frames here and there during the actual render. The VDBs seem to be fine, the same botched render frames look fine in the viewport render. These anomolies are happening only during the actual render. See the attached image (middle right image) for what is happening please.\n\nThe same sequence renders without any issues in Houdini.\n\n(Cycles)\n\n\n\n\nHere is an actual movie render (with Eevee) that shows how horrible this issue is.\n[borkedframes_0001-0200.mp4](borkedframes_0001-0200.mp4)\n\n\n[#88966.blend](T88966.blend)\n[steam.4.0001.vdb](steam.4.0001.vdb)\n\n- Open .blend file\n- Render sequence with CUDA\n\nIssue should appear within 30 frames\n", "Extreme motion blur not rendering correctly in Cycles\nOperating system: Windows-10-10.0.16299 64 Bits\nGraphics card: x2 GeForce GTX 980/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 419.17\n\n\nBroken: version: 2.80 (sub 51)\nWorked: (optional)\n\n\nI'm trying to recreate the zig-zag motion blur of bullet tracers seen in WWII aircraft combat footage.\nThis is a test showing the effect I'm aiming for rendered with Redshift\n \n\nIn Blender the blur seems to be angular as if it does not obey the sub frame sampling.\n\n\nParticles must be baked otherwise they do not render properly\nAlso Blender crashes several times while rendering this scene. (CPU & GPU)\n\nPlease see attached simple Blend file.\n[BlurTest.blend](BlurTest.blend)\n\n\n\n", "Cycles viewport unneccessary update calls\nOperating system: Ubuntu 20.04\nGraphics card: Nvidia MX150\n\nBroken: 2.83.17, 8a8c61190b06, Beta and 2.90.2, 8f21f867e715, Alpha\nWorked: 2.82a official\n\n[#76970.blend](T76970.blend)\n\n* set scene renderer to Cycles\n* configure two viewports and a shader editor in a single screen\n* set one viewport to cycles preview render\n* set second viewport to workbench texture color\n* add two image texture nodes and load them with different textures\n* by clicking on the texture node both viewports are redrawn.", "Incorrect Areas affected by Timeline Playback settings\nVersion: 2.80 rc2 (Linux 64)\n\n[Timeline_bug.blend](Timeline_bug.blend)\n\nTo reproduce:\n\n - Open the example file (enable \"Load UI\" too)\n - With the mouse cursor in the 3d view, play the animation using Shift + Space (or Alt + A, whatever your keymap is)\n - Ignore the 3d view. Notice which animation editors show a moving playhead.\n - Inspect the playback settings in the timeline editor.\n - Disable \"All 3d Viewports\" and repeat step 2.\n - Now Enable \"Active Editor Only\" and repeat step 2.\n\nWith certain settings, the Dope Sheet and Timeline show playback even when the Graph Editor and NLA do not. It's as if these windows think they are 3d viewports!", "Cycles: Motion blur artifacts\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 472.88\n\nBroken: version: 3.2.2\n\n\n**description of error**\nWhen rendering an animated rigged model on cycle **with motion blur** ; if the model has **auto smooth enable AND a subdivision modifier**, then it is possible that strange artefact will appear during the render. (doesn't happen in the viewport) \n- This bug do not happen if auto smooth is disable or doesn't have a subdivision modifier. \n- The bug also only appear during some specfic poses. So this is mainly noticiable when rendering an animation.\n- Those artefact often appear near sharp edge.\n- Only happen on Cycle, Eevee render seem fine.\n- [Edit] changing the shutter values of the blur will sometime remove the bug on some frames OR make the bug appear on frame that was fine before.\n\nHere 2 Render exemple with and without autosmooth , both has motion blur enabled :\n \nAnd here an animation where you can notice the bug.\n[Animated0035-0318.mp4](Animated0035-0318.mp4)\n\n\n[MotionBlurCycleArtefact.blend](MotionBlurCycleArtefact.blend)\nIn this blender file, rendering the frame 302 will make the bug happen. \n- If you disable auto smooth on the model (the main body and not the cloth) then the render will be fine.\n- If you disable the subdivision modifier on the model then the render will be fine.\n- If you disable the subdivision modifier AND the auto smooth on the model then the render will be fine.\n\nThank\n", "Cycles: transparent object has black line where it intersect with other objects\nOperating system: Windows-10-10.0.18363 64 Bits\nGraphics card: GeForce RTX 2080 Ti/PCIe/SSE2 NVIDIA Corporation 4.6.0 NVIDIA 461.40\n\nBroken: 2.91.2\n\n\nWhen rendering transparent objects that intersect with other objects, black lines will appear on the intersecting area.\n\n1. Open a default scene.\n2. Add a plane to the scene.\n3. Rotate the plane a little.\n4. Move the plane up to make it intersect with the top face of the default cube.\n5. Make sure the view or camera view is very close to the intersecting area and lower the Clip Start value to 0.00001\n6. Set the renderer to Cycles and render the view.\n\n\n\n[Cycles_Black_Line_Transparency.blend](Cycles_Black_Line_Transparency.blend)" ]

[ "2.8 Vector render layer pass not working \nHello\n\nVector renderlayer pass is not showing up once the render is done, it render empty with no information at all. \nits quite a big deal, impossible to have any motion blur information when post processing motion blur in for example nuke.\n" ]

[ "Viewport Render (and Render Animation) color looks completely off from the Viewport\nOperating system: Darwin-19.0.0-x86_64-i386-64bit 64 Bits\nGraphics card: AMD Radeon Pro 560X OpenGL Engine ATI Technologies Inc. 4.1 ATI-3.2.24\n\nBroken: version: 2.82 (sub 6)\n\nRendering the viewport animation gives a very different result from the actual viewport.\n(Also: Viewport Anti-Aliasing isn't taken into consideration on the Viewport Render).\n\n||\n| -- | -- |\n\nThe problem is more noticeable when the Viewport Shading is set to \"Flat\", even though it also shows in \"Studio\". It delivers very dull colors with less contrast and a blue-ish tone.\nChanging between Cycles and Eevee doesn't affect the Viewport Render (from my tests).\n\n||\n| -- | -- |\nflat:\n||\n| -- | -- |\n\n- Open attached file.\n- Go to View -> Viewport Render Image\n[Viewport Colors.blend](Viewport_Colors.blend)", "Issues with HDRIs and render result pixel values that exceed half float limits\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 512.15\n\nBroken: version: 3.1.2\n(Also tested in latest 3.3 Alpha)\n\nSome lookdev HDRI's give the scene a colored tint, that also does not disappear after lowering the strength\n\nI imported this HDRI to preview lighting in my Eevee scene: lenong_2\n\nAt first, this looks okay in blender:\n\n\nBut after lowering the strength, you will notice that there is a yellow light that should not be there, as well as a white spot from where the sun used to be, even if strength=0\n\n\nI found a workaround by adjusting the exposure of the HDRi in advance by -1 in affinity photo. This makes the image only a tiny bit darker, but the end result is as you would expect it.\nNo yellow light, no sun.\n\n\nI tried this with another HDRi from HDRIhaven (from another creator, this one specifically ehingen_hillside) and the issue also existed there. \n\nHere is the blend file for testing:\n[HDRI_Test.blend](HDRI_Test.blend)\nAnd the HDRI fixed by me\n[lenong_2_1k_fixed.hdr](lenong_2_1k_fixed.hdr)", "Hair BSDF\nThis is a tough topic, I need to investigate it more.\nIdea: Dual Scattering using shadowmaps.\n\nEstimate : 2 weeks\n", "EEVEE AOV of full transparency is bad compared to Cycles\nOperating system: win10\nGraphics card: 1050ti\n\nBroken: 3.4\n\nAOV of full transparent shader is barely noticeable. Cycles works fine.\n\n\n[eevee_aov_transparent.blend](eevee_aov_transparent.blend)\nFor default scene:\n\n - Set fully transparent shader (alpha clip or hashed) and add AOV layer for it\n - Render EEVEE -> AOV layer is bad\n\n", "OSL on the GPU displays all black for Voronoi Distance > Bump Height\nSystem:\nWindows 11\nRTX3090Ti gpu\n\nBump output is pitch black when using Voronoi (Distance) output into Bump Height. Colorramp in between does not fix it. Noise output into Bump Height works fine.\n(Blender 3.5 in cycles)\n\nedit: more useful info + screenshot + isolated blend file\n\nRunning OSL on the CPU works fine, only the new GPU support seems affected.", "Display \"enum\" attributes as strings/icons in the spreadsheet\nThe numbers displayed currently are cryptic.\n- `handle_type`\n- `curve_type`\n- `normal_mode`\n- `knots_mode`\nIdeally the integer value would still be displayed in the tooltip.", "Color attribute ignores user color if it was added for mesh without faces\nOperating system: win10\nGraphics card: 1050ti\n\nBroken: 3.2, 3.5\n\nColor attribute automatically replaces user color with white (1, 1, 1) for mesh without faces. But if you create attribute for a plane and then delete only face, user color remains.\n\n - Open attached file\n - In `Properties` > `Mesh` > `Color Attributes` remove and add a new non-white color attribute for the vertex domain - Note that the mesh color is updated accordingly.\n - In `Properties` > `Modifiers`, apply its *Mask* modifier (so that the original mesh becomes wire)\n - Repeat second step (\"remove and add a color attribute\")\nColor becomes white\n[color_attribute_problem.blend](color_attribute_problem.blend)\n\n", "Adding noise to smoke simulation causes colors to change dramatically\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 950/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 527.56\n\nBroken: version: 3.5.0\nWorked: Not since mantaflow introduced in 2.82, worked with \"high res\" back in in 2.8\n\nWhen creating a quick smoke simulation with default settings (except changing the \"smoke color\" on the emitter object to purple. Adding Noise will make the smoke above the emitter turn yellow while the rest stays purple\n\n.blend has been attached with this reproduction.\n1. Delete everything from default scene.\n2. Add ico sphere.\n3. Select sphere, \"quick smoke\".\n4. Set sphere smoke color to purple.\n5. Enable noise in domain properties.\n6. Add area light to light up the smoke sim.\n7. Add \"color\" attribute to principled volume in the default domain node set up.\n\n", "Color Attribute not affecting \"bump\" shader node\nOperating system: Windows 10\nGraphics card: RTX 3080\n\nBroken: 3.50 beta (03 March 23)\n\nUsing a mesh's color attribute as an input to the \"bump\" shader node does not result in the expected behavior. Using the color attribute for practically every other value (including the normal map node, roughness, metalness etc.) works as expected.\n\nCreate a mesh and paint a color attribute with, for example, a black-and-white pattern.\nGive the mesh a material and add the color attribute.\nPlug the color attribute into the \"height\" input of the bump node, and view the result. There is no difference compared to the \"height\" input being left disconnected.\n\nOther inputs (such as a noise texture) result in the expected behavior.\n\n", "Improve volume render passes\nCycles volumes mostly do not write to render passes currently. Given the transparency it's not always useful or meaning to do this, but we can do better.\n\nA few known issues:\n- [ ] Cryptomatte is not written for volumes. We could accumulate cryptomatte contributions of objects while ray-marching.\n- [ ] Mist is not written, which causes the background value to be used when the volume is opaque. More useful would be to either write it based on the volume position weight by transparency, or to write mist for the object behind. (#91280)\n- [ ] There is no Volume Color pass currently, equivalent to the Diffuse/Glossy/Transmission Color passes. We do now write albedo for denoising, a similar thing could be added for volumes but it's unclear if this is useful for compositing in practice, especially with overlapping volumes.\n", "Outliner: 2020 GSoC Follow-up Tasks\nRemaining tasks from #77408 (Continued Outliner Improvements Design)\n- [x] Prevent setting Scene Collection color from python\n- [x] Group the double undo steps for changing modes\n- [x] Collection colors in other themes\n- [x] Remove remaining usage of the old collection icon (`ICON_GROUP`)\n- [ ] Set scene camera from context menu\n- [ ] Collection Exclude column overlap\n- [x] Grease pencil draw mode icon draws green in outliner mode column\n- [x] Box select from the gutter in object mode", "Quick Smoke does not assign color attribute to Principled Volume\nOperating system: Windows 10 64 bit\nGraphics card: Intel HD Graphics\n\nBroken: 2.90.1\n\nWhen running the Quick Smoke effect on something, the resulting automatic volume material does not have the color attribute assigned. Thus, any change of smoke color from the physics tab will not manifest by default: All smoke will simply be grey. You can manually fix this problem by typing \"color\" in the color attribute field of the Principled Volume node.", "EEVEE Volume draw distance broken with emission shader\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce RTX 2070 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 432.00\n\nBroken: version: 2.83.4\nWorked: none\n\nWhen using an Emission shader as Volume shader in EEVEE, draw distances smaller than 100 produce unecpected behaviour.\nUsing Principled Volume works fine as long as Density is not 0, if it is 0 the behavior breaks in the same way.\n\n- Open attached blend (Cube with emission shader as Volume)\n- Change Volumetrics End distance to 70m\n- Zoom in and out in EEVEE viewport, the cube does disappear at a distance range inbetween the set clipping range\n[eevee_volume_drawdistance_bug.mp4](eevee_volume_drawdistance_bug.mp4)\n[eevee_volume_drawdistance_bug.blend](eevee_volume_drawdistance_bug.blend)\n\n", "File Output node outputs incorrect dimensions in some cases\nOperating system: Linux 64-bit\n\n2.92\n\nWhen connected to certain composite node trees, the File Output node will output images with incorrect and unexpected dimensions. This appears to have something to do with the blur node, but I suspect the issue might occur with other nodes that alter the image buffer dimensions.\n\n[composite bug.blend](composite_bug.blend)\n\n - Open and inspect attached file\n - Adjust output file paths to whatever is convenient\n - Render a frame\n - Review the outputted image\n\n", "Add option to show an image in black & white in image editor.\nCurrently we have channel sizzling in the image editor.\n\nAlthough this sounds an easy task I would prefer to have a color transform function for this OCIO. But that would need some discussion with Brecht + Troy." ]

[ "Wrong socket color for shader sockets, red instead of green\n*Affected:*\nOperating system: Linux-5.4.0-56-generic-x86_64-with-debian-bullseye-sid 64 Bits\nGraphics card: SVGA3D; build: RELEASE; LLVM; VMware, Inc. 3.3 (Core Profile) Mesa 20.0.8\n\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 1080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 457.30\n\nBroken: version: 2.92.0 Alpha\nWorked: Currently checking, likely caused by one of the geometry node related commits.\n\nThe color of the shader sockets is currently red instead of green on my Linux build.\n\n- Open the shader editor" ]

[ "Clicking \"Bake All Dynamics\" causing various crashes.\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 536.99\n\nBroken: version: 3.6.2\n\nUsing \"Bake All Dynamics\" is beyond finicky and crashes Blender in many different use cases. Many little \"tricks\" are needed to make it work. I am using Disk Cache for all below situations, but I believe the crashes happen with or without this.\n\nSituations where it crashes:\nClicking \"Bake All Dynamics\" within the Cloth cache settings. It seems it's required to click \"Bake All Dynamics\" button within the context of the Rigid Body system, or else it crashes instantly.\n\nClicking \"Bake All Dynamics\" in the context of hair dynamics seems to do nothing. Same goes for \"Update All to Frame\"\n\nIf a rigid body simulation has been baked, clicking \"Delete All Bakes\", then clicking \"Bake All Dynamics\" immediately will also crash blender. The only workaround for this is to \"move\" an object even if using a zero value, before clicking \"Bake All Dynamics\" again. This prevents the crash. \n\n\nCreate a Rigid Body Sim, create a Cloth Sim. Check \"Disk Cache\" as an option. Try the various methods listed above.\n\n", "Killed particle's position is reset to the initial position when rendered\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce RTX 2080 with Max-Q Design/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 461.40\n\nBroken: version: 2.92.0\n\nI am encountering a strange particle behavior.\nWhen a particle collides with a plane where collision is enabled and \"kill particles\" is checked, the particle position seems to be reset to its initial position. This can not be observed in the viewport. This unexpected position reset is observed only when rendering the images.\nAs shown in this video:\n[0001-0050.mp4](0001-0050.mp4)\n\n[debug.blend](debug.blend)\nBake particle simulation and render images. You will get the same video as shown above.\n", "Rigid body : the animation of the angle constraint parameters does not work with an animated object\nOperating system: windows 10\nGraphics card: gtx 1070\n\nBroken: 2.92\nWorked: don't know \n\n\nRigid body : A rigid body \"animated\" object + a another rigid body object connected with hinge rigid constraint. The animation of the angle constraint parameters does not work with an animated object. I have bugs sometimes also with other parameters like breakable . (see the blend)\n\n[rigidconstraint.blend](rigidconstraint.blend)\n\n", "Buggy floor grid when using a micrometer unit scale\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1060 6GB/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 452.06\n\nBroken: version: 2.83.5\n\nWhen working with micrometer unit, setting up unit scale up to 0.0001 works ok. But, from 0.00001 micrometer value and grid scale changes to weird condition.\n\n\n- In Scene Settings set unitscale to 0.00001 or less (metric r imperial)\n\n", "Rigidbody cache is not invalidated when constraints are adjusted.\nOperating system: Ubuntu 20.04 x64\nGraphics card: RTX 2080\n\nBroken: 3.2\nWorked: Unknown\n\n\nWhen working with motor and slider constraints, adjusting them does not (always?) invalidate the existing rigidbody cache, causing the previously cached behavior to be retained.\nThis behavior does not persist on file reload (eg, saving and reloading the file with modified constraints corrects the issue), and anything else that will invalidate the cache also \"corrects\" the problem (ie, adjusting a simulated object).\n\nStarting with the included blend file.\n[repro.blend](repro.blend)\n 1. Press play on the timeline. Observe that the cube moves towards the +X axis.\n 2. Rotate the empty hosting the slider constraint (select from the outliner, or select the empty in the -Y quadrant) so that the local X axis points in another direction; for example, rotate 90d around the Z axis so it points in the +Y direction.\n 3. Rotate the empty hosting the motor constraint (select from the outliner, or select the empty in the +Y quadrant) so that the local X axis matches the direction of the slider constraint.\n 4. Press play on the timeline. Observe that the cube still moves towards the +X axis.\n\n[vokoscreenNG-2022-06-28_13-37-49.webm](vokoscreenNG-2022-06-28_13-37-49.webm)", "Edge Sliding with subdivision surface is extremely laggy (caused by \"bad\" UVs and UV Smoothing in the modifier)\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 527.56\n\nBroken: version: 3.4.0\n\nSo I've been modelling a raptor the last week and I noticed that viewport performance in edit mode was extremely laggy with subdivision surface modifier enabled. I checked the face count and it was just 150k faces which is not a lot at all. It's definitely a modifier problem because when I apply the subd modifier the lag goes away completely even when sliding twice the amount of edges around. I thought that was maybe a hardware issue so I opened task manager and while sliding edges with subd on I get like 4 fps with severe stuttering to top it off. All that while the cpu was at 10% usage and 5% gpu usage with memory barely touched (6gb used out of 64gb). I have nothing else running in the background like an ativirus or music app, I have the latest version of windows 10, latest gpu and chipset drivers, latest bios etc... I even have gpu subdivision enabled without autosmooth to make sure it actually works properly. I even run an sfc/scannow to check for other kinds of issues but nothing changed. Subd in edit mode however is not only slow when sliding edges and vertices but also in adding loop cuts or insetting faces. It is extremely frustrating so it would be great if you could fix this. I know edit mode was not made to handle massive polycounts but 150k is again not a whole lot and definitely comfortable to run on my PC. Here is the file to see for yourselves.\n\n- Open attached .blend\n- Execute Edge slide/loop cuts/inset faces\n[Broken Subd.blend](Broken_Subd.blend)\n\n", "rigid-body bake to keyframes with visibility keyframes \nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 1060 6GB/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 461.72\n\nBroken: version: 2.92.0\n\n\n\nI have a rigid-body object with visibility keyframes. and When I try to bake to keyframes I get an error:\n\n```\nPython: Traceback (most recent call last):\n File \"C:\\Program Files\\Blender Foundation\\Blender 2.92\\2.92\\scripts\\startup\\bl_operators\\rigidbody.py\", line 174, in execute\n bpy.ops.anim.keyframe_insert(type='BUILTIN_KSI_LocRot', confirm_success=False)\n File \"C:\\Program Files\\Blender Foundation\\Blender 2.92\\2.92\\scripts\\modules\\bpy\\ops.py\", line 132, in __call__\n ret = _op_call(self.idname_py(), None, kw)\nRuntimeError: Ошибка: Нет подходящей информации о контексте для активного ключевого набора\n\n\nlocation: <unknown location>:-1\n\nbpy.context.space_data.system_folders_active = 2\n «bug.blend» сохранено\nbpy.ops.rigidbody.bake_to_keyframes(frame_start=1, frame_end=250)\nPython: Traceback (most recent call last):\n File \"C:\\Program Files\\Blender Foundation\\Blender 2.92\\2.92\\scripts\\startup\\bl_operators\\rigidbody.py\", line 174, in execute\n bpy.ops.anim.keyframe_insert(type='BUILTIN_KSI_LocRot', confirm_success=False)\n File \"C:\\Program Files\\Blender Foundation\\Blender 2.92\\2.92\\scripts\\modules\\bpy\\ops.py\", line 132, in __call__\n ret = _op_call(self.idname_py(), None, kw)\nRuntimeError: Ошибка: Нет подходящей информации о контексте для активного ключевого набора\n\n\nlocation: <unknown location>:-1\n```\n\n\n\n\n[bug.blend](bug.blend)\n\n", "Cloth simulation does not work for a basic plane, having some specific location values.\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: AMD Radeon R5 M335 ATI Technologies Inc. 4.5.13587 Core Profile Context 19.50.26 26.20.15026.1\n\nBroken: version: 3.6.1\nWorked: Nothing found, same error with 3.3, 3.4, 3.5, 4.0 versions\n\nCloth physics in Blender seems to have a bug. It does not work correctly when the cloth object has some specific locations. Exact steps to reproduce the error are given below. The same cloth physics without any change seems to again work when the object is moved to another location. I could not establish any specific pattern between the location and the working/non-working status, but changing the location randomly solves the issue - although temporarily. If we again move it back to the previous location, cloth physics again stops to work. This is happening with all versions of Blender from 3.3 to 4.0 as far as I could test. I am reporting this officially for 3.6.1 latest version.\n\n1. Start with the default scene and delete the default cube.\n2. Add a plane, change X-rotation to 90 and Y-location to -2 (minus 2).\n3. Go to edit mode, right-click and subdivide, set subdivision level to 6.\n4. Select only the top-left & bottom-left vertices (blend file is attached for reference).\n5. Create a new vertex group and assign the selected vertices to this vertex group.\n6. Back to the object mode, add cloth physics with all default options.\n7. Enable self collisions, and select the above vertex group in the Pin group.\n8. Now run the simulation, or bake it and run. The cloth physics does not work.\n9. Now change the Y-location to zero. Cloth physics starts working after a few frames.\n10. Again change the Y-location to minus 2. Cloth physics stops working like before.\n\n", "Cloth simulation of Ngons are broken with Angular Bending Springs\nOperating system:Win\nGraphics card:RTX 3060 Ti\n\nBroken: 2.83, 3.2.2, 3.4\n\n\nThe attached file includes two similar meshes, one of which has an Ngon. A cloth simulation with the default settings was added except for some pinned vertices.\nThe mesh with the Ngon doesn't simulate correctly as shown in this video:\n\n[20220829-125207.mp4](20220829-125207.mp4)\n\n\nPlay animation:\n\n[cloth bug.blend](cloth_bug.blend)\n", "not possible to make multiple fabrics with different properties collide\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 512.15\n\nBroken: version: 3.1.0\nWorked: n\n\n1. By default, it is not possible to make multiple fabrics collide\n2. Use \"jion\" to combine multiple fabrics to generate collision, but it will lose different cloth properties at per object.\n3. Adding collision modifier while using fabric modifier will lead to very unexpected results.\n\nI have created a simple example to reproduce this problem.\n\n - > open the file\n [test_some_cloth.blend](test_some_cloth.blend)\n- > paly animation.\n\n", "Blender 2.90.1 crashes after rendering the first few frames of animation.\n**System Information** \nOperating system: windows 10 64 bit\nGraphics card: Nvidia GeForce GTX 1650\n\nBroken: (2.90.1 3e85bb34d0d7; master)\nWorked(didn't): (only tried on 2.8.3 and did not work at all)\n\nWhen I try to render the animation, after about 50 or so frames(varies randomly), blender crashes with no alerts or anything that tells me what the error is, it simply closes.\nI've rendered another project, and that did indeed render, so I'm left confused why this short animation would cause a crash.\n\nEdit: I disabled motion blur, and it rendered fine. However, I would really like to know why blender won't render if motion blur is enabled. The renderer is Eevee and while my laptop isn't a super computer, it should be able to handle it. Thank you all in advance!\n\n- Open attached file\n- render the animation\n[cube-loop-tut-V1.0.blend](cube-loop-tut-V1.0.blend)", "Skin modifier can produce undesirable geometry\nBroken: blender-2.74-65328fa-win32\nWorked: never as the skin modifier always had problems like that\n\nThe skin modifier sometime fail and generate very bad geometry\n\n[skinbug.blend](skinbug.blend)\nload the attached blend, and see the problem as shown in the screenshot {[F177512](screenshotbug01.jpg)}(the underlying vertices are all symmetrical, as that base was done with the mirror modifier) \nTest with the various x/y/z options in the skin modifier and see how that buggy geometry can move in other place of the generated mesh\n\nOnly way to fix is to try to either move around the vertices or to rescale them (CTRL+A with the skin modifier) , not real solutions if you need all the branches to be +/- the same diameter and at exact locations.", "Blender 2.71 (rc2) crashes when a UV sphere is added\nMac OS X 10.10 (beta) / AMD Radeon HD 6750M 512 MB\n\nBroken: 2.71 (release candidate 2)\nWorked: -\n\nBlender 2.71 (rc2) crashes when a UV Sphere is added to a existent project in Edit Mode.\n\nUsing the attached file, select the figure (stickman), go to Edit Mode and add a UV Sphere. After executing these steps, Blender stops responding and it's necessary to force quit and reopen. \n\nIt seems that's something to do with the modifiers (in the attached file, I'm using Mirror, Skin and Sub-surf). One or more of them are causing the crash when the UV Sphere is added. If I create a new project with no modifiers, I'm able to add UV Sphere. \n", "Subdiv hangs blender\n**System Information** DESKTOP-HR30UII\nOperating system:Windows 11 Home Single Language\nGraphics card:2GB NVIDIA GRAPHICS DRIVER\n\n\nBroken: Blender Version 3.5\nWorked: ?\n\nBLENDER 3.5 KEEPS NOT RESPONDING WHILE I USE MODIFIERS \n\nON BLENDER ADD MESH AND ADD MODIFIERS TO SUBDIVISION SURFACE THEN CHANGING THE VALUE OF LEVELS VIEWPORT VALUE ABOVE 4 .THEN ONWARDS BLENDER COMPLETELY STOPS WORKING AND I NEED TO RESTART IT AGAIN\n\n", "Flashes in Eevee render with Mantaflow viscosity enabled and Motion Blur\nWindows 7 Pro 64-bit\nNvidia GeForce GTX 1060 6gig\n\nBroken: Blender 2.92.0 alpha / c7e92e379d62 / 2021-01-02\n\nExperimented with the new viscosity setting in Mantaflow.\nRendered the test with Eevee and ended up with output where the surface of the fluid was occasionally 'flashing' in certain frames.\nMy only suspicion is that something it going wrong with the normals of the mesh created by Mantiflow.\nPlease see attached video clip and Blend file.\n[viscosity flashes.mp4](viscosity_flashes.mp4)\n\n- Open attached file\n- Bake simulation (maybe you need to set `modular` and `Is Resumable`)\n- Bake Mesh (maybe you need to set `modular` and `Is Resumable`)\n- Render Animation in Eevee\n\n[viscosity flashes.blend](viscosity_flashes.blend)\n\n\n" ]

[ "Cloth collision with rigidbody object doesn't work\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: Radeon RX550/550 Series ATI Technologies Inc. 4.5.13596 Core Profile Context 20.11.3 27.20.11027.5002\n\nBroken: version: 2.92.0 Alpha\nWorked: Never(2.80+)\n\nCloth collision with rigidbody object doesn't work\n\n[cloth.blend](cloth.blend)\n\n- Open file\n- Add collision physics to selected object\n# Play animation\n\nIt doesn't work.\n\n" ]

[ "2.8: Python API: Changing area type on frame change crashes Blender\nOperating system: Windows 10\nGraphics card: Quadro P4200/PCIe/SSE2\n\nBroken: 2.81a (also crashes in 2.79 and 2.80)\n\nI am attempting to sample the individual frames of an image on frame change, using a solution similar to the one described here: 5. I cannot seem to set the viewer area, however, from this context, as `bpy.context.screen` is `None`, and `bpy.context.window_manager.windows- [x].screen` seems to crash Blender in some situations.\n\nFrom the default startup file, run the following basic code:\n\n```\nimport bpy\n\ndef on_frame_change(scn):\n bpy.context.window_manager.windows[0].screen.areas[0].type = \"IMAGE_EDITOR\"\n\nbpy.app.handlers.frame_change_pre.append(on_frame_change)\n```\n\nRun the `Render Animation` operator to crash Blender.\n\nNote: This code runs fine when playing the animation in the 3D Viewport. It seems that this bug relates to the render context only.", "Regression: Crash on deleting shape key from Blender File view.\nOperating system: macOS-13.5-x86_64-i386-64bit 64 Bits\nGraphics card: Metal API AMD Radeon Pro 575 1.2\n\nBroken: version: 4.0.0, 3.2\nWorked: 3.1\n\nCaused by a71a513def202b119328035dbd68e86c2c47f7ac\n\nOutliner on Blender File when delete cause crash guide.\n\n- Open attached file\n- Add shape key\n- Switch outliner to Blender file display mode\n- Expand Shape key\n- select `key` then delete it from context menu\n\nr2jrCca\n\nYoutube:\nOlegeJrBbmE\n\nMy Dropbox file:\nLola%20Bunny.zip?dl=0\n\n- - -\nFirst create new ticket I unable to paste picture this here.\nI did drag there 250MB and (Remove) cause appear <html><head></head><body></body><html> please fix without <html> k5RFuuy\n\n", "Blender crashes if simplify is turned on while baking multires\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: AMD Radeon RX 5700 XT ATI Technologies Inc. 4.5.14742 Core Profile Context 21.7.2 27.20.22021.1002\n\nBroken: version: 3.0.0 Alpha\nWorked: --\n\n\nBlender crashes if simplify is turned on while baking multires data\n\n\n- Grab any object and add multiresolution to if, subdivide at least once.\n- Activate simplify, and turn the max subdivisions on viewport to 0\n- Create a material and try to bake the multiresolution. Blender will crash instead of giving an error message or just baking normally \n\nTest File:\n[multires_crash.blend](multires_crash.blend) ", "Crash when rendering animation in Workbench (with workbench shadows)\nOperating system: Windows 10 20H2\nGraphics card: Nvidia GTX1080\n\nBroken: 2.93.0\nWorked: 2.92.0\n\nWhen rendering the attached file as animation Blender ends without notice after the second rendered frame.[mirandaAnimNeu3.blend](mirandaAnimNeu3.blend)\n\nOpen the file mirandaAnimNeu3.blend. Click Render | Render Animation.\n\nThe animation is set to start at Frame 466 and to last until frame 4900. After rendering of frame 467 Blender crashes without notice. No crash file is created. Blender just ends.\n\nBoth rendered images are saved. It is interesting to note that the earlier images of this animation are rendered within seconds but the render times of the last images require minutes although no noticeable content becomes visible. Yesterday I have rendered all 4900 frames at 50% size without anti-aliasing, which worked fine. Each image rendered in a second or two.\nSorry that the blend-file is so big...", "Cycles Rendering of project causes crash of program.\nOperating system: Windows-10-10.0.22621-SP0 64 Bits (Windows 11)\nGraphics card: NVIDIA GeForce RTX 3070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 537.13\n\nBroken: version: 3.6.2\n\nTrying to render, pressing F12 of the file to render project it crashes. I'm using Blender 3.6.2\n\nTrying to render, pressing F12 of the file to render project it crashes. I'm using Blender 3.6.2\nNot sure if you need textures, can provide if needed.", "Blender crashes in edit mode in wireframe view\nOperating system: Linux Mint 21.1\nGraphics card: geforce gtx 1060\n\nBroken: 3.6.1, branch: blender-v3.6-release, commit date: 2023-07-17 12:50, hash: 8bda729ef4dc, type: release\nWorked: 3.5.1\n\nOn the attached file, selecting the bear head and going into edit mode, then switching to wireframe, or switching to edit mode while in wireframe mode, blender crashes and exit.\n\nSelect the bear head, switch to edit mode in wireframe. It crashes. It doesn't happen with a default cube in an empty file. The bear has geometry node hairs attached to it on the fur layer, but the geometry nodes are disabled. It also has an armature.\n\nThe problem also happens on windows10 on the same computer.", "2.80 FBX export simplify key not working as expected\nThe 'Simplify' keys option doesn't appear to work correctly\n\n**Goal** \n- to export all channels with keys.\n- keep keys that are the same value (flat curves).\n- not add keys to channels that don't have keys.\n\nWith 'simplify' set to 0.01 up to 1 - all keys that have a flat curve are removed, even rotations that are not 0,0,0.\nWith 'simplify' set to 0 (disabled) - all animate-able channels are keyed, even if there are no curves/keys on them. \n\nThis means there is no way to export with keys on only the channels I animated. \n\n**Should this be fixed???**\nSimplify shouldn't remove curves that are flat (especially if the values are not default) - or it should be a check box.\nDisabling 'simplify' shouldn't add keys to un-keyed channels - we have 'Force Start/End keying' for that.\n\n**Current workaround**\nI can get what I want by ensuring all bones with currently flat curves have some small tweak to make them not flat. Bit of a PITA as you might expect. \n\nIf these can't be fixed or aren't considered bugs then we don't have full control over animation exported from Blender in FBX files :( \n", "Lost hotkey with save preset\n2.80\n\n\nI finally can replicate problem from 2.7x and it is still problem in 2.80. When you change some combination of hotkeys and save it to the preset. You lose some of them. On my GIF example Numpad 1 dissapear.\n\nSometimes it come back after restart Blender, but sometimes it is just gone.\n\n\n\n", "Crash. Script changing values of a prop attached to object by an operator popup crashes the app\nOperating system: Windows 10 Pro\nGraphics card: NVidia GTX 1070 Ti\n\nBroken:\n(example: 2.80, edbf15d3c044, blender2.8, 2018-11-28, as found on the splash screen)\nWorked: (optional)\n\nWhen used from a plugin, operator that uses wm.invoke_props_popup(self, event) to draw(and change) properties from a property group attached to an object (constantly) crashes the app.\nWhen same code is used from the blender scripting area, the app does not crash, but after initial change of the param, the property is grayed out(seems like a bug as well).\nError : EXCEPTION_ACCESS_VIOLATION\nAddress : 0x00007FF7DDBEA080\nModule : c:\\Blender28\\program\\blender.exe\nThe terminal process terminated with exit code: 11\n\n1) Create a testing_grounds.py file\n2) Fill it with code provided below\n3) Install as an addon\n4) Select an object\n5) Launch the operator(through space, typing \"TEZZT\")\n6) Change the value of the property by clicking and dragging the value;\n7) Repeat step (6) a few times if the app hasn't crashed yet\n\n```\nbl_info = {\n \"name\" : \"TestingGrounds\",\n \"author\" : \"ChieVFX\",\n \"description\" : \"\",\n \"blender\" : (2, 80, 0),\n \"location\" : \"\",\n \"warning\" : \"\",\n \"category\" : \"Generic\"\n}\n\nimport bpy\n\nclass SomeParams(bpy.types.PropertyGroup):\n float_value : bpy.props.FloatProperty()\n\nclass TestOp(bpy.types.Operator):\n bl_idname=\"testing_ground.tezzt\"\n bl_label=\"TEZZT\"\n bl_options={'REGISTER', 'UNDO'}\n\n def invoke(self, context, event):\n wm = context.window_manager\n return wm.invoke_props_popup(self, event)\n \n def execute(self, context):\n print(\"Im fine\")\n return {'FINISHED'}\n \n def draw(self, context):\n layout : bpy.types.UILayout = self.layout\n obj = context.object\n some_params = obj.some_params\n layout.prop(some_params, \"float_value\", text=\"Prop popup test\")\n \nclasses = [\n SomeParams,\n TestOp\n]\n\ndef register():\n for cls in classes:\n bpy.utils.register_class(cls)\n \n bpy.types.Object.some_params = bpy.props.PointerProperty(type=SomeParams)\n\n\ndef unregister():\n for cls in classes:\n bpy.utils.unregister_class(cls)\n \nif __name__ == \"__main__\":\n register()\n bpy.ops.testing_ground.tezzt('INVOKE_DEFAULT')\n```\n", "Cycle crash\n[System Information](https://projects.blender.orgattachment)\n\nBlender version: 3.4.1\n\nCrash when switching from Material Preview to Rendered in Viewport Shading and GPU Compute was set for Cycles. In CPU mode it's OK.\n\n\nOpen project.\nSelect CPU in Cycles settings.\nSwitch from Material preview to Rendered in Viewport shading.\nNow it works.\nChange to GPU Compute in Cycles settings.\nCrash...\n\n", "Sculpt mode - program crash: brush smooth + graphics tablet\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1050 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 531.30\n\nBroken: version: 3.6.0\n\nUnless explicitly specified in \"preferences > input > tablet > tablet API > wintab or windows ink\", it crashes when using smooth brush. If I use the \"automatically\" setting.\nFor the bug to recur, the brush must take a force or size parameter from the pressure force on the graphics tablet. If I use the mouse, the error does not occur.\nThis error is repeated in any scene and with any object. I created a cube, remesh it and used a smooth brush in sculpt mode. Blender crashes to completion.\n\ngraphics tablet: xp-pen star 06\n\n", "Blender crash in bmesh.to_mesh() while using it to modify weights\nOperating system: Windows-10-10.0.18362 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 431.70\n\nBroken: version: 2.82 (sub 1)\nWorked: (optional)\n\nCrash in bmesh.to_mesh() if using it to modify weights in weightpainting while playing animation\n\n*Remove any lights or cameras, keep just one cube in the scene (for the simple script to work).\n*Subdivide it 5 times and apply the modifier.\n*Add the script below to the script editor and run it\n*Go to weight paint mode for the cube\n*Run the animation and try to paint. If you just click the left mouse button it will paint a dot and the script will work, but if you hold and drag, Blender crashes.\n\n\n```\nimport bpy, bmesh\n\n\ndef add_handler():\n # remove existing handlers\n old_handlers = []\n for h in bpy.app.handlers.frame_change_post:\n if \"paint_test\" in str(h):\n old_handlers.append(h)\n for h in old_handlers: bpy.app.handlers.frame_change_post.remove(h)\n # add new handler\n bpy.app.handlers.frame_change_post.append(paint_test)\n\ndef paint_test(scene):\n for ob in scene.objects:\n me = ob.data\n \n print(str(scene.frame_current) + \": Creating Bmesh\")\n bm = bmesh.new()\n bm.from_mesh(me)\n \n group_index = ob.vertex_groups.active_index\n dvert_lay = bm.verts.layers.deform.active\n \n for vert in bm.verts:\n dvert = vert[dvert_lay]\n try: dvert[group_index] -= 0.1\n except: pass \n \n print(str(scene.frame_current) + \": Writing Bmesh\")\n bm.to_mesh(me)\n bm.free()\n \nadd_handler()\n```", "Decals crash Eevee on AMD graphics card\nOperating system: Linux 6.2.7 Manjaro \nGraphics card: AMD 7900XT\n\nBroken: 3.3LTS, 3.4.1\n\nUsing decals will crash blender in Eevee. cycles works\n\nas an example file please see the .blend that was linked here by the decalmachine creator: 66438\n\nThe issue there is unrelated but the file can serve to show the issue I am having. The moment I switch to rendered view or materials preview blender will crash. I see no error message in console. The log files states just: \n\n```\nbpy.context.space_data.shading.type = 'RENDERED' # Property\nbpy.context.space_data.context = 'RENDER' # Property\nbpy.context.scene.render.engine = 'BLENDER_EEVEE' # Property\n \n```\n\n", "Geometry Nodes related memory leak and exception access violation\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 4070 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 535.98\n\nBroken: version: 3.5.1\nWorked: \n\nBlender freezes, memory usage spikes and eventually you get exception access violation crash.\n\nOpen attached file and enable Layer 3. Blender will freeze, memory usage will spike and eventually lead to exception access violation crash. Crash log is generated.\n\n[gnodes bug1.blend](attachment)\n\nTo stop crashing disconnect Twig Object from Object Info node. I think this started happening after I deleted the \"fir_twig\" object, but there was still a hidden reference in the Object info. However, I was not able to replicate this in a new file and the crash stops happening if I delete / modify some of the node groups, even in unrelated ways. I am pushing Blender to the limits and this is a heavily reduced file.\n\n", "crash while setting vertices positions using foreach_set\nOperating system: macOS-12.6.6-x86_64-i386-64bit 64 Bits\nGraphics card: Intel(R) Iris(TM) Graphics 6100 Intel Inc. 4.1 INTEL-18.8.6\n\nBroken: version: 3.6.0\n\ncrash while setting vertices positions using foreach_set\n\nstart blender, run script\n\n```python\nimport bpy\nimport numpy as np\n\nl = 2 ** 28\nme = bpy.data.meshes.new('mesh')\nme.vertices.add(l)\na = np.random.random((l, 3))\nme.vertices.foreach_set('co', a.ravel())\no = bpy.data.objects.new('mesh', me, )\nbpy.context.view_layer.active_layer_collection.collection.objects.link(o)\n```\n\noutput:\n```\nWriting: /tmp/blender.crash.txt\n[1] 13339 segmentation fault /Applications/Blender/blender-3.6.0.app/Contents/MacOS/Blender\n```\n\nin blender 3.4 while using 2**27 length output is this\n```\nError: Array length mismatch (expected -402653184, got 402653184)\nError: Python: Traceback (most recent call last):\n File \"/Text\", line 8, in <module>\nRuntimeError: internal error setting the array\n```\n\ncontents of `/tmp/blender.crash.txt` is just me running script, nothing interesting\n\ninteresting bit from crash report (i can provide in full)\n\n```\nThread 0 Crashed:: Dispatch queue: com.apple.main-thread\n0 Blender \t 0x10bbaef04 foreach_getset + 1124\n1 Blender \t 0x112f45046 cfunction_call + 134\n2 Blender \t 0x112f01015 _PyObject_MakeTpCall + 373\n3 Blender \t 0x112fdf95b call_function + 971\n4 Blender \t 0x112fdbc41 _PyEval_EvalFrameDefault + 24817\n5 Blender \t 0x112fd5a22 _PyEval_Vector + 146\n6 Blender \t 0x112fd5978 PyEval_EvalCode + 120\n7 Blender \t 0x10bb9cd2a python_script_exec + 506\n8 Blender \t 0x10c6754b0 text_run_script + 64\n9 Blender \t 0x10b600ce6 wm_operator_invoke(bContext*, wmOperatorType*, wmEvent const*, PointerRNA*, ReportList*, bool, bool) + 582\n10 Blender \t 0x10b600a03 wm_operator_call_internal(bContext*, wmOperatorType*, PointerRNA*, ReportList*, wmOperatorCallContext, bool, wmEvent const*) + 515\n11 Blender \t 0x10b602025 WM_operator_name_call_ptr_with_depends_on_cursor + 165\n12 Blender \t 0x10c36bffd ui_apply_but_funcs_after(bContext*) + 285\n13 Blender \t 0x10c366fc2 ui_handler_region_menu(bContext*, wmEvent const*, void*) + 434\n14 Blender \t 0x10b604a4f wm_handlers_do_intern(bContext*, wmWindow*, wmEvent*, ListBase*) + 1007\n15 Blender \t 0x10b604020 wm_handlers_do(bContext*, wmEvent*, ListBase*) + 48\n16 Blender \t 0x10b6033ba wm_event_do_handlers + 2554\n17 Blender \t 0x10b5f91b0 WM_main + 32\n18 Blender \t 0x10adea41e main + 878\n19 dyld \t 0x12788152e start + 462\n```\n\n" ]

[ "Shapekey transfer cause the crash of blender\nOperating system: Window 10 64bit\nGraphics card: PNY Nvidia GTX 1060\n\nBroken:\nblender-2.80-80ffa73b3f87-win64\n\nI attached a blend file that contains the problem , so basically i am trying to transfer shapekeys from a model named \"Pattern\" to another model named \"Pattern1\" that have no shapekeys that cause the crash of blender and outputs this in the console :\n\nError : EXCEPTION_ACCESS_VIOLATION\nAddress : 0x00007FF76B4C0963\nModule : C:\\Users\\test\\Downloads\\blender-2.80-80ffa73b3f87-win64\\blender-2.80.0-git.80ffa73b3f87-windows64\\blender.exe\n\nThis problem happens for those models only for some reason\n\nSelecting the model \"Pattern\" first then selecting the model \"Pattern1\"\nOn shapekeys panel i press on arrow and select \"Transfer Shape key\" \nand it crashes \n\n\n\n[bug.blend](bug.blend)" ]

[ "Geometry based lights do not generate shadow on a shadow catcher surface\nOperating system: Windows 10\nGraphics card: RTX 3060\n\nBroken: 3.0.0 and further versions\nWorked: (2.93.10 LTS)\n\nGeometry based lights do not generate shadow on a shadow catcher surface.\n\n- Open attached file\n- Switched to viewport rendering\n[#103795-3.blend](T103795-3.blend)\n\nBlender 2.93 LTS:\n\n\nBlender 3.x:\n", "Cycle Shader Displacement does not distinguish Instances by Geometry Nodes\nOperating system: Windows-10-10.0.17134-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 970M/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 466.11\n\nBroken: version: 3.0.0 Beta\n\nThe color of shader is different on each instance, but the shader displacement has all the same result for each one.\n\n\n[DISPLACEM.blend](DISPLACEM.blend)", "Workbench Shadow Improvement\n### Respect World Space Lighting option\nCurrently, the shadow is always in world space. Use view space orientation when matcaps is used or when studio lights use view space orientation.\n\n### Follow the brightest light\nCurrently, the shadow is not following any of the lights direction in the studio lighting config. Selecting the most bright light seems like a good default. Otherwise, we can add a flag in the studio-lights to tag which one the shadow should follow.\nA toggle option to force the use of the custom shadow direction should be added (disabled by default in new files, enabled in older files).\n\n\n### Follow the matcap\nWhen matcap brightest spot can be extracted, we should make the shadow follow this direction.\n\n", "Improve visualisation and manipulation of lights in the viewport\nThe idea is to better control/visualize light parameters in the viewport. Currently we can only control the spot angle of spot lights, the size of area lights and the light direction. \nConcept drawings by @eyecandy for reference, not the final design.\n**Tasks**\n- [ ] visualize light color (translucent filled? opaque outline? center circle?)\n\n- [ ] register gizmo undo\n- [x] make scaling transform follow the cursor exactly\n\nAdd gizmos to\n- [ ] spot light\n - [x] radius\n \n - [x] make scalable even when radius is zero?\n - [x] blend \n - [ ] spot angle via new gizmo type instead of arrow \n- [ ] area light\n - [x] constraint scaling in X/Y directions \n - [ ] beam spread (only supported in Cycles) \n - [ ] clip end/custom distance (only in Eevee) \n- [ ] sun lamp\n - [ ] angular diameter \n- [x] point light\n - [x] size ", "Workbench shadows disappear and reappear depending on what's selected.\nOperating system: Linux-6.5.3-arch1-1-x86_64-with-glibc2.38 64 Bits, X11 UI\nGraphics card: NVIDIA GeForce RTX 3090/PCIe/SSE2 NVIDIA Corporation 4.6.0 NVIDIA 535.104.05\n\nBroken: version: 4.0.0 Alpha\n\nWorkbench shadows disappear and reappear depending on what's selected. Makes no real sense as to why. This behaviour seems odd and I don't know if it was ever supposed to be like this.\n\n\n1. Open a new default scene and raise the default cube to sit on the floor.\n2. Add a mesh plane and scale it up to create a floor underneath it to display the cube's shadow.\n3. Change the render engine to Workbench and activate shadows in viewport shading popover menu.\n4. Optionally, toggle shadows in workbench render properties, as well.\n5. Select cube and watch the shadow disappear.\n6. Select the plane and watch the shadow reappear.\n\n", "World aov output nodes dont work unless background node's colour is >= 0.0001\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 536.99\n\nBroken: version: 3.6.3 Release Candidate\nWorked: Unknown\n\nWorld aov output nodes dont work unless background node's colour is >= 0.0001\n\n\n1. Open attached file\n2. go to the compositor of 'scene' and render still image\n3. the custom world aov is blank\n4. go to scene.001 and render\n5. the world aov now outputs correctly\n\nBoth scene's are identical, the only difference is scene.001's world background node's colour is set to 0.0001. The custom AOV is of type Value rather than colour. This is problematic because it breaks compositor setups that check for values > 0 etc.\n\n", "CPU render will look different to GPU render and fails randomly\nWin 10, GTX 880M, GTX 1080TI Founders Edition\n\nBroken: blender-2.79.0-git.2ca1f29-windows64\nWorked: na\n\nAttached you'll finde a scene what has issues with GPU rendering. The scene is using the old shaders with a bevel shader (OSL).\n\nThe preview will not start or maybe it takes a very long time. Here nothing happened.\nIn GPU mode the preview starts as expected.\n\nWhen I try to render with F12 in CPU mode it doesn't start or maybe it will start after a long time. Here nothing happened after 10 minutes.\nWhen I render a small region with CPU it takes a long time till it starts to render.\nWhen I start rendering in GPU mode it starts fast as expected ignoring OSL of course.\n\nIn the ZIP and attached you'll see a picture with a split screen. Left side is GPU. Right CPU. The results are different. Transparent parts with DOF will have extreme noise.\n\nLoad the scene and press F12. Here CPU render will not start as expected.\nSwitch to GPU and press F12. Rendering will start.\nDefine a small region where the transparent parts are. Press F12. It will take really long but start. Result has noise at the transparent parts.\nThis might be a bug in the Cycles render engine or an issue with the shader tree.\n\n[170902_Blender_279_Cycles_CPU_Issue.zip](170902_Blender_279_Cycles_CPU_Issue.zip)\n\n\n\n", "MacOS, Metal: Banding artefacts in viewport when using curves in color management\nOperating system: macOS-13.2.1-x86_64-i386-64bit 64 Bits\nGraphics card: Metal API AMD Radeon Pro Vega 64 1.2\nScreen: iMac 27 inch Retina 30bit\n\nBroken: version: 3.5.0\nWorked: same version with openGL viewport API\n\nIn the \"wanderer\" demo file you get white lines in the bright parts of the scene. They are visible in rendered view as well as in the final render (but only inside blender). \nWhen \"Use Curves\" in \"Color Management\" is turned off, they dissapear. \nWhen switching back to openGL api everything looks ok.\n\nOpen the demo file wanderer wanderer.blend\n\n", "Optix: Motion Blur seems to glitch out while rendering\n\nOperating system: Windows 11 Pro 22H2 22621.525\n\nGraphics card: NVIDIA GeForce RTX 3080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 526.98\n\n\nBroken: version: 3.3.2 LTS / 3.5.0 Alpha (Broken in both)\n\n\nHaving motion blur on while rendering with Optix seems to not work as expected in this scene. It renders out with artifacts (Alpha / black squares). Things work fine **when you turn off motion blur or switch to Cuda**.\n\n \n\n**Steps to reproduce the bug:**\nOpen the attached fine and render frame #82 with Optix enabled.\n\n[Keyframe remove 1_010.blend](Keyframe_remove_1_010.blend)\n", "Viewport material is darker than the rendered version\nOperating system: Windows-8.1-6.3.9600 64 Bits\nGraphics card: Intel(R) HD Graphics 4400 Intel 4.3.0 - Build 10.18.14.5067\n\nBroken: version: 2.82 (sub 3)\nConfirmed in 2.81 release\n\n\nThe material consists \na transparent shade mixed with a translucent shade, mixed with an emission shade with\nfrenel node multiplied with a noise texture as the factor\n\nThe sample sizes are the same\n\nJust render\n\n\n[necro_core_reduced.blend](necro_core_reduced.blend)", "modifier display/render driver not responding.\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1050/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 527.56\n\nBroken: version: 3.4.0\nWorked: unknown\n\nIf I have my Mask display/render driver from another objects display/render setting, changing controlling obj disp/render doesn't take effect without coaxing.\n\n1. Open the attached file. \n2. In outliner, disable the viewport display of the Cube object (drives Mask visibility of part of Sphere object)\n3. For the Mask modifier, slide the Threshold a little. Display effect now takes place.\n\nThis all works without a problem if I animate the Cubes visibility. This appears to be a problem for any/all modifier display drivers (e.g., Subdiviison). Sometimes I just need to hover over a value like Threshold, or just over time it responds to the change.\n\n", "Inconsistency between CPU and OptiX GPU when using OSL hash function \nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 4090/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 527.56\n\nBroken version: 3.5.0 Alpha\nWorked: NA\n\nWhen using the Open Shading Language \"hash\" noise function there is an inconsistency between the CPU and OptiX GPU implementation. The most noticeable is that the OptiX output is grey scale while the CPU implementation has colour.\n\n|CPU|OptiX|\n| -- | -- |\n|||\n\n1. In user preferences change your Cycles rendering backend to OptiX (requires a OptiX compatible GPU)\n2. Download the attached blend file and open it.\n3. Enter the rendered viewport and compare the results with CPU rendering and OptiX GPU rendering.\n\n[OptiX CPU OSL Hash Inconsistency.blend](OptiX_CPU_OSL_Hash_Inconsistency.blend)", "Incorrect interaction glass with visibility of other objects\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1060 6GB/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 471.96\n\nBroken: version: 2.93.3\n\nIncorrect interaction glass with visibility of other objects. I have objects which disappear if I look at them at a certain angle through the glass.\n\nGif of the trouble \n\n\nTest file:\n[#91424.blend](T91424.blend)\n", "IES led bar not working\nOperating system: WINDOWS 10\nGraphics card: GTX 1080 \n\nBroken: blender 3.6 alpha \n\n\nHello, I'm working for a lighting company, we have led bars and their IES doesn't work on blender (I use cycle) the point light to which I apply the IES doesn't take into account that it's a light bar and the light comes from a single point. I've put together a comparison with Vray used on 3DSMAX (software we want to leave to migrate to blender) so that you can understand what I'm explaining in more detail.\n\nBased on the default startup or an attached .blend file (as simple as possible).\n\n", "Strange Cycles Area Light artifact when using Motion Blur in animation\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3060 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 536.23\n\nBroken: version: 3.6.0\n\nPer the attached screenshot, two astronauts are at the windows of a moving spacecraft. There are four Area Lights inside the spacecraft between the astronauts and windows acting as fill lights for the astronauts. The strange artifact appears when Motion Blur is enabled, causing the Area Lights to appear to be moving at a different velocity from the spacecraft. The result is that the Area Lights themselves become visible, especially the green light at the far left window; in reality it's behind the post between the windows.\n\nIs this a known issue? If yes, any way to fix it short of disabling Motion Blur? Thanks.\n\n\n" ]

[ "View Transform and Look settings are ignored in LookDev mode\nWindows 10, Quadro 4000\n\nBroken: c86b5fa820d181c2beabdf4147ac17cb6ff8149b\nWorked: before mentioned commit\n\nSince the aforementioned commit, any change of View Transform or Look is entirely ignored by Workbench and LookDev mode, which to me makes no sense at all. I've created this bug report to check if everyone is fine with this change. Also Exposure and Gamma settings are ignored by LookDev. Knowing the artists on my team, I'm sure they do tweak these settings while creating shaders, to see how materials react on light condition changes. Also, we're using an external OCIO configuration instead of the one shipped with Blender (we use the original Filmic GIT repo and made some own modifications there). With this commit, LookDev becomes rather limited to us.\n\nI appreciate the separation of Eevee and LookDev a lot, since I can easily check what happens if I don't use scene lights, and try out other environments. But if I cannot use the same View Transform as with Eevee Render Mode, I have to do a lot of tweaks either again, or not use LookDev at all to be on the safe side.\n\nEven in Workbench mode I'm not entirely sure if this change is a welcome one, but I guess we could live with it. Haven't checked if this change goes through down to OpenGL renderings too. If so, the colors of my masks (which I generate using Random Color mode with flat shading) become mangled - they really need to be exported as sRGB EOTF.\n\n\n\nStart Blender using Factory settings, Go to LookDev mode, and change View Transform, Look, Exposure or Gamma - nothing changes. Only Display Device is considered. Note that at the moment you have to rotate your 3D View to see CM changes updating." ]

[ "Timeline scrubbing is laggy when Preview tab is open in Material Properties\nOperating system: Win10\nGraphics card: RTX3070\n\nBroken: 3.0\nWorked: N/A\n\n\n\nHaving the little preview tab open in Material Properties when scrubbing the timeline causes heavy lag, noticeable even on the default cube file. \n\nThe material can be a mesh material or a grease pencil material, but the result is the same.\n\nVideo example: watch?v=1Fw6Y-XzMyE\n\n", "Crash while using several rigid bodies and some force fields.\nOperating system: Darwin-18.7.0-x86_64-i386-64bit 64 Bits\nGraphics card: AMD Radeon R9 M370X OpenGL Engine ATI Technologies Inc. 4.1 ATI-2.11.20\n\n\nBroken: version: 2.80 (sub 75)\nWorked: (optional)\n\n\nThe scene is pretty simple, a house with some elements used as rigid bodies and some force field, vortex and wind force are applied.\n\nWhen I play the animation, it starts and around frame 140-200 it crashes.\n\nAttached blend file. The only thing to do is to play the animation and wait for the crash.\nAttached crash report.[Blender_2019-08-26-215345_MacBook-Pro-de-Kote.crash](Blender_2019-08-26-215345_MacBook-Pro-de-Kote.crash)\n\n[casaFalla.blend](casaFalla.blend)\n\n", "Decals crash Eevee on AMD graphics card\nOperating system: Linux 6.2.7 Manjaro \nGraphics card: AMD 7900XT\n\nBroken: 3.3LTS, 3.4.1\n\nUsing decals will crash blender in Eevee. cycles works\n\nas an example file please see the .blend that was linked here by the decalmachine creator: 66438\n\nThe issue there is unrelated but the file can serve to show the issue I am having. The moment I switch to rendered view or materials preview blender will crash. I see no error message in console. The log files states just: \n\n```\nbpy.context.space_data.shading.type = 'RENDERED' # Property\nbpy.context.space_data.context = 'RENDER' # Property\nbpy.context.scene.render.engine = 'BLENDER_EEVEE' # Property\n \n```\n\n", "AMD - Blender Crashes When I Start Render - Cycles GPU Compute\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: AMD Radeon RX 6800 XT ATI Technologies Inc. 4.5.0 Core Profile Context 23.7.1.230626\n\nBroken: version: 3.6.1\n\n[Whenever I start Rendering, It crashes when using cycles on GPU compute. Doesn't happen on CPU option.]\n\n[On render engine, select cycles. Then set it to GPU Compute. Then click on render on top left corner. Then select render image.]\n\n", "Regression: Crash when Sculpting on a mesh from a GP Lineart Modifier\nOperating system: Windows-10-10.0.22621-SP0 64 Bits (W11)\nGraphics card: NVIDIA GeForce RTX 4090/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 531.68\n\nBroken: version: 3.2, 4.0.0 Alpha\nWorked: 3.1\n\n\nWhen trying to sculpt a mesh in an assigned collection that a grease pencil lineart modifier references and the layer it is drawing to contains Grease Pencil strokes - it will instantly crash Blender. \n\n\n1. Download and extract zip to get the test blend file\n2. Load attached file with Load UI off\n3. Select the bust, it is in a GP lineart modifier collection \n4. Go to sculpt mode\n5. Try sculpt\n6. Crash\n\nAlternative steps:\n1. Create new scene\n2. Create a sphere\n3. Create a Grease Pencil object\n4. Add sphere to a new collection\n5. Select the GP object and add a lineart modifier\n6. Reference the sphere collection in the modifier\n7. Assign layer and material to the modifier\n8. Draw in the same layer in Draw mode of the GP\n9. switch to sculpt mode\n10. Try sculpt over sphere surface\n11. Instant crash\n\n", "Mantaflow: Minimum timesteps affects dissolve rate\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: AMD Radeon R9 200 Series ATI Technologies Inc. 4.5.13587 Core Profile Context 20.2.2 26.20.15019.19000\n\nBroken: version: 2.91.0 Alpha\n\nI was testing the dissolve feature described [here ](settings.html#dissolve). Time: \"Speed of smoke’s dissipation in frames.\" but it seems wrong. Let me show this short video (don't look at my test about flow consistency in domain settings):\n[0000-0150.mp4](0000-0150.mp4)\nAs you can see, changing the Maximum Timesteps does not change the Dissolve velocity, even changing the substeps of the inflow object doesn't change the dissolve behaviour.\nBut changing the Minimum Timesteps (forcing mantaflow to iterate more on each frame) it clearly messes up the behaviour. It seems to act faster than desired! It seems the Dissolve Time parameter works on steps and not frame.\n\n[MineExplosion.blend](MineExplosion.blend)\n\nBake simulations and play animation.\n", "Blender crashes with certain geometries (e.g. tweaked Jiggly Pudding simulation demo file)\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 4090/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 531.79\n\nBroken: version: 4.0.0 Alpha\nWorked: Unsure\n\nWith certain geometries, Blender will freeze then crash. I have included a demonstration in the video attached below:\n<video src=\"attachment\" title=\"Jiggly Pudding Crash.mp4\" controls></video>\n\n1. Download the Jiggly Pudding demo from demo-files/\n2. Follow the steps outlined in the video attached above. The steps in the video are as follows:\n a. Open the Jiggly Pudding file.\n b. Select the Pudding and open a geometry node editor.\n c. Inside the Simulation zone there is a node with the title \"Scale\". Change the Scale from 0.85 to 10 or something higher.\n d. Open a 3D viewport (if you haven't already) and press play in the time line.\n e. Blender will freeze and crash.\n\n", "Crashing while viewing animation in Cycles\nOperating system: macOS-13.0.1-arm64-arm-64bit 64 Bits\nGraphics card: Apple M1 Max Apple 4.1 Metal - 83\n\nBroken: version: 3.4.0\n\nWill crash seconds after playing animation in viewport when Cycles is selected and in render view. Both CPU and GPU crash. Does not crash when removing the Subdivision Surface Modifier so polycount may be a factor? This is a brand new Apple M1 Max computer from BestBuy with nothing else installed. Also tried initializing a render to video file and crashed as well. (Note: This is an error strictly with Apple M1 computers as I have tested it on an Apple intel MacBook Pro and it works just fine.) Thank you and happy holidays!\n\n1. Load file. It is a mesh plane with default cloth and subdivision modifiers acting on a wind force.\n2. Make sure Cycles is selected as Render Engine.\n3. Select Viewport Shading to Rendered (to view cycles render)\n4. Push play. Will crash or freeze seconds after play.\n\n[Cloth_Cycles_Crash.blend](Cloth_Cycles_Crash.blend)\n\n[Blender 3.4 Crash Report 12.13.22.rtf](Blender_3.4_Crash_Report_12.13.22.rtf)\n[Blender 3.5 Crash Report 12.14.22.rtf](Blender_3.5_Crash_Report_12.14.22.rtf)", "Crash on Python API bpy.ops.mesh.rip_move()\nOperating system: Windows 10\nGraphics card: RTX3080\n\nBroken: 3.3.2, 3.4.0, 3.5.0-alpha\n\nCrash on calling `bpy.ops.mesh.rip_move()` from Python API.\n\n - Open new Blender file\n - Switch to Scripting workspace\n - Create file with\n```\nimport bpy\nbpy.context.area.ui_type = 'VIEW_3D'\nbpy.ops.mesh.rip_move()\nbpy.context.area.ui_type = \"TEXT_EDITOR\"\n```\n - Delete all objects\n - `Shift-A` to add cylinder\n - Tab` to edit mode\n - Click edge or vertex selection mode\n - Select an edge or vertex\n - Run script\n - Crash\n\nManually doing a `V` rip move and then using the \"reports\" list to copy over the exact command to line #3, also results in a crash\n[blender.crash.txt](blender.crash.txt)", "Mantaflow APIC creates wrong droplet behaviour\nOperating system: Windows-10-10.0.19043-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 472.39\n\nBroken: version: 3.0.0 Alpha\n\nIf you create a big ball of water with the APIC solver, it breaks apart when falling. FLIP keeps the ball spherical until it touches the collision and APIC deforms it incorrectly immediately.\n\n**FLIP** (left) vs **APIC** (right) \ntimesteps were set to MIN 4 and MAX 8\n[Drop_FLIP_vs_APIC.mp4](Drop_FLIP_vs_APIC.mp4)\n\n\n\n\n - Create any primitive (I used an Icosphere in my case)\n - Perform a Liquid quick effect on that primitive\n - Change the simulation from FLIP (where it works) to APIC (where it doesn't work)\n - Simulate and observe\n\n\nBased on the default startup or an attached .blend file [Droplet.blend](Droplet.blend)\n\n", "GPencil: Blender crash when GP3 enabled & you open the default 2D animation file\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3060 Ti/PCIe/SSE2 NVIDIA Corporation 4.6.0 NVIDIA 536.23\n\nBroken: version: 4.0.0 Alpha\nWorked: 3.6\n\nBlender crash when GP3 enabled & you open the default 2D animation file & use the Draw tool\n\n- enable GP3 under the experimental features\n- save preferences and restart Blender\n- do File > New > 2D Animation\n- try to draw something.\n\nSince my 2D animation file is customized, I included it below.\n\n", "Cycle crash\n[System Information](https://projects.blender.orgattachment)\n\nBlender version: 3.4.1\n\nCrash when switching from Material Preview to Rendered in Viewport Shading and GPU Compute was set for Cycles. In CPU mode it's OK.\n\n\nOpen project.\nSelect CPU in Cycles settings.\nSwitch from Material preview to Rendered in Viewport shading.\nNow it works.\nChange to GPU Compute in Cycles settings.\nCrash...\n\n", "use-after-free crash after setting scene's render.engine inside custom render engine context\nOperating system: Ubuntu 20.04.2LTS\n\nBroken: 2.93.0 Beta, branch: master, commit date: 2021-05-19 10:14, hash: a294670bacb1, type: release\nAlso Broken: 0efb627bbd3ac3e53afb95591ae4e244bd1f8af4\n\nBlender crashes when setting some scene's render.engine property inside `__init__` or the `view_update` context of a custom render engine to a different one than the custom engine.\n\n[T88404_asan_report.txt](T88404_asan_report.txt)\n**Code to reproduce the error**\n```\nimport bpy\nimport bgl\n\nclass CustomRenderEngine(bpy.types.RenderEngine):\n bl_idname = \"CUSTOM\"\n bl_label = \"Custom\"\n bl_use_preview = True\n \n def __init__(self):\n return\n scene = bpy.data.scenes.new(\"foo\")\n scene.render.engine = \"CYCLES\"\n bpy.data.scenes.remove(scene)\n\n def __del__(self):\n pass\n\n def view_update(self, context, depsgraph):\n scene = bpy.data.scenes.new(\"foo\")\n scene.render.engine = \"CYCLES\"\n bpy.data.scenes.remove(scene)\n\n def view_draw(self, context, depsgraph):\n pass\n\ndef get_panels():\n exclude_panels = {\n 'VIEWLAYER_PT_filter',\n 'VIEWLAYER_PT_layer_passes',\n }\n\n panels = []\n for panel in bpy.types.Panel.__subclasses__():\n if hasattr(panel, 'COMPAT_ENGINES') and 'BLENDER_RENDER' in panel.COMPAT_ENGINES:\n if panel.__name__ not in exclude_panels:\n panels.append(panel)\n\n return panels\n\n\ndef register():\n bpy.utils.register_class(CustomRenderEngine)\n\n for panel in get_panels():\n panel.COMPAT_ENGINES.add('CUSTOM')\n\n\ndef unregister():\n bpy.utils.unregister_class(CustomRenderEngine)\n\n for panel in get_panels():\n if 'CUSTOM' in panel.COMPAT_ENGINES:\n panel.COMPAT_ENGINES.remove('CUSTOM')\n\nif __name__ == \"__main__\":\n register()\n```\n\n1) Copy code from above into a new script file inside of Blender\n2) Run the script\n3) Select \"Custom\" as Render Engine inside Render Properties\n4) Select \"Material Preview\" as display method in the viewport.\n\n[crash.blend](crash.blend)\n\nCrash log:\n```\n# Blender 2.93.0, Commit date: 2021-05-19 10:14, Hash a294670bacb1\nEngine 'CUSTOM' not available for scene 'Scene' (an add-on may need to be installed or enabled) # Error\nbpy.ops.text.run_script() # Operator\n\n# backtrace\n./blender(BLI_system_backtrace+0x20) [0xa63c950]\n./blender() [0xf6082a]\n/lib/x86_64-linux-gnu/libc.so.6(+0x46210) [0x7f7ce17be210]\n./blender(GPU_framebuffer_blit+0x2f) [0x8664faf]\n./blender() [0x1485fc3]\n./blender() [0x14597fe]\n./blender(DRW_draw_render_loop_ex+0x242) [0x1459a72]\n./blender(view3d_main_region_draw+0x8f) [0x1d3570f]\n./blender(ED_region_do_draw+0x851) [0x189c771]\n./blender(wm_draw_update+0x52e) [0x130cc9e]\n./blender(WM_main+0x30) [0x130aa20]\n./blender(main+0x319) [0xe881c9]\n/lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xf3) [0x7f7ce179f0b3]\n./blender() [0xf5d1bc]\n\n# Python backtrace\n```", "Crash when run the script code follow\nOperating system: Mac 11.4\nGraphics card: Intel Iris 1536 MB\n\nBroken: 3.0.1(2022-01-26)\n\nBlender crashed when running the script followed in \"Scripting Window\":\n\n```\nimport bpy\n\nfor obj in bpy.data.collections['Bg_objects'].all_objects:\n obj.is_holdout = True\n```\n\nAddressSanitizer error log:\n```lines=20\n==34714==ERROR: AddressSanitizer: heap-use-after-free on address 0x00015d792c38 at pc 0x00010523645c bp 0x00016fdfa1d0 sp 0x00016fdfa1c8\nREAD of size 8 at 0x00015d792c38 thread T0\n - 0 0x105236458 in rna_iterator_listbase_next rna_access.c:4677\n - 1 0x1053cc940 in Collection_all_objects_next rna_collection_gen.c:209\n - 2 0x105234920 in RNA_property_collection_next rna_access.c:3690\n - 3 0x105f6c330 in pyrna_prop_collection_iter_next bpy_rna.c:7107\n - 4 0x11366ad40 in _PyEval_EvalFrameDefault ceval.c:4001\n - 5 0x113664fe0 in _PyEval_Vector ceval.c:5065\n - 6 0x113664f38 in PyEval_EvalCode ceval.c:1134\n - 7 0x105f1812c in python_script_exec bpy_interface_run.c:121\n - 8 0x105f18d5c in BPY_run_text bpy_interface_run.c:216\n - 9 0x10a275444 in text_run_script text_ops.c:773\n - 10 0x10a275284 in text_run_script_exec text_ops.c:812\n - 11 0x103035044 in wm_operator_invoke wm_event_system.c:1335\n - 12 0x103033cf4 in wm_operator_call_internal wm_event_system.c:1530\n - 13 0x10303ad6c in WM_operator_name_call_ptr wm_event_system.c:1577\n - 14 0x10303b978 in WM_operator_name_call_ptr_with_depends_on_cursor wm_event_system.c:1765\n - 15 0x109722ff8 in ui_apply_but_funcs_after interface_handlers.c:1008\n - 16 0x10971b1d4 in ui_handler_region_menu interface_handlers.c:11383\n - 17 0x10304ab2c in wm_handler_ui_call wm_event_system.c:698\n - 18 0x103047864 in wm_handlers_do_intern wm_event_system.c:3025\n - 19 0x1030426f4 in wm_handlers_do wm_event_system.c:3145\n - 20 0x1030406f0 in wm_event_do_handlers wm_event_system.c:3675\n - 21 0x1030043ec in WM_main wm.c:623\n - 22 0x10000ba94 in main creator.c:544\n - 23 0x1454710f0 in start+0x204 (dyld:arm64e+0x50f0)\n\n0x00015d792c38 is located 8 bytes inside of 80-byte region [0x00015d792c30,0x00015d792c80)\nfreed by thread T0 here:\n - 0 0x145a890f0 in wrap_free+0x98 (libclang_rt.asan_osx_dynamic.dylib:arm64e+0x3d0f0)\n - 1 0x1151b28e8 in MEM_lockfree_freeN mallocn_lockfree_impl.c:116\n - 2 0x114b67940 in BLI_freelistN listbase.c:463\n - 3 0x1003acbf8 in collection_object_cache_free collection.c:816\n - 4 0x1003aca74 in BKE_collection_object_cache_free collection.c:826\n - 5 0x100e4b080 in BKE_main_collection_sync_remap layer.c:1366\n - 6 0x1059bcf0c in rna_Object_hide_update rna_object.c:370\n - 7 0x105257fa8 in rna_property_update rna_access.c:2069\n - 8 0x105257bac in RNA_property_update rna_access.c:2138\n - 9 0x105f50268 in pyrna_py_to_prop bpy_rna.c:2083\n - 10 0x105f54308 in pyrna_struct_setattro bpy_rna.c:4453\n - 11 0x1135de558 in PyObject_SetAttr object.c\n - 12 0x113667dd0 in _PyEval_EvalFrameDefault ceval.c:2850\n - 13 0x113664fe0 in _PyEval_Vector ceval.c:5065\n - 14 0x113664f38 in PyEval_EvalCode ceval.c:1134\n - 15 0x105f1812c in python_script_exec bpy_interface_run.c:121\n - 16 0x105f18d5c in BPY_run_text bpy_interface_run.c:216\n - 17 0x10a275444 in text_run_script text_ops.c:773\n - 18 0x10a275284 in text_run_script_exec text_ops.c:812\n - 19 0x103035044 in wm_operator_invoke wm_event_system.c:1335\n - 20 0x103033cf4 in wm_operator_call_internal wm_event_system.c:1530\n - 21 0x10303ad6c in WM_operator_name_call_ptr wm_event_system.c:1577\n - 22 0x10303b978 in WM_operator_name_call_ptr_with_depends_on_cursor wm_event_system.c:1765\n - 23 0x109722ff8 in ui_apply_but_funcs_after interface_handlers.c:1008\n - 24 0x10971b1d4 in ui_handler_region_menu interface_handlers.c:11383\n - 25 0x10304ab2c in wm_handler_ui_call wm_event_system.c:698\n - 26 0x103047864 in wm_handlers_do_intern wm_event_system.c:3025\n - 27 0x1030426f4 in wm_handlers_do wm_event_system.c:3145\n - 28 0x1030406f0 in wm_event_do_handlers wm_event_system.c:3675\n - 29 0x1030043ec in WM_main wm.c:623\n\npreviously allocated by thread T0 here:\n - 0 0x145a89380 in wrap_calloc+0x9c (libclang_rt.asan_osx_dynamic.dylib:arm64e+0x3d380)\n - 1 0x1151b13b8 in MEM_lockfree_callocN mallocn_lockfree_impl.c:222\n - 2 0x1003b5754 in collection_object_cache_fill collection.c:754\n - 3 0x1003b5d20 in collection_object_cache_fill collection.c:775\n - 4 0x1003b53b0 in BKE_collection_object_cache_get collection.c:786\n - 5 0x1053cc5d8 in rna_Collection_all_objects_begin rna_collection.c:54\n - 6 0x1053cc2d0 in Collection_all_objects_begin rna_collection_gen.c:200\n - 7 0x105233a88 in RNA_property_collection_begin rna_access.c:3674\n - 8 0x105f6c158 in pyrna_prop_collection_iter_CreatePyObject bpy_rna.c:7076\n - 9 0x105f67ef0 in pyrna_prop_collection_iter bpy_rna.c:7083\n - 10 0x11357f400 in PyObject_GetIter abstract.c:2813\n - 11 0x11366ac34 in _PyEval_EvalFrameDefault ceval.c:3958\n - 12 0x113664fe0 in _PyEval_Vector ceval.c:5065\n - 13 0x113664f38 in PyEval_EvalCode ceval.c:1134\n - 14 0x105f1812c in python_script_exec bpy_interface_run.c:121\n - 15 0x105f18d5c in BPY_run_text bpy_interface_run.c:216\n - 16 0x10a275444 in text_run_script text_ops.c:773\n - 17 0x10a275284 in text_run_script_exec text_ops.c:812\n - 18 0x103035044 in wm_operator_invoke wm_event_system.c:1335\n - 19 0x103033cf4 in wm_operator_call_internal wm_event_system.c:1530\n - 20 0x10303ad6c in WM_operator_name_call_ptr wm_event_system.c:1577\n - 21 0x10303b978 in WM_operator_name_call_ptr_with_depends_on_cursor wm_event_system.c:1765\n - 22 0x109722ff8 in ui_apply_but_funcs_after interface_handlers.c:1008\n - 23 0x10971b1d4 in ui_handler_region_menu interface_handlers.c:11383\n - 24 0x10304ab2c in wm_handler_ui_call wm_event_system.c:698\n - 25 0x103047864 in wm_handlers_do_intern wm_event_system.c:3025\n - 26 0x1030426f4 in wm_handlers_do wm_event_system.c:3145\n - 27 0x1030406f0 in wm_event_do_handlers wm_event_system.c:3675\n - 28 0x1030043ec in WM_main wm.c:623\n - 29 0x10000ba94 in main creator.c:544\n```\n\n- Open attached file\n- Go to Scripting workspace\n- Run the above script\n[bug.zip](bug.zip)", "Crush while baking a cloth simulation\nOperating system: macOS-13.1-arm64-arm-64bit 64 Bits\nGraphics card: Metal API Apple M1 Pro 1.2\n\nBroken: version: 3.6.2\n\nI was performing the baking of a cloth simulation. During the initial and subsequent baking attempts, everything worked fine. However, when I tried to re-bake the simulation after moving the cursor in the timeline, the program crashes. I tries to bake that cloth in previous versions. It didn't work as well\n\nTo trigger the crash, you simply need to move the cursor in the timeline and then initiate the baking process.\n\nUnfortunately, i cannot add the file (size is 200mb and i can't make it smaller ), so than i added the crash txt file and a link to the file on google drive\nview?usp=sharing\nThank you!\n\n" ]

[ "Crash While Scrubbing Viewport render of fire sim\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce RTX 2070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 441.66\n\nBroken: version: 2.83 (sub 2)\nWorked: (optional)\n\nWhile scrubbing the timeline while cycles viewport rendering of a fire sim blender crashes\n\nIt appears intermittent but is repeatable, my guess is some thread conflict or race condition\n\n- Quick Smoke --> Shange type to fire\n- Set cache type to final\n- bake all\n- Set render to cycles\n- viewport Render\n- scrub timeline\n\n" ]

[ "Cycle crash\n[System Information](https://projects.blender.orgattachment)\n\nBlender version: 3.4.1\n\nCrash when switching from Material Preview to Rendered in Viewport Shading and GPU Compute was set for Cycles. In CPU mode it's OK.\n\n\nOpen project.\nSelect CPU in Cycles settings.\nSwitch from Material preview to Rendered in Viewport shading.\nNow it works.\nChange to GPU Compute in Cycles settings.\nCrash...\n\n", "crash when i try to make instances real for nodes\nhi , in attached files all info about the crash , its when i try to make instances real for nodes modifier addon scifi pro + flex\nso can i fix it ?\n\n", "Blender 3.5 crash on render with Intel 4600 integrated GPU\nOperating system: WIndows 10 x64\nGraphics card: Intel 4600\n\nBroken: 3.5 \nWorked: 3.4.1\n\n\nBlender crashes when attempting to render in Evee. Cycles seems to work.\n\n", "Crash when selecting line art\nOperating system: Linux mint 21. Blender installed using flatpak.\nGraphics card: See attached system_information.txt\n\nBroken: version: 3.5.1\n\nBlender Crashes.\n```\nbacktrace\n\nPython backtrace\nFile \"/app/blender/3.5/scripts/startup/bl_ui/space_view3d.py\", line 704 in draw\nFile \"/app/blender/3.5/scripts/modules/bpy_types.py\", line 903 in draw_ls\n```\n\n```Py\nbpy.ops.outliner.item_activate(deselect_all=True) # Operator\nbpy.ops.outliner.item_activate(deselect_all=True) # Operator\n```\nWith attached blend file, in the outliner, select cameras, line-art then try to change the color of the background.\n\n", "Crash when change material on render with python\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce RTX 2080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 460.79\n\nBroken: version: 2.8+, 2.9 - 2.93.0 Alpha\n\nHey :)\nI’m trying to change material on a few objects during \"frame change\" with python.\nIn a viewport everything works fine but during Animation Rendering(cycles or eevee) it crashes immediately or a few frames later.\nMy system can’t handle more than 500 cubes.\n\n\n- Open \"change_mat.blend\"\n- Run script \"change_mat\" from Text Editor\n- Change Frame in Time-Line (0,1,2) // All cubes in the scene/viewport change their active material (Red, Green, Blue)\n- Press \"Render Animation\" => Crash // If it does not crash then duplicate more cubes\n\n[change_mat.blend](change_mat.blend)\n[change_mat.crash.txt](change_mat.crash.txt)\n\nMany Thanks!\n", "Regression: Crash with boolean modifier\nOperating system: Fedora Linux 36 (Workstation Edition)\nGraphics card: Nvidia Geforce RTX 3080\n\nBroken: 3.2.1, 3.3 alpha\nWorked: 3.1\n\nCrash with boolean modifier\n\n- Open .blend file\n- In boolean modifier tab -> select `dot cutter` for boolean operation\n[yin_yang-v1.blend](yin_yang-v1.blend)\noriginal file: [F13292564](yin_yang.blend)", "Keyed Particle Duration Crash\nOperating system: Windows 7 64\nGraphics card: nVidia GeForce GTX 970\n\nv.2.80.60\nDate: 2019-05-10 22:21\nHash: de9d846353bd\n\nBlender will sometimes crash when a keyed physics type particle system references more than one particle system. This usually occurs when attempting to change the duration setting of the second particle system.\n\nIn the provided file, change the duration setting in the particle system settings. Blender will crash.\n\n[Keyed Particle Duration Crash.blend](Keyed_Particle_Duration_Crash.blend)", "Crash when baking Dynamic Paint Image Sequence multiple times in loop\nBroken: 2.90.0 Alpha\nWorked: blender-2.80-windows64\n\nBlender crashes when baking dynamic paint image sequence in for loop. Depending on wait time crash will have differens stack traces.\n\n[#74052.blend](T74052.blend)\n\n- Open file\n- Run script", "Blender crashed when increasing Strand Steps in the Viewport Display on the Particle tab\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: Quadro RTX 3000 with Max-Q Design/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 535.98\n\nBroken: version: 3.6.0\nWorked: I dunno\n\nBlender crashed when increasing Strand Steps in the Viewport Display on the Particle tab\n\n<video src=\"attachment\" title=\"Crash when particle edit.mp4\" controls></video>\n\n1, Open the attached Blender file\n2, Rekey it to have a higher keypoints\n3, Increase the Strand Steps in the Viewport Display on the Particle tab\n\n", "Texture Paint: Color Palette operator issues with Undo/Redo -- Vertex Paint is fine though\nOperating system: Darwin-15.6.0-x86_64-i386-64bit 64 Bits\nGraphics card: ATI Radeon HD 5770 OpenGL Engine ATI Technologies Inc. 4.1 ATI-1.42.15\n\nBroken: version: 2.83.0\n\n\nIn Texture Paint mode (contrary to vertexpaint mode or gpencil), the palette operators dont work with Undo/Redo\n\nFrom the default startup file:\n- go to vertexpaint mode\n- (a) in the Active Tool tab in the Properties Editor, under Brush Settings > Color Palette, add a palette\n- (b) add a color to the palette\n- (c) Undo --> should be working [color is removed again]\n- so far everything is working\n- go to texture paint and do the same a, b, c --> color is not removed again, Undo is not working\n\nThis is what is probably the cause for the following more specialized original report (but also there the same thing works fine in vertex paint mode)\n\n## original report\n\nWhile in texture mode users press S to activate a color picker (not sure why this isnt the GUI nor the tools palette?)\nWhen its active, we have the options menu (used to be F6 i believe) There is this option \"Add to Palette\".\nThis however always adds colors. When you uncheck it, it will enable itself the next time your use the S key and pick a color.\n\n01. Open the attached scene.\n02. Start texture paint mode\n03. Use S key to enable color picker\n04. Options menu (bottom left) shows \"Add option to palette\" > Deactivate it.\n05. Use S key again > Notice colors are still being added but option is unchecked.\n06. When you let go of the S key, its checked again\n\n\n\n\n[IMport Palette - Blank.blend](IMport_Palette_-_Blank.blend)", "Blender crashes when using an EnumProperty to set Space\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3090/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.94\n\nBroken: version: 3.4.0 Alpha\n\nIn this script, I'm using `context.area.type` with a property to change the Space type. It caused Blender to crash.\n\n1. Running this script from the Text Editor (below)\n2. Switch to Graph Editor or Dope Sheet\n3. Click new property buttons on right of Header\n\n```\nimport bpy\nfrom bpy.types import PropertyGroup\nfrom bpy.props import EnumProperty\n\ndef set_space(self,value):\n context = bpy.context\n\n for area in context.screen.areas:\n if area != context.area:\n break\n\n override = {'region': area.regions[0]}\n\n if value == 0:\n context.area.type = \"GRAPH_EDITOR\"\n elif value == 1:\n context.area.type = \"DOPESHEET_EDITOR\"\n\nclass SWITCH_PG_switcher_props(PropertyGroup):\n space: EnumProperty(\n name=\"Space Target\",\n description=\"Switch to space\",\n items=(\n ('GRAPH_EDITOR', \"\", \"Graph Editor\", \"GRAPH\", 0),\n ('DOPESHEET_EDITOR', \"\", \"Dopesheet Editor\", \"ACTION\", 1),\n ),\n default='GRAPH_EDITOR',\n set = set_space\n )\n\ndef draw_enum(self, context):\n layout = self.layout\n scene = context.scene\n layout.prop(scene.switcher, \"space\", icon_only=True, expand=True)\n\ndef register():\n bpy.utils.register_class(SWITCH_PG_switcher_props)\n bpy.types.Scene.switcher = bpy.props.PointerProperty(type=SWITCH_PG_switcher_props)\n bpy.types.DOPESHEET_HT_header.append(draw_enum)\n bpy.types.GRAPH_HT_header.append(draw_enum)\n\ndef unregister():\n del bpy.types.Scene.switcher\n bpy.types.DOPESHEET_HT_header.remove(draw_enum)\n bpy.types.GRAPH_HT_header.remove(draw_enum)\n bpy.utils.unregister_class(SWITCH_PG_switcher_props)\n\nif __name__ == \"__main__\":\n register()\n```\n\nSetting the `context.area.type` in Python seems to cause others' scripts to crash:\n[Performing a Translate Operation after switching the context area crashes Blender ](1164763)\n", "Selecting vertices and centering the model crashes Blender\nOperating system: Windows 10\nGraphics card: RTX 2070 \n\nBroken: 03015a9b222e\n\n\nSelecting some vertices and trying to center the selection crashes blender. See the crash log\n\n[crash.txt](crash.txt)\n\nOpen the attached .blend\nSelect the object\nGo into edit mode\nSelect all vertices\nCenter (press Numpad .)\n\n\n\n\n\n[_13122021_2147_15.blend](_13122021_2147_15.blend)", "Resetting singular values in the Themes disables them\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 471.68\n\nBroken: version: 2.93.5\n\nInstead of globally resetting the default themes, if I use the \"Reset to default\" for a color or a value, it simply puts it to black/0.0.\n[2021-10-27 14-12-52.mp4](2021-10-27_14-12-52.mp4)\n\nGo to themes and right click a color or value.\n[Based on the default startup]\n\n", "Crash when selecting object on Mac\nOperating system: macOS-12.5.1-x86_64-i386-64bit 64 Bits\nGraphics card: AMD Radeon Pro 5700 XT OpenGL Engine ATI Technologies Inc. 4.1 ATI-4.8.101\n\nBroken: version: 3.3.0\n\nSystem crash trying to open einar_hair.blend file\n\nDownload einar_hair.blend fle from ?asset=6072\nSelect any object.\n\nBlender will crash. Also hair is not visible. It's possible, that OS will crash as well. ", "Color Attribute not affecting \"bump\" shader node\nOperating system: Windows 10\nGraphics card: RTX 3080\n\nBroken: 3.50 beta (03 March 23)\n\nUsing a mesh's color attribute as an input to the \"bump\" shader node does not result in the expected behavior. Using the color attribute for practically every other value (including the normal map node, roughness, metalness etc.) works as expected.\n\nCreate a mesh and paint a color attribute with, for example, a black-and-white pattern.\nGive the mesh a material and add the color attribute.\nPlug the color attribute into the \"height\" input of the bump node, and view the result. There is no difference compared to the \"height\" input being left disconnected.\n\nOther inputs (such as a noise texture) result in the expected behavior.\n\n" ]

[ "Intermittent crash clicking material color selector \nWindows 8.1, GTX 770 and 780Ti's, 16GB RAM, i7-4930k cpu and i7-4770 cpu's\n\n\nBroken: 2014-07-01 23:50 Hash 5898abe\nWorked: N/A\n\nIntermittent crash clicking material color selector \n\nSadly there is no way we have figured out how to reproduce the following error, though it is occurring on 4 different PC's. When it occurs it is the same crash scenario with the same action. \n\nClicking on the selection dot that should popup the color chooser under the Material tab, Surface dialogue, Color.\n\nThe modelers who are having the problem report the problem began this past Friday July 4.\n\nThe error randomly occurs on NVidia stable and beta as we thought that might be the problem.\n" ]

[ "Filesystem accidental creations of directories\nOperating system: Windows 10 Pro (64-bit), version 1809, OS build 17763.615\nOperating system: Ubuntu 18.04.2 LTS 64-bit\nGraphics card: Nvidia GTX 650\n\nWindows Build version: 2.80 (sub 75), branch: master, commit date: 2019-07-29 14:47, hash: f6cb5f54494e, type: \nWindows Build version: 2.81 (sub 1), branch: master, commit date: 2019-08-05 13:33, hash: 8f1a55831c89, type: Release\nUbuntu Build version: 2.80 (sub 75), branch: master, commit date: 2019-07-29 14:47, hash: f6cb5f54494e, type: Release\n\nThe window pop-up asking for confirmation of a filesystem change, e.g creating a directory, seeks cursor and a click gets immediately accepted as a confirm.\n\nFile -> Open\nClick the \"file path\" area (\"C:\\Users\\%USERNAME%\\\" or similar)\nType \"C:\\Users\\%USERNAME%\\something_random\" and click in the file browser.\nResult: Created a directory called something_random.\n\nBased on the default startup nothing other than Blender needed.", "Crash on render\n**System Information** MSI BRAVO 15\nOperating system: WINDOW 11\nGraphics card: RX 6600\n\n**Blender Version**3.6\n\nWhen attempting to render a scene in Blender, the software consistently crashes, causing frustration and hindrance to the rendering process. This issue is hindering users from efficiently creating their projects and requires immediate attention to maintain a stable and reliable rendering environment.\n\n[debug-log](https://projects.blender.orgattachment)\n[system info](https://projects.blender.orgattachment)", "Cycles HIP binaries disabled due to compiler bugs\n### Tasks\n\n- [x] Enable on Windows (#107098)\n- [x] Enable on Linux (#107662)\n- [ ] Enable Radeon VII (gfx906)\n\n### Background\n\nThere are known bugs in HIP compiler that are causing random build failures when making changes to the Cycles kernel. This is preventing developers from efficiently making improvements to Cycles.\n\nCycles AMD GPU rendering were disabled in Blender 3.6 alpha, so that ongoing work like Principled v2 is not blocked. Additionally light tree sampling is not available for HIP in Blender 3.5. We hope this can be resolved later on in the 3.6 release cycle.\n\nRelevant FIXME's in LLVM amdgpu backend (in fact I suspect we hit the same ones back when we were using OpenCL).\nSIRegisterInfo.cpp#L221\nSIRegisterInfo.cpp#L296\n\nMost recent issues:\n#104085 (#104720 )\n#104669\n#104097\nD16714", "lost 90% of my progress\nOperating system:\nGraphics card:\n\nBroken: (example: 2.80, edbf15d3c044, master, 2018-11-28, as found on the splash screen)\n\n\nBased on the default startup or an attached .blend file (as simple as possible).\n\ni was 90% through my build and had forgetten to save although i havent made saving every five minutes my priority anyways i was almost done when it freezes then crashes and all of mt work is now gone this has already happened once but i didnt lose anything so i therefore didnt report it\n\n", "adding drivers in the compositor always results in a dependency cycle which often causes blender to hang when reopening the saved .blend file.\nOperating system: Windows-10-10.0.17763-SP0 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 442.92\n\nBroken: version: 2.90.0 Alpha\n\nadding drivers in the compositor always results in a dependency cycle which causes blender to hang when reopening the saved .blend file.\n\n\nJust to clarify, the append issue I mention in the video only happens if you copy and paste 3d objects from a file with the dependancy cycle problem into a new file, save the new file, and then try and append that new file into another new file, in this instance it also results in a missing data block if trying to re-open the file you copied into.\n\n[compositor dependancy cycle on all drivers.blend](compositor_dependancy_cycle_on_all_drivers.blend)\n\n[2020-06-02 17-50-21.mp4](2020-06-02_17-50-21.mp4)\n\n", "Assert with Array() in master\nDebian stable. Git build from master today (I think 2ee6891)\n\n(Can't re-save the attached file with compression because....)\n\nReproduce:\n1. Load this blend :)\n\n[fff.blend](fff.blend)\n\n\n```\n#2 0x000000001206d923 in _BLI_assert_abort ()\n at .../blender-git/blender/source/blender/blenlib/intern/BLI_assert.c:46\n#3 0x0000000004e27bfe in blender::Array<int, 32l, blender::GuardedAllocator>::operator[] (this=0x7fffffffbfd0, index=-1)\n at .../blender-git/blender/source/blender/blenlib/BLI_array.hh:197\n#4 0x0000000004e2669e in mesh_render_data_polys_sorted_build (mr=0x7fffe747f308, \n cache=0x7fffdc48b808)\n at .../blender-git/blender/source/blender/draw/intern/draw_cache_extract_mesh_render_data.cc:235\n#5 0x0000000004e25b43 in mesh_render_data_polys_sorted_ensure (mr=0x7fffe747f308, \n cache=0x7fffdc48b808)\n at .../blender-git/blender/source/blender/draw/intern/draw_cache_extract_mesh_render_data.cc:194\n#6 0x0000000004e25aab in mesh_render_data_update_polys_sorted (mr=0x7fffe747f308, \n cache=0x7fffdc48b808, \n data_flag=(MR_DATA_LOOP_NOR | MR_DATA_LOOPTRI | MR_DATA_POLYS_SORTED))\n at .../blender-git/blender/source/blender/draw/intern/draw_cache_extract_mesh_render_data.cc:177\n#7 0x0000000004e1ff02 in blender::draw::mesh_extract_render_data_node_exec(void * __restrict__) (task_data=0x7fffe68d5920)\n at .../blender-git/blender/source/blender/draw/intern/draw_cache_extract_mesh.cc:528\n#8 0x00000000122049f6 in TaskNode::run (this=0x7fffe5f61ec0, UNUSED_input=...)\n at .../blender-git/blender/source/blender/blenlib/intern/task_graph.cc:80\n#9 0x0000000012204895 in TaskNode::TaskNode(TaskGraph*, void (*)(void*), void*, void (*)(void*))::{lambda(tbb::flow::interface11::continue_msg)#1}::operator()(tbb::flow::interface11::continue_msg) const (__closure=0x7fffe5e8ed58, input=...)\n at .../blender-git/blender/source/blender/blenlib/intern/task_graph.cc:56\n```", "AMD - Blender Crashes When I Start Render - Cycles GPU Compute\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: AMD Radeon RX 6800 XT ATI Technologies Inc. 4.5.0 Core Profile Context 23.7.1.230626\n\nBroken: version: 3.6.1\n\n[Whenever I start Rendering, It crashes when using cycles on GPU compute. Doesn't happen on CPU option.]\n\n[On render engine, select cycles. Then set it to GPU Compute. Then click on render on top left corner. Then select render image.]\n\n", "Crash on startup with LibDecor on Wayland\nAlmaLinux 9.2, probably Rocky Linux 9.2 and RHEL 9.2 as well, although I did not try those too. \"Bug for bug compatible...\"\n\nBroken: Blender 3.6.0 Beta (hash 2533576802f5 built 2023-05-25 23:31:27)\n\nBlender 3.6.0 does not use Wayland on AlmaLinux 9.2, stating Blender from the commandline produces the following messages:\n\n```\n[antonraves@mail libdecor]$ /opt/blender/blender\nGHOST/Wayland: wl_registry@2: error 0: invalid version for global zwp_pointer_gestures_v1 (15): have 1, wanted 3\nPlugin \"GTK3 plugin\" uses conflicting symbol \"png_free\".\nlibdecor-cairo-WARNING: Could not get required globals\nFailed to load plugin 'libdecor-cairo.so': failed to init\nNo plugins found, falling back on no decorations\nRead prefs: \"/home/antonraves/.config/blender/3.6/config/userpref.blend\"\nWriting: /tmp/blender.crash.txt\nSegmentation fault (core dumped)\n```\n\nInstall Blender into the /opt/blender directory and launch Blender from the command line with:\n\n`/opt/blender/blender`\n\nAnd you will see that bit of error messages appear:\n\n```\n[antonraves@mail libdecor]$ /opt/blender/blender\nGHOST/Wayland: wl_registry@2: error 0: invalid version for global zwp_pointer_gestures_v1 (15): have 1, wanted 3\nPlugin \"GTK3 plugin\" uses conflicting symbol \"png_free\".\nlibdecor-cairo-WARNING: Could not get required globals\nFailed to load plugin 'libdecor-cairo.so': failed to init\nNo plugins found, falling back on no decorations\nRead prefs: \"/home/antonraves/.config/blender/3.6/config/userpref.blend\"\nWriting: /tmp/blender.crash.txt\nSegmentation fault (core dumped)\n```\n\nDe-install libdecor from the system and Blender will run on that same system but with X11...\n\n", "Cycles OSL with OptiX\n**Status**\n\nSee also [D15902](D15902).\n\n- [x] Generate OSL closures using macros and a template file (8611c37f97)\n- [x] Remove separate OSL attribute map (ef7c9e793e)\n- [x] Make OSL implementation independent from SVM (a45c36efae)\n- [x] Update to latest OSL (to benefit from 7b4c7cffd7412a0f9b3d056f0731a9c9dcb1cf63)\n- [x] Implement all Cycles closures\n- [ ] Implement supported built-ins\n- [x] Support for textures from file paths (OIIO)\n- [x] Remove OSL dependency on OptiX and CUDA\n- [x] Contribute OSL patches upstream (1627)\n- [x] Find out why commit c8dd33f5a37b6a6db0b6950d24f9a7cff5ceb799 in OSL broke rendering in Cycles (097a13f5be)", "Make Local feature is undocumented\nBroken: version: 2.83.0\nWorked: Never\n\n**Description of error**\nMake Local feature is undocumented\n\nRight click on collection -> \"ID Data\" -> \"Make Local\" <---- no action (not intuitively obvious what the action would be)\n\n", "Can't uninstall completely and clean\nwindows 7 SP1 64 bit\n\nBroken: 2.75a c27589e\n\nCan't uninstall completely when used \"Uninstall blender\" from Start menu, and need to manually delete\n\n\nthe old version (with .exe packed ) has an option that let's you select \"leave user setting\" or \"remove all things\" when you uninstall, but the new version (with .msi packed) no this option for uninstall, and will leave some files on your computer, by the way, it's have't choice for your install, e.g. create a desktop shortcut, use blender for .blend flies, and create start menu, all blender default at install, and I can't custom setting for this, although this no affect for our used, but i'm a little OCD :D\n\nsorry for my bad English :)\n", "Translated Interface when enable Tooltips\nOperating system: Windows\nBlender Version: 2.83.15\n\nIf you enable translation \"Tooltips\" then some interface elements are translated.\nThey must be translated if you enable translation \"Interface\", not \"Tooltips\".\n\n", "Installation fails when desktop permission to write is denied\nOperating system: Windows 10\nGraphics card: N/A\n\nBroken: 3.6 lts\nWorked: Unknown\n\n**Short Description**\n\nBlender setup error : An error occurred while attempting to create the directory DESKTOP_PATH\nOptions are Retry or Cancel.\nSelecting cancel give a confirmation prompt and then rolls back the installation.\n\n- - -\nI do not want every program I install to create icons on my desktop so have remove permissions to prevent this which is why this error occurs. Even so setup should not fail simply because a desktop Icon can not be created.\n\n", "Windows/msi installer: Blender casing in the installer is lower case\nOperating system: Windows\nGraphics card: Irrelevant \n\nBroken: 3.4/3.5\nWorked: No idea\n\n\nThe Blender .msi installer on windows has a lower case 'blender' in the title bar\n\n\n\n\nUse the installer.\n\nEverything on the blender repository side is configured to have a capital B, when I build an installer locally it indeed has the right capitalization. @brecht showed me part of the build bot scripts when this last came up that forced that specific variable to lower case so this needs to be fixed on the bots to either by it having the right capitalization or to stop it from overriding this specific value.\n\n", "Linux: SSL/HTTPS Request Unable to Get Local Issuer Certificate\nOperating system: Linux-5.15.0-52-generic-x86_64-with-glibc2.35 64 Bits (Ubuntu 22.04.1)\nGraphics card: SVGA3D; build: RELEASE; LLVM; VMware, Inc. 4.1 (Core Profile) Mesa 22.0.5\n\nBroken: version: 3.3.1\nWorked: version: 2.79b, date: 2018-03-22 14:10, hash: `f4dc9f9d68b`\n\nHTTPS requests made via urllib always result in `urllib.error.URLError: <urlopen error [SSL: CERTIFICATE_VERIFY_FAILED] certificate verify failed: unable to get local issuer certificate (_ssl.c:997)>`\nCert paths point to locations that do not exist on the machine and seem to be specific to the build environment: `/home/sybren/buildbot-builder/linux_glibc217_x86_64_cmake/build_deps/deps/Release/ssl/certs`\nWhereas I'd expect it to be `/usr/lib/ssl/certs` which is the case in version 2.79b and in a separate python 3.10 interpreter.\n\nTo view cert paths:\n```\nimport ssl\nssl.get_default_verify_paths()\n```\nTo make HTTPS request:\n```\nfrom urllib import request\nrequest.urlopen('https://blender.org').status\n```" ]

[ "master compilation broken on linux - blenkernel/intern/cdderivedmesh.c: In function ‘CDDM_merge_verts’\n/home/ins/blender/source/blender/blenkernel/intern/cdderivedmesh.c: In function ‘CDDM_merge_verts’:\n/home/ins/blender/source/blender/blenkernel/intern/cdderivedmesh.c:3235:6: error: ‘v1’ undeclared (first use in this function)\n```\n v1 = (vtargetmap[last_valid_ml->v] != -1) ? vtargetmap[last_valid_ml->v] : last_valid_ml->v;\n ^\n```\n/home/ins/blender/source/blender/blenkernel/intern/cdderivedmesh.c:3235:6: note: each undeclared identifier is reported only once for each function it appears in\n/home/ins/blender/source/blender/blenkernel/intern/cdderivedmesh.c:3236:6: error: ‘v2’ undeclared (first use in this function)\n```\n v2 = mlv;\n ^\n```\n/home/ins/blender/source/blender/blenkernel/intern/cdderivedmesh.c:3287:9: warning: declaration of ‘v1’ shadows previous non-variable [-Wshadow]\n```\n uint v1 = (vtargetmap[last_valid_ml->v] != -1) ? vtargetmap[last_valid_ml->v] : last_valid_ml->v;\n ^\n```\n/home/ins/blender/source/blender/blenkernel/intern/cdderivedmesh.c:3288:9: warning: declaration of ‘v2’ shadows previous non-variable [-Wshadow]\n```\n uint v2 = (vtargetmap[first_valid_ml->v] != -1) ? vtargetmap[first_valid_ml->v] : first_valid_ml->v;\n ^\n```\nmake- [x]: *** [source/blender/blenkernel/CMakeFiles/bf_blenkernel.dir/intern/cdderivedmesh.c.o] Error 1\nmake- [x]: *** Waiting for unfinished jobs....\nmake- [x]: *** [source/blender/blenkernel/CMakeFiles/bf_blenkernel.dir/all] Error 2\nmake- [x]: *** [all] Error 2\nmake: *** [all] Error 2" ]

[ "Blender crash after maximizing/unmaximizing an editor (debug build only)\n\nBroken: master - debug build\nWorks: master - release build\nOS: Tested on both Windows and Linux\n\nWith a debug build, Blender crashes after maximizing an editor + moving mouse + unmaximizing.\nIt does not seem to be reproducible with a release build.\n\n\n- Open Blender default scene\n- Maximize the 3D viewport with CTRL+SPACE\n- Move the mouse\n- Unmaximize with CTRL+SPACE\n\n**Debug info**\nThe crash happens in `wm_event_do_handlers` when calling `wm_region_tag_draw_on_gizmo_delay_refresh_for_tweak(wmWindow *win, bScreen *screen)`.\nIn that case, `screen` points to invalid (freed?) memory.", "Blender Crashes on Render (Save As), then refuses to start\nOperating system: Windows 10\nGraphics card: NVIDIA GeForce RTX 3060\n\nBroken: 3.6.1\nWorked: N/A\n\nFirst observed with the attached file. I tried to save a png image render from it; when I clicked \"Save As,\" Blender crashed and then... wouldn't open. It seemed to initiate startup, because the little black window that usually notified me that it was reading my userprefs.py file would blink into existence just long enough for me to register what it said, then close itself and act like I hadn't clicked it.\n\nRestarting my computer fixes the problem temporarily, until the next crash. The crashes are inconsistent, but seem to happen only when I hit main menu items like Open or Save. This includes previous versions, apparently, as I still had 3.5 installed and it does the same quick blink.\n\nThe only thing I could think of was that that Read line in the black window referred to finding the userprefs file in C:\\ when I installed everything into M:\\, but I tried copying the required path over to C:\\ with no effect and also it gives me that same message when it's working and starts up fine.\n\nI also tried repairing the 3.6.1 install, and uninstalling/reinstalling, to no effect.\n\nBased on the default startup or an attached .blend file (as simple as possible).\n\n", "Hair particles crashing when baking or animating\nOperating system: windows 10 64 , windows 8 64\nGraphics card: intel, nvidia\n\nBroken: blender-2.81-9bd0d8b5505a-windows64\nWorked: (optional)\n\nwhen trying to bake a hair particles the system hangs giving this message\n\nError : EXCEPTION_ACCESS_VIOLATION\nAddress : 0x00007FF7CEBD24A4\nModule : C:\\Users\\dribeiro\\Desktop\\blender-2.81-9bd0d8b5505a-windows64\\blender.exe\n\n\nopen the file\nclick 'play' on timeline or bake on hair particles\n\nNEW INFO\nremoving the hair particles the animation runs normally\n[giselle.blend](giselle.blend)", "Blender 3.5.1 + Mac + Metal + Crash\nOperating system: MONTEREY 12.6.3\nGraphics card: AMD Radeon Pro 5500 XT 8 GB\n\nBroken: Blender 3.5.1\nWorked: Blender 3.4.1\n\nBlender 3.4.1 will render a given scene with \"RENDER PROPERTIES\" -> \"DEVICE\" set to \"GPU COMPUTE\" in 1 min 58 sec\n\nBlender 3.5 and 3.5.1 will crash on the same scene set on \"RENDER PROPERTIES\" -> \"DEVICE\" set to \"GPU COMPUTE\" - \nIt has to be run with \"RENDER PROPERTIES\" -> \"DEVICE\" set to \"CP\" and now takes 5 min 53 sec.\n\nI have lost computing power with the latest versions.\n\nJust run the file with GPU BACKEND turned on and \"RENDER PROPERTIES\" -> \"DEVICE\" set to \"GPU COMPUTE\"\n\nI could not include the file as it is a project I am working on - if there is no other way to check up on this issue I will try to create another file that I can send\n\n", "Blender 3.3 crashing unnecessarily/ unexpectedly/ frequently/ no apparent reason until now\nOperating system: Windows 10 Pro\nGraphics card: Intel i3\n\nBlender 3.3\n\n\n\nWhenever I open blender with some projects...whenever I enter material preview mode or try to render (sometimes even in solid mode...still slow) Whenever I try to apply a material and the crashes even when the project is simple and not finished or being rendered....it was working just fine and started to act weirdly during the last couple of days...I even tried, manually, to allocate more RAM to the program but that just crashed the whole thing. \n\n\nExact steps for others to reproduce the error:\n\nopen the blender file I attached as an example (The project that blender doesn't tolerate) try to apply a texture or render \n\nNOTE: I'm looking for a software solution...I don't think reproducing the error on a RTX 3090 is fair...so...we'll probably look for software solution or tweak the settings.\n\n", "Crash when importing Collada .dae file\nOperating system: macOS 10.15.7 (19H114)\nGraphics card: AMD Radeon Pro 5500M 8 GB graphics\n\nBroken: 2.91.0 (2.91.0 2020-11-25)\nWorked: N/A\n\nI exported the attached .dae file from SceneKit on iOS. When I try to open it in Blender, it crashes. It's worth noting that [Apple's documentation ](1523577-write) for the SceneKit → .dae export says it should work only on macOS, but I was able to use it on the iOS simulator. It's possible that I'm using some sort of unsupported configuration, but Blender should at least report an error instead of crashing. And if there's something wrong with the .dae file, I'd like to report it to Apple.\n\n1. Open Blender.\n2. Choose **File** → **Import** → **Collada (Default) (.dae)**.\n4. Choose the attached Juggler.dae file.\n5. Try to import with default options: all checkboxes are Unchecked under Import Data Options, Armature Options, and Keep Bind Info.\n6. Crash.\n\n[juggler.dae](juggler.dae)\n\n[Blender_2020-12-29-143229_A9918.crash](Blender_2020-12-29-143229_A9918.crash)", "Frequent render crashes in 3.5\nOperating system: macOS-12.3.1-x86_64-i386-64bit 64 Bits\nGraphics card: Metal API AMD Radeon Pro 5500M 1.2\n\nBroken: version: 3.5.0\n\nI've been encountering a lot of crashes in 3.5 when rendering in the background. Meaning I start a rendering, and move over to something like a browser. At random time Blender 3.5 just crashes. I've included a crash log file.\n\n\n", "Blender 3.5.0 Win11 RTX3090 Fatal Crash on activating Cycles\n@deadpin thanks for replying.\n\nDriver Version: 31.0.15.3161 (NVIDA 528.24) DCH Win1164 installed\n\nCycles now working without issue thank u.\n\n_Originally posted by @Jaymi-McManus in /blender/blender/issues/106953#issuecomment-923094_\n\n>>>>>\n\nUpdate: \n\n1. Closed the ticket too soon, after adding some simple color, specular, metallic, changes in BSDF and then navigating between layout and other tabs, fatal crash occurs again\n\n2. Updated NVIDIA drivers to 31.0.15.3161 (NVIDIA 531.61) DHC / Win11 64 the issue still persists\n\nUpdate logs attached.", "Blender crashes: OpenVDB math::Transform::read throws exception. \nOperating system: Darwin-19.5.0-x86_64-i386-64bit 64 Bits\nGraphics card: AMD Radeon Pro 5500M OpenGL Engine ATI Technologies Inc. 4.1 ATI-3.9.15\n\nBroken: version: 2.83.2\n\nBlender keeps crashing.\n\nI have a project were I created an animation of modified arrays. Everytime I play it it crashes\n\nCrash report: P1536\n\ncrashed thread: \n```\n\tThread 35 Crashed:\n0 libsystem_kernel.dylib \t0x00007fff726c733a __pthread_kill + 10\n1 libsystem_pthread.dylib \t0x00007fff72783e60 pthread_kill + 430\n2 libsystem_c.dylib \t0x00007fff7264e8a2 __abort + 139\n3 libsystem_c.dylib \t0x00007fff7264e817 abort + 135\n4 libc++abi.dylib \t0x00007fff6f8ae458 abort_message + 231\n5 libc++abi.dylib \t0x00007fff6f89f8bf demangling_terminate_handler() + 262\n6 libobjc.A.dylib \t0x00007fff713da5b1 _objc_terminate() + 104\n7 libc++abi.dylib \t0x00007fff6f8ad887 std::__terminate(void (*)()) + 8\n8 libc++abi.dylib \t0x00007fff6f8b01a2 __cxxabiv1::failed_throw(__cxxabiv1::__cxa_exception*) + 27\n9 libc++abi.dylib \t0x00007fff6f8b0169 __cxa_throw + 113\n10 org.blenderfoundation.blender \t0x0000000110291c61 openvdb::v7_0::math::Transform::read(std::__1::basic_istream<char, std::__1::char_traits<char> >&) + 4801\n11 org.blenderfoundation.blender \t0x000000011026082b openvdb::v7_0::io::Archive::readGrid(std::__1::shared_ptr<openvdb::v7_0::GridBase>, openvdb::v7_0::io::GridDescriptor const&, std::__1::basic_istream<char, std::__1::char_traits<char> >&) + 1787\n12 org.blenderfoundation.blender \t0x0000000110286589 openvdb::v7_0::io::File::Impl::unarchive(openvdb::v7_0::io::File const&, std::__1::shared_ptr<openvdb::v7_0::GridBase>&, openvdb::v7_0::io::GridDescriptor const&, openvdb::v7_0::io::File::Impl::NoBBox) + 57\n13 org.blenderfoundation.blender \t0x0000000110282c91 openvdb::v7_0::io::File::readGridByName(std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> > const&, openvdb::v7_0::math::BBox<openvdb::v7_0::math::Vec3<double> > const&) + 465\n14 org.blenderfoundation.blender \t0x0000000110282ab6 openvdb::v7_0::io::File::readGrid(std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> > const&) + 54\n15 org.blenderfoundation.blender \t0x00000001128a2cc9 MANTA::updateGridFromVDB(std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> >, float*, bool) + 361\n16 org.blenderfoundation.blender \t0x0000000112890985 MANTA::updateGridFromFile(std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> >, float*, bool) + 869\n17 org.blenderfoundation.blender \t0x0000000112890cee MANTA::updateNoiseStructures(FluidModifierData*, int) + 558\n18 org.blenderfoundation.blender \t0x000000011287e2b5 manta_update_noise_structures + 21\n19 org.blenderfoundation.blender \t0x000000010df34cd0 BKE_fluid_modifier_processDomain + 3168\n20 org.blenderfoundation.blender \t0x000000010df316ce BKE_fluid_modifier_do + 142\n21 org.blenderfoundation.blender \t0x000000010e3ad4f6 applyModifier + 54\n22 org.blenderfoundation.blender \t0x000000010dea5b0a mesh_calc_modifiers + 2778\n23 org.blenderfoundation.blender \t0x000000010dea4d6b mesh_build_data + 123\n24 org.blenderfoundation.blender \t0x000000010dea4035 makeDerivedMesh + 133\n25 org.blenderfoundation.blender \t0x000000010dfe1281 BKE_object_handle_data_update + 1089\n26 org.blenderfoundation.blender \t0x000000010dfe1704 BKE_object_eval_uber_data + 52\n27 org.blenderfoundation.blender \t0x000000010e9f36ba DEG::(anonymous namespace)::deg_task_run_func(TaskPool*, void*, int) + 138\n28 org.blenderfoundation.blender \t0x00000001142cadbc task_scheduler_thread_run(void*) + 364\n29 libsystem_pthread.dylib \t0x00007fff72784109 _pthread_start + 148\n30 libsystem_pthread.dylib \t0x00007fff7277fb8b thread_start + 15\n```\n", "LineArt; Crash after press play in a scene with LineArt modifier\nOperating system: Linux-5.4.0-105-generic-x86_64-with-glibc2.27 64 Bits\nGraphics card: GeForce GT 630/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 390.144\n\nBroken: version: 3.2.0 Alpha\n\nNo idea what crashes blender. Just press play and wait some seconds.\n\nI would love to help solving this issue, so when someone could remind me where the crashreport usually lands in Linux, i could share that as well.\n[Narrator.blend](Narrator.blend)\n\n", "Blender Crash. Code marked as unreachable has been executed.\nOperating system: Windows-11-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1080/PCIe/SSE2 NVIDIA Corporation 4.6.0 NVIDIA 536.67\n\nBroken: version: 3.6.1, 4.0.0 Alpha\nWorked: didn't test\n\nOpen the .blend and go to frame 60. Crash\n\nThis is a .blend from a random user. So, sadly, I don't know the exact steps to reproduce.\nIt just said: \"Code marked as unreachable has been executed. Please report this as a bug.\"\n\n", "2.9.1 Close without error message\nOperating system: Windows 10, 10gb ram, ssd\nGraphics card: Nvidia GTX 650 1gb vram\n\nBroken: (2.83 LTE and 2.90 current available version)\nWorked: (2.82)\n\nWhen rendering with GPU AND CPU, the app crashes, works perfectly fine on CPU only. On GPU only too but slower than older versions.\n\n\n```\nblender.exe :0x00007FF7A5B76440 ccl::kernel_cpu_sse41_filter_finalize\nblender.exe :0x00007FF7A57A9220 ccl::CPUDevice::denoising_solve\nblender.exe :0x00007FF7A57DA470 ccl::DenoisingTask::reconstruct\nblender.exe :0x00007FF7A57D8ED0 ccl::DenoisingTask::run_denoising\nblender.exe :0x00007FF7A57A7240 ccl::CPUDevice::denoise_nlm\nblender.exe :0x00007FF7A57AB590 ccl::CPUDevice::thread_render\nblender.exe :0x00007FF7A57ABA30 ccl::CPUDevice::thread_run\nblender.exe :0x00007FF7A57A60B0 ?_Do_call@?$_Func_impl_no_alloc@V<lambda_d884e08a7877f415f05ce8ffda8f97b4>@@X$$V@std@@EEAAXXZ\nblender.exe :0x00007FF7A7D5F520 tbb::internal::function_task<std::function<void __cdecl(void)> >::execute\ntbb.dll :0x00007FFBF5F037A0 tbb::recursive_mutex::scoped_lock::internal_try_acquire\ntbb.dll :0x00007FFBF5F037A0 tbb::recursive_mutex::scoped_lock::internal_try_acquire\ntbb.dll :0x00007FFBF5EF51D0 tbb::interface7::internal::isolate_within_arena\ntbb.dll :0x00007FFBF5EFA490 tbb::task_scheduler_init::terminate\ntbb.dll :0x00007FFBF5F019C0 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFBF5F019C0 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFC08611FA0 o_exp\nKERNEL32.DLL :0x00007FFC0B647960 BaseThreadInitThunk\nntdll.dll :0x00007FFC0B7AA250 RtlUserThreadStart\n```\n\nBased on the default startup or an attached .blend file (as simple as possible).", "Crash while switching to edit mode with EEVEE + Geometry Nodes\nOperating system: macOS-10.15.7-x86_64-i386-64bit 64 Bits\nGraphics card: NVIDIA GeForce GT 650M OpenGL Engine NVIDIA Corporation 4.1 NVIDIA-14.0.32 355.11.11.10.10.143\n\nBroken: version: 3.0.0\n\nWhen switching to edit mode in while the viewport's set to render it crashes (see attached video)\n[blender crash 2021-12-17 21-54-00.mp4](blender_crash_2021-12-17_21-54-00.mp4)\n\n1 load the attached file\n2 switch to rendered view in the viewport\n3 switch to edit vertex mode\n4 that's a crash!\n\n[transitions_001_blender3.blend](transitions_001_blender3.blend)", "Blender crashes as soon as an image texture is plugged in the material BSDF node\nOperating system: Windows 10\nGraphics card: NVIDIA GeForce GTX 1080 Ti\nSee attached log for more detail\n\nBroken: 3.5.1, but also 4.0.0 Alpha, did not try intermediate versions\nWorked: ??? The earliest available build, 2.93.18, also causes the same crash\n\nThe second any kind of image texture is plugged in a material, Blender crashes without any warning. Occurs when creating a new image texture, when plugging an existing file, or even when trying to plug a UV grid for unwraping purposes\n\n**Similar tickets**\n#76193\n#76377\n\n1. Start Blender.\n2. Create new default file.\n3. Select the default cube.\n4. Switch to Shading workspace.\n5. Add Image Texture node.\n6. Create new image or select an existing one.\n7. Plug Color output to the Base Color input of the Principled BSDF node.\n8. Watch Blender shut down.", "Blender Crashing When Painting Stroke into a \"Moving Animated Character\" (heap-use-after-free)\nBroken: 4.0 alpha\nBroken: version: 2.79 - 2.82 (sub 7)\nWorked: Never\n\nI have this experiment of making LIVE talking head, via OSC. And I noticed that if I tried painting a stroke on a talking moving animated head, it will crash. Even with timeline not running.\n\nI got a video (Might need iPhone X with MOM addon to test this....) :\njhHdG6ZXFDU\n\n- Open attached file\n- Run animation\n- Paint over the animated object\n[paint_animated_shapekey_bug.blend](paint_animated_shapekey_bug.blend)\n\n\n" ]

[ "Blender crashes on opening file explorer window (Open, Save as, Import, Export)\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 471.96\n\nBroken: version: 2.93.3\n\nWhenever I open a file browser window from Blender (e.g.: Export, Import, Open, Save As, etc.), it crashes. For example when I click on File>Open, it seems like many file explorer windows are being opened in quick succession on top of each other until the \"Blender File View\" titled window appears on top and the program immediately exits. This whole process takes about 3-4 seconds and renders the program unresponsive.\n\nThe error occurs when the desktop is extended to my TV (4K TV, Marked by Windows as \"Screen 1\", set as secondary screen, connected via HDMI through the video card), and the (Export, Import, Open, etc.) file browser window is opened while Blender's window is maximized on \"Screen 2\" (FullHD monitor, set as the primary screen, connected via DVI-I through the video card).\n\nIf Blender is maximized on the extended desktop on the TV, the error will not occur. \nIf Blender is not maximized on the primary monitor screen, the error will not occur.\nIf the TV is set as the primary screen, Blender can be maximized on either of the screens and the error will not occur.\n\n\n1. Extend Windows desktop so that primary screen is FullHD monitor, while extended screen is 4K TV.\n2. Start Blender\n3. Maximize Blender window on FullHD monitor primary screen.\n3. Click File>Open\n\n" ]

[ "Node muting connects mismatching sockets\nBroken: 3.2.0 beta\n\n\n - Add *Float Curve* node in any editor (*Shader, Geometry Nodes*) and connect output to anything else\n - Plug *Value* nodes into input sockets (both factor and value)\n - Mute *Float Curve* -> inner noodle connects Factor to Value\n\nOr in provided file mute *Float Curve*\n\n[wrong_mute.blend](wrong_mute.blend)\n", "When changing the width of frame, it also changes frame location.\nIf you create frame and set it into (0,0) location. Then grab right side of the frame node and make frames width bigger. Now using python read the frame \nlocation again and it is not (0,0) anymore. \n\n", "Keyframing Stretch property in IK won't work properly - iTaSC Solver\nDoesn't work in all versions 2.79b, 2.83, 3.00\n\nWhen using iTaSC solver for an armature, you can't keyframe or use a driver for the bone stretch property. If you do it will be buggy and won't work as expected.\n\nThe Stretch property I'm talking about\n\n\n1. Make a a small armature with iTaSC Solver\n2. Add IK Bone Constraint\n3. Animate the stretch property whether in the bone constraint or the bone property\n4. Keyframe your rig and see that it's not behaving properly\n\n**note: I know it may seem incomprehensible but if you click constantly \"Update Dependencies\" in the drivers panel you will see the expected result**\n\n[Bug report.blend](Bug_report.blend)", "Normal map support in Cycles standalone\nWindows 8.1 64-bit\n\n**Cycles Version**\nd553b8de7180\n\n\nIn the latest version available in the Cycles standalone repository, commit d553b8de7180, using a normal map node in a shader graph will cause Cycles to crash. This behavior was not present in release 1.7.0 but has been added since then.\n\nThis looks to be caused by an improperly initialized member variable in the ccl::NormalMapNode type. Upon creating an object of type NormalMapNode, the ustring named 'attribute' is initialized such that attribute.c_str() will return a nullptr.\n\nCycles crashes when the nullptr used to construct a std::string in NormalMapNode::attributes in this bit:\n\n```\t\n\t\telse {\n\t\t\tattributes->add(ustring((string(attribute.c_str()) + \".tangent\").c_str()));\n\t\t\tattributes->add(ustring((string(attribute.c_str()) + \".tangent_sign\").c_str()));\n\t\t}\n```\n\n\nI have attached a simple cycles standalone scene that triggers this error here: [sample_scene.xml](sample_scene.xml)", "Skipping node auto-attachment doesn't work with custom keymap\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3060 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 526.86\n\nBroken: version: 3.5.0 Alpha\nWorked: none\n\nBy default holding `ALT` while dragging a node disables auto-attachment. However, this doesn't work when using a custom keymap that has been imported from previous version. The prompt doesn't show which key to hold, and holding `ALT` just makes the prompt flicker like crazy, but doesn't disable auto-attachment.\n\n\n\n\n**Steps to reproduce the bug**\n\n - Switch to a keymap that has been created with an earlier Blender version\n - Move a node in any node editor\n - Hold `ALT`\n\n\n\n", "Small Custom Bone Shapes have some Wireframes Artifact.\nWindows 10 64 Bits.\nGraphics card GeForce GTX 1080 Ti \n\nBroken: version: 2.82.6, hash: a854840e76ae\nWorked: (optional)\n\nSmall custom bone shape has wireframe glitching, the Scale is based on real world but if you resize it in/out u can see it happen.\n\n{[F8257584](Shape.gif) size = full}\n\n\n\n[Bug.blend](Bug.blend)\n", "Dynamic paint sub-steps don't work when the brush position is set in geometry nodes\n\nI believe this **may** be the same issue as T40597\n\nIn order to have a brush follow a potentially very large collection of curves in sequence, I'm using geometry nodes to set the position, which avoids the pain of having to join the curve paths or painstakingly tweak the offsets and influence keyframes for each and every one.\n\nThis works, but with sub-steps enabled it seems that the nodes are still only being updated once/frame. Or, if they are being updated, then the `seconds` input from the `Scene Time` node is being derived from the `frames` value which is clamped to an integer.\n\n\n[text_path_test.blend](text_path_test.blend)\n", "Keyed Particles don't render as expected\nOperating system: Windows-10-10.0.17763 64 Bits\nGraphics card: Intel(R) HD Graphics 4000 Intel 4.0.0 - Build 10.18.10.5069\n\nBroken: version: 2.81 (sub 16)\nWorked: (optional)\n\n[KeyedParticlesJL.blend](KeyedParticlesJL.blend)\n\n- Open file\n- Go to frame 105\n- Render\n- Compare render to viewport\n\nRender is different to viewport\n\nThat's what I expected:\n[ViewportRender.mp4](ViewportRender.mp4)\n\nThat's what I get when rendering with Eevee:\n[EeveeRender.mp4](EeveeRender.mp4)\n\nThat's what I get when renderng with Cycles:\n[CyclesRender.mp4](CyclesRender.mp4)\n", "Weighted normals modifier produces different results on objects of different scales\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: GeForce GTX 1050 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 457.51\n\nBroken: version: 3.3.0\n\n\nWeighted normals modifier is not consistant on small objects \n\nOpen Blender\nCreate a cube 1mmx1mmx1mm, scale applied, \nadd bevel and weighted normals modifiers\nYou can notice the différence by doing the same process on a 1mx1mx1m cube\n\nI can do a work around by modifying the unit scale parameter but I think it should work on 1mm objects, like a lot of watch pieces. \n\n[weightednormals_defect.blend](weightednormals_defect.blend)\n", "Depsgraph deferred evaluation design\nFor heavy modifiers, evaluation will take a lot of time, while the UI is locked, which affects user experience. Deferred (background) evaluation will allow:\n\n1. UI to be responsive while calculation is happening in the background.\n2. Showing progress in the UI.\n3. Other interactive stuff (?)\n\nThis function will eventually be needed anyway so I set up a task here while I poke around the code to find a doable design.\n", "Transform generalized custom-data correction support\n\n**Motivation:**\n\nHaving to manually adjust custom-data is tedious\nwhile there is a UV correction option,\nit's limited to vertex & slide.\n\nThis proposal is to generalize the option so it can be used for any\ntransform mode.\n\n\n**Solution:**\n\nThis proposes to add two kinds of custom-data correction methods.\n\n- Correct based on surrounding geometry, this is suitable for\n```\nvertex & edge slide.\n```\n *The existing logic for vertex & edge slide can be used here.*\n\n- Project Custom-Data\n\n```\nThis would project geometry onto the surfaces of the original mesh\nto calculate the custom-data.\nThis would allow geometry.\n```\n\n```\nThis could be used for other transform modes besides vertex/edge slide.\n```\n\n\n**Limitations:**\n\nThis option works best when adjusting small regions of geometry and often won't work well moving large regions of faces.\n\nThere is a limit to how much UV's can be corrected in this case without re-unwrapping/projecting\n*(which is beyond the scope of this proposal).*\n\n\n**User Interface:**\n\nThese would be shown in the tool options panel, we could keep the transform option too (although this isn't essential).\n\n----\n\nThere is a risk that users accidentally leave this enabled and transform their UV's when moving the geometry.\n\nSince we have other fairly obscure options (transform origins for example), it may be best to show these are enabled while transforming (as we do for auto-keyframe).\n", "Relative space\nCurrently when users change the render size percentage of a scene the compositor tree needs to be adjusted as a lot of composite nodes aren’t relative. This has a lot of impact on performance of the user and the system. This is part of T74491.\n\nA solution can be to introduce a Pixel factor to any buffer in the compositor. The term Pixel Factor is not really correct, but I couldn’t come up with a better term. (suggestions are welcome).\nThe pixel factor is the scale factor between the used input buffer and the input buffer that would have been used when the render resolution percentage would have been set to 100%.\n\nFor render layers the pixel factor is the same as the render resolution percentage. For movie clips and images a pixel factor of 1 is used.\n\nNodes that use parameters in pixels could use this pixel factor to adjust its effect. Nodes that use multiple input buffers would select a pixel factor for the output buffer. This depends on the node, for example when mixing a render layer on a movie clip, the pixel factor closest to the pixel factor that would be used by the compositor node would be selected for the output buffer.\n\nThe compositor output node would scale the input buffers to match the scene render percentage.\n\nThere are nodes where the pixel factor won’t work:\n\n```\n Glare Node: Iteration factor could be scaled, but could lead to different results.\n Filter Node: Uses a Convolution filter. When working on scaled down buffers this could lead to visual artifacts\n Despeckle: Uses arithmetic with the neighbour pixels. Might not have visual artifacts.\n```\n\nThere is new compo implementation compositor-up but it seems that this branch doesn’t tackle this part of the ticket.\n\nQuestions:\n\n```\n Is having a pixel factor the way to solve the problem?\n What should be done with the nodes that cannot work with a pixel factor?\n Should the Image/Movieclip pixel factor be dependent on the compositor output resolution? (Quality vs performance)\n", "Keyframing Corner pinning in compositing is inconsistent\nOperating system: windows 10\nGraphics card: Nvidia geforce GTX 1050 Ti\n\nBroken: 3.3.0\n\n\nWhen using the background X symbols to drag and drop corners with corner pinning, only the topmost value of each vector is keyframed\nExample image(wanted to edit a meme):\n\n\n- Open attached file\n- Single Render to Update Compositor\n- Move the Corner Pins\nor\n- Arrange nodes as seen.\n```\n{F13601679}\n```\n- Select the corner pin node\n- Try to keyframe the location of the second image/video, using auto keyframes and the background viewport manipulators.\nOnly the topmost (X) value gets keyframed\n\n[Corner pin.blend](Corner_pin.blend)\n", "Frame counter while rendering does not respect blender's DPI setting\n4K screen, Windows 10, GEFORCE 960M\n\nBroken: bac1279\n\nThe cursor frame counter that displays while rendering is minuscule on a 4K screen, regardless of the DPI setting in Blender", "Occasional render glitches when using simulation zones on an object without a material.\nThis bug has an easy workaround, mostly posting this so anyone running into the same issue knows the solution.\n\nOperating system: Linux-5.15.0-79-generic-x86_64-with-glibc2.35 64 Bits, X11 UI\nGraphics card: NVIDIA GeForce RTX 3080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 530.30.02\n\nBroken: version: 3.6.2\nWorked: N/A, 3.6 is the first version with simulation nodes.\n\nIn Eevee, rendering an object with a simple simulation zone applied in geometry nodes and no material assigned to the object results in some rendered frames being blank or displaying odd glitches.\n\nThe workaround is to simply assign a material to the object.\n\nCheck out the attached .blend file for reproduction.\n\n" ]

[ "Keyframes and drivers on custom nodes properties seem not working\nOperating system: Darwin-18.2.0-x86_64-i386-64bit 64 Bits\nGraphics card: Intel(R) Iris(TM) Graphics 540 Intel Inc. 4.1 INTEL-12.4.7\n\n\nBroken: version: 2.80 (sub 74)\nWorked: (optional)\n\n\nI'm not able to animate the properties of custom nodes. Keyframes do not work properly: when I set one all the others seem to disappear and only the last value set as a keyframe is shown on every time step.\n\n1. Go to the text editor, import the 'Custom Nodes' template and run the script.\n2. Open the 'Custom Node Tree' area.\n3. Add a 'Custom Node', in the 'Some Nodes' category.\n4. Add a keyframe for 'my_float_prop' on frame 1.\n5. Change the frame, change 'my_float_prop' value and set a keyframe.\n6. Now I should see (at least it was like this in the last version) the property changing while scrolling through the timeline between the two keyframes I just set, but I don't. The property keeps the last value.\n", "Drivers do not update Properties in custom pyNode trees\nN/A\n\nBroken: 2.83\nWorked: Never?\n\nWhen using a custom pyNode tree, setting a property to use a driver does not update that property.\n\nRequires an add-on that uses a custom pyNode tree. (You can use mine from BakeWrangler_1_0_RC5.zip). Open the 'Bake Node Editor' and create a new node tree. Right click on any property and 'Copy as New Driver', then right click on some other property in the node tree and 'Paste as Driver'. Changing the original property will not update the driven property. (It is also possible to modify the driven property, it generates an error message but changes the value any way)" ]

[ "Crash when importing a file using python from the command line with an asset browser open\nOperating system: Linux-5.15.0-40-generic-x86_64-with-glibc2.35 64 Bits\nGraphics card: AMD Radeon RX 5700 XT (navi10, LLVM 13.0.1, DRM 3.42, 5.15.0-40-generic) AMD 4.6 (Core Profile) Mesa 22.0.1\n\nBroken: version: 3.3.0 Alpha\nWorked: -\n\nBlender crashes with segfault when trying to import a model using `blender --python-expr` when the startup file or the opened file has an asset browser open.\n\nDownload [sample model ](Sponza) from KhronosGroup github.\n\nRun blender with:\n\n```\nblender <path/to/file/with/asswet/browser/open> --python-expr 'import bpy; bpy.ops.import_scene.gltf(filepath=\"/<PATH>/Sponza.gltf\")'\n```\n\nBlender crashes with the follow stack trace:\n\n```\n* thread #1, name = 'blender', stop reason = signal SIGSEGV: invalid address (fault address: 0x5c0)\n * frame #0: 0x0000000004b21c10 blender`filelist_file_ex(filelist=0x0000000000000000, index=-1, use_request=true) at filelist.c:2231:36\n frame #1: 0x0000000004b22070 blender`filelist_file(filelist=0x0000000000000000, index=-1) at filelist.c:2277:10\n frame #2: 0x0000000004b2914e blender`ED_fileselect_active_asset_get(sfile=0x00007fffdaf50988) at filesel.c:472:30\n frame #3: 0x0000000004b2f636 blender`file_listener(listener_params=0x00007fffffffe7a0) at space_file.c:467:38\n frame #4: 0x0000000003ca4f3c blender`ED_area_do_listen(params=0x00007fffffffe7a0) at area.c:163:5\n frame #5: 0x0000000003259301 blender`::wm_event_do_notifiers(C=0x00007ffff3f84388) at wm_event_system.cc:619:11\n frame #6: 0x000000000324f100 blender`WM_main(C=0x00007ffff3f84388) at wm.c:626:5\n frame #7: 0x0000000002a73fe9 blender`main(argc=4, argv=0x00007fffffffe978) at creator.c:547:5\n frame #8: 0x00007ffff7dba290 libc.so.6`__libc_start_call_main + 128\n frame #9: 0x00007ffff7dba34a libc.so.6`__libc_start_main@@GLIBC_2.34 + 138\n frame #10: 0x0000000002a73ba5 blender`_start + 37\n```\n\n**Additional information**\n\nBlender does not crash if the same command is run in the python console when blender is opened.", "Startup.blend file is not loaded on startup if userpref.blend does not exist.\nOperating system: Windows-10-10.0.17763 64 Bits\nGraphics card: Intel(R) UHD Graphics 620 Intel 4.5.0 - Build 26.20.100.7000\n\nBroken: version: 2.81 (sub 3)\nWorked: (optional)\n\nStartup.blend file is not loaded on startup if userpref.blend does not exist.\n\n- Remove userpref.blend file from it's usual location\n- Start blender\n- Dismiss splash screen by clicking on 3D viewport\n- Make some change to the file (move the cube)\n- Save as startup file\n- Close Blender\n# Open blender\n\nStartup file is not loaded", " Startup File: Header Collapse Menu not restored after restarting blender\nOperating system: windows 10 64bit\nGraphics card: Nvidia gtx m860\n\nBroken:\nversion: 2.80 (sub 37), branch: blender2.8, commit date: 2018-12-16 00:27, hash: 4488056f3d9, type: Release\n\nWorked: (optional)\n2.79b\n\nwhen using Collapse Menu on the header and restarting blender the setting is ignored and the header menu is expanded again.\n\nBased on the default startup or an attached .blend file (as simple as possible).\n\n1. open blender 2.8\n2. collapse the header menu\n \n3. save startup file so we save the header configuration\n\n4. restart blender\n5.the Header menu is expanded again after a restart, other headers like the editor menu stay collapsed after a restart.\n\n", "Material Tab freeze still\n|**Operating system**:|Linux version 5.11.11-200.fc33.x86_64 ([email protected]) (gcc (GCC) 10.2.1 20201125 (Red Hat 10.2.1-9), GNU ld version 2.35-18.fc33)\n| -- | -- |\n|**Graphics card**:|OpenGL version string: 4.3 (Compatibility Profile) Mesa 20.3.5\n||NVIDIA Corporation GP104 [GeForce GTX 1080] (rev a1) (prog-if 00 [VGA controller])\n\nBroken: 2.83lts - 2.92.0 (Blender version has been downloaded from the Snap Store).\nThis same freeze also happens if using Blender from the Software store (2.91.2) or via the \"dnf\" command.\nWorked: N\\a\n\nIn Eevee, when opening the material tab, either blender crashes or my entire system freezes up and becoming completely unusable and requires a hard reset.\nwatch?v=te6IaP12Gbg&ab_channel=WhitequillRiclo\n\nPerhaps it is related to the compilation of shaders in the material icon.\n\n- Open the Material tab", "Reference image opacity value doesn't work with Blender 2.83+ (Intel HD 630 with Mesa 20, regression from D6729)\nOperating system: Mageia 8 x86_64 (Linux)\nGraphics card: Intel HD Graphics 630, Mesa 20.1.0\n\nBroken: 2.83.0, 2.90.0-aed11c673efe\nWorked: 2.82a, 2.83-ce76e17584ee\n\nIn Blender 2.82a, when using a reference image with opacity between 0.0 and 1.0, the image is properly blended using the opacity value, provided that \"Use Alpha\" is enabled. (I.e. an image with opacity 0.20 will be less visible than an image with opacity 0.80, though both are transparent - like in most image editors.)\nA few weeks back I used 2.83-ce76e17584ee, and there the \"Use Alpha\" setting was renamed to \"Transparency\", but the behavior stayed the same.\n\nOpacity working in Blender 2.82a:\n\n\nAfter updating to 2.83.0 (stable) now, I see that opacity no longer works as intended, it behaves like an on/off switch without any blending.\nAn image will look the same with opacity 0.0001 or 1.0 (and will be invisible with opacity 0).\n\nOpacity not working in Blender 2.83.0 (same .blend):\n\n\nCurrent 2.90 alpha seems to suffer from the same issue.\n\n- Add > Image > Reference\n- Pick an image\n- In object properties, enable Transparency, and try different values of Opacity. Opacity doesn't impact the blending of the reference image in the viewport (unless it's 0, which makes it disappear).", "Blender Instantly Crashes When Rendering Cycles Or Clicking Wrong Setting\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nCPU: 12th Gen Intel(R) Core(TM) i7-12700KF, 3600 Mhz, 12 Core(s), 20 Logical Processor(s)\nGraphics card: AMD Radeon RX 6800 XT ATI Technologies Inc. 4.5.0 Core Profile Context 23.9.1.230821\n\nBroken: version: 3.6.2\n\nWhenever Blender attempts to render in cycles mode it exits immediately. It doesn't try for a few seconds, there's no loading, just click and instantly the process dies. This happens whether Preferences>System>Cycles Render Devices is set to \"None\", \"HIP\" with my GPU selected, or \"HIP\" with my CPU selected. Regardless of the cycles render device setting when I turn on viewport shading it works as it defaults to Eevee. However, if I try to change it to cycles, the moment I click cycles Blender crashes.\n\nThis same instant crash happens when I try to click OptiX or CUDA in Preferences>System>Cycles Render Devices instead on None or HIP.\n\n\n- Start Blender with a default project\n- Press Z then 8 to go into rendering mode\n- On the sidebar click render properties\n- Click the dropdown to change render engine\n- Click Cycles\n- Should see it crash instantly\n\nTo see the settings crash\n- Start Blender with a default project\n- Navigate to Edit>Preferences>System>Cycles Render Devices\n- Click between None, HIP, and OneAPI without issue\n- Click CUDA or OptiX and see the instant crash\n\n", "Blender 2.90/ 2.91 Sculpt Vertex Paint - Stencil not working\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 750 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 446.14\n\nBroken: version: 2.90.0 Beta, branch: master, commit date: blender-2.90.0-0e280b96ca82-windows64\nWorked: I think this last work in 2.83.2 or 2.83.1\n\nSculpt Vertex Paint - Stencil not working\n\n\n[BlenderBug.blend](BlenderBug.blend)\n\nAdded colored image to texture\nChange texture mapping to stencil\nImage is coming out as black and white on the screen and image is not being transferred on the object.\n\nI have tried changing Blender Mode and Color Space but it didn't work.\n\nAlso I am not sure if this suppose to happen but the image for stencil disappears when I tick Color Ramp under Color in Texture Panel.\n\n", "Blender 3.0.0 breaks compatibility of rigs with the previous version\nOperating system: Windows 11 Pro 21H2\nGraphics card: Nvidia RTX 3070\n\nBroken: 3.0.0\nWorked: 2.93.6\nBlender 3.0.0 breaks compatibility of rigs with the previous version.\n\n1. Open Agent 327 shot 10_03_B-agent_dodges animation file (10_03_B.anim.blend)\n2. Observe that the rigs are completely broken\n\n\n", "Grid doesnt show\nOperating system:\nGraphics card:\n\nBroken: (example: 2.80, edbf15d3c044, master, 2018-11-28, as found on the splash screen)\n\n\nBased on the default startup or an attached .blend file (as simple as possible).\n\n", "Blender 3.6.3 SegFault Crash at startup on MacOS Ventura 13.5.2\nOperating system: MacOS Ventura 13.5.2 (22G91)\nGraphics card: Intel Iris Plus Graphics 650 1536 Mo\n\nBroken: 3.6.3 (3.6.3 2023-09-21)\nWorked: None tried on this machine\n\nCrash on startup with `Segmentation fault: 11`\n\nEither:\n- Double-click on Blender application icon\nOr:\n- command line : `/Applications/Blender.app/Contents/MacOS/Blender` with or without option `--gpu-backend`\n\n```\nThread 0 Crashed:: Dispatch queue: com.apple.main-thread\n0 Blender \t 0x109d2903d (anonymous namespace)::Local::Local() + 29\n1 Blender \t 0x109d28fe9 memory_usage_init + 25\n2 Blender \t 0x109d26f09 MEM_init_memleak_detection + 9\n3 Blender \t 0x101e098c3 main + 227\n4 dyld \t 0x7ff815a5241f start + 1903\n```", "Blender 3.5.0 Win11 RTX3090 Fatal Crash on activating Cycles\n@deadpin thanks for replying.\n\nDriver Version: 31.0.15.3161 (NVIDA 528.24) DCH Win1164 installed\n\nCycles now working without issue thank u.\n\n_Originally posted by @Jaymi-McManus in /blender/blender/issues/106953#issuecomment-923094_\n\n>>>>>\n\nUpdate: \n\n1. Closed the ticket too soon, after adding some simple color, specular, metallic, changes in BSDF and then navigating between layout and other tabs, fatal crash occurs again\n\n2. Updated NVIDIA drivers to 31.0.15.3161 (NVIDIA 531.61) DHC / Win11 64 the issue still persists\n\nUpdate logs attached.", "MetalRT: Rendering being unresponsive with AO node\nOperating system: Ventura 13.4.1\nGraphics card: Apple M1 Max 32 Core\n\nBroken: Blender 3.6 LTS\nWorked: Blender 3.5.1\n\nAfter downloading the newest Version of Blender it seems that using the Ambient Occlusion Node need leads to Blender being unresponsive. Sample count stays on 0 and neither Viewport nor F12 Render is starting as expected. Trying to quit, Blender crashes completely.\n\n- Add new material\n- Add Ambient Occlusion Node\n- Select cycles and GPU\n- Enable MetalRT\n- Start Viewport Rendering.\n\n", "Select Loop Inner-Region issue\nWin8 64bit | GTX680\n\nBroken: 2.77 509270e\n\nWhen working with face loop, this tool doesn't always work as expected.\n\n\n\n\nNot sure if this can be a bug. Please check.", "Shader compilation freeze Blender on Intel Arc770\nOperating system: Windows 10\nGraphics card: Intel Arc770 16Gb (Resizable Bar enabled)\n\n3.6.0\nDate: 2023-06-27 08:08\nHash: c7fc78b81ecb\nBranch: blender-v3.6-release\n\n**Intel driver version**\n31.0.101.4665\nReleased 11.08.2023 (latest on moment of this post)\n\nAll modes of viewport on new scene (default, for example) works fine: solid, material preview, display render. But, if I open my scene, works only 1 and 3 option. Building a shaders freezes Blender. It's around 800+ materials on scene, and GeForce cards works fine with that. The video card works at full power, with enabled Resizable Bar function. Cycles render (oneAPI) also works fine, but i can't switch the viewport on material preview. Blender freezes when shaders start compile.\n\n", "CMD terminal stays open after opening blender\nOperating system: Windows-10-10.0.25247-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2080 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 526.86\n\nBroken: version: 3.5.0 Alpha\n\nI'm not sure why it';s happening but it's not happening for 3.3 version\n\n\n\n" ]

[ "System freezes completely on staring Blender 2.8 (Ubuntu using RX 5700)\nOperating system: Ubuntu 18.04 LTS and 19.04\nGraphics card: Sapphire Radeon RX 5700 (\"reference card\")\n\nBroken: tested (2.81 Alpha, 900a9a4b06a6, master, 2019-09-25) and (2.80 Release, 2019-07-30)\nWorked: (2.79.b, 2019-09-24 last updated in repository)\n\nIf I try to open Blender the screen (to be more exact both of my screens) freeze. The audio of music or videos I had opened in other applications still play but everything is frozen except my mouse-cursor. But not a single window reacts to clicks or any key-presses. I can only end this state of my system by rebooting the whole computer. So I can't practically use the current version of Blender.\n\nIf you have a Navi-GPU from AMD and their amdgpu-pro-drivers installed (which support Ubuntu 18.04) on Ubuntu 18.04 or 19.04, you should be able to reproduce it by executing the \"blender\" called binary from Blender 2.80. The other binary \"blender-softwaregl\" does not cause this error in any version tested but is rather slow in performance (but I guess this is expected).\n\n**My thoughts what causes the problem**\nThe reason I post this problem around here and not only to the side of AMDs support fixing the drivers is that I think I know what's causing this problem. Because Blender 2.79 still seems to work fine on Ubuntu, I thought it could have something to do with Blender using Uniform Buffer Objects or even Shader Storage Buffer Objects from OpenGL in a wrong way. This thought came to me when I tried to develop an own application on my system using OpenGL 4.3 and its SSBOs in a shader but I made the mistake to call the shader before binding a SSBO to the appropriate Shader Storage Buffer Block. The result was the same error I got from calling Blender. So maybe it could just be fixed like in my own code by swapping some few lines in order which would remove this undefined behavior which causes this problem with the current Navi-drivers from AMD. Because I think AMD will probably need more time than you guys. Keep up the great work!" ]

[ "Blender crash after maximizing/unmaximizing an editor (debug build only)\n\nBroken: master - debug build\nWorks: master - release build\nOS: Tested on both Windows and Linux\n\nWith a debug build, Blender crashes after maximizing an editor + moving mouse + unmaximizing.\nIt does not seem to be reproducible with a release build.\n\n\n- Open Blender default scene\n- Maximize the 3D viewport with CTRL+SPACE\n- Move the mouse\n- Unmaximize with CTRL+SPACE\n\n**Debug info**\nThe crash happens in `wm_event_do_handlers` when calling `wm_region_tag_draw_on_gizmo_delay_refresh_for_tweak(wmWindow *win, bScreen *screen)`.\nIn that case, `screen` points to invalid (freed?) memory.", "F12 render all 600 frames complete cause crash.\nOperating system: macOS-13.5-x86_64-i386-64bit 64 Bits\nGraphics card: AMD Radeon Pro 575 OpenGL Engine ATI Technologies Inc. 4.1 ATI-4.14.1\n\nBroken: version: 3.6.3 Release Candidate\n\nF12 render all 600 frames complete cause crash.\n\n\n\n\nBlender 3.6.0 crash\nBlender 4.0.0 is fine\n\n", "Crashes when Switching to Material Viewport / Rendered Viewport / Rendering\nOperating system: Windows 10\nGraphics card: 2 Nvidia GTX 1660 Super's\n\n[blender_system_info.txt](blender_system_info.txt)\n\nBroken: blender 2.8 / 3.4\n\n\nBlender runs completely fine when editing and building a model but once rendering is involved, even in preview, it immediately closes with no explanation. This crash will also happen with the default cube. This had not happened before and have tried Updating blender, reinstalling, simplifying, etc but no change and can reproduce the issue on my computer. i can have it in material mode if i have 0 assets in the scene but once i add something it crashes, i am suspecting a bad pathing issue as i had recently refreshed my computer but do not know how to fix it or give blender a hard reset.\n\n{[F14170509](blender.crash.txt)}[blender_debug_output.txt](https://archive.blender.org/developer/F14170500/blender_debug_output.txt)\nOpen Blender, have any Physical asset inside the scene, switch to rendered / Material Viewport, Crash\n\nI have included the project I was working on and the default cube as they create the same issue.\nPlease note that I have installed the most recent Blender program onto my laptop and opened the files attached and did not have any crash happen so it seems the issue is solely between my computer and the blender program.\n\n[Default Cube.blend](Default_Cube.blend) [Skell Rig base.blend](Skell_Rig_base.blend)", "Render crash\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: Radeon RX 560X ATI Technologies Inc. 4.5.0 Core Profile Context 22.8.1.220810\n+ AMD Ryzen 5 3550H with Radeon Vega Mobile Gfx\n\nBroken: version: 3.3.0 Beta\nWorked: 3.2.2\n\nBlender every time crash when switch to render mode in viewport.\nRender set to Cycles with GPU Compute.\nVega 8 selected as device in preferences.\nIf I select RX 560X in preferences, GPU Compute is greyed in Render preferences (also interesting why).\n\nAny project with simple cube crashed.[test11.crash.txt](test11.crash.txt)\n", "Crash when sculpting with a texture, deleting the image of the texture and exiting sculpt mode\nOperating system: Windows 10\nGraphics card: RTX 2070 Super\n\nBroken: version: 3.4, 3.5, 3.6\nBroken: version: 3.3.3\nBroken: 3.1.0\nWorked: 3.0.1\nWorked: 2.93 LTS\n\nCaused by 9111ea78ac\n\nBlender crashes when a texture image is deleted and you exit sculpt mode.\n\nYou can test it inside the attached blend file. I basically just attached an image to a sculpting brush.\n\n1. Open the blend file below\n2. draw on the surface of the sphere\n3. delete the image inside the outliner\n4. exit sculpt mode\n\nCrash occurs because `item->image->runtime.cache_mutex` is null inside `BKE_image_pool_free`\n```\nStack trace:\nblender.exe :0x00007FF713D04C10 pthread_mutex_lock\nblender.exe :0x00007FF70E6EE880 BKE_image_pool_free\nblender.exe :0x00007FF70E6B5CF0 BKE_sculptsession_free\nblender.exe :0x00007FF70F1B67A0 ED_object_sculptmode_exit_ex\nblender.exe :0x00007FF70F1B85F0 sculpt_mode_toggle_exec\nblender.exe :0x00007FF70E953220 wm_operator_invoke\nblender.exe :0x00007FF70E9528D0 wm_operator_call_internal\nblender.exe :0x00007FF70E955880 WM_operator_name_call_ptr\nblender.exe :0x00007FF70F171800 ED_object_mode_set_ex\nblender.exe :0x00007FF70F148710 object_mode_set_exec\nblender.exe :0x00007FF70E953220 wm_operator_invoke\n```", "Crash/Bugs with material's Surface dropdown + undo\nOperating system: Linux-4.15.0-45-generic-x86_64-with-debian-buster-sid 64 Bits\nGraphics card: GeForce RTX 2080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 435.21\n\nBroken: version: 2.83 (sub 4)\nAlso confirmed to crash in 2.80.\nWorked: 2.79b does not crash (works as expected)\n\nWhen changing the surface/volume in material properties (the list view of the shader nodetree) and undo, opening the dropdown directly afterwards again can crash Blender.\n\n- Download repro blend file (it's just a Cube with a Principled BSDF) [shaderlistRepro.blend](shaderlistRepro.blend)\n- In Properties Editor, click on Surface (Principled BSDF) and select any option in the dropdown\n- Move your mouse over the Surface dropdown and keep it there!\n- Press CTRL+Z to undo\n- Click on the Surface (Principled BSDF) dropdown again (your mouse may not leave the dropdown box)\n- Blender crashes\n\nThe same happens when done with the Volume dropdown instead.\n\n[2020-02-24 15-09-16.mp4](2020-02-24_15-09-16.mp4)\n\n**Possibly related issue**\nIn another file, doing the exact same steps as above has some other effects:\n- Either it spawns this popup (clicking any of these options crashes Blender) \n- Or it spawns a blank popup \nThis is the reproduction file which caused this: [shaderlistRepro2.blend](shaderlistRepro2.blend)\n\nWhich effect happens is really inconsistent and I am not sure how to reproduce one or the other.", "Open Blender\nHello\n\nI can't open the app, i dowload all versión windows, windows store and portable.\n\nsxstrace.exe\n\nI have windows 10, but i can open the app,\n\nI hope you can help me\nThanks\n\n", "Quick Smoke crash guide February 3, 2023\nOperating system: macOS-13.2-x86_64-i386-64bit 64 Bits\nGraphics card: Metal API AMD Radeon Pro 575 1.2\n\nBroken: version: 3.5.0 Alpha\n\nBake ALL button within Sample 128 from Frame 1 to 220 cause CRASH GUIDE\n\nBake ALL button within Sample 128 from Frame 1 to 220 cause CRASH GUIDE\n\n[image.png](image.png)\n[image.png](image.png)\n[Blender-2023-02-03-113746.ips](Blender-2023-02-03-113746.ips)\nTest File: [3.5.zip](3.5.zip)\n```\nCall Stack: Main thread\n__psynch_cvwait\n_pthread_cond_wait\ntake_gil\nPyEval_RestoreThread\nPyGILState_Ensure\npy_timer_execute\nBLI_timer_execute\nwm_event_do_notifiers\nWM_main\nmain\nstart```", "simulation in new scene causes crash in original scene's simulation\nOperating system: Windows-7-6.1.7601-SP1 64 Bits\nGraphics card: GeForce GTX 1080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 442.19\n\nBroken: version: 2.83 (sub 6)\nWorked: (optional)\n\n**steps**\n- make cube an active rigid body. \n- play simulation\n- then click on new scene button and on full copy.\n- in the new scene, rewind playhead to frame 0\n- run the simulation.\nBlender should crash.\n\nThere are more ways to cause crash when full scene copy is done (see comments), but I guess these will stem from single issue.", " Startup File: Header Collapse Menu not restored after restarting blender\nOperating system: windows 10 64bit\nGraphics card: Nvidia gtx m860\n\nBroken:\nversion: 2.80 (sub 37), branch: blender2.8, commit date: 2018-12-16 00:27, hash: 4488056f3d9, type: Release\n\nWorked: (optional)\n2.79b\n\nwhen using Collapse Menu on the header and restarting blender the setting is ignored and the header menu is expanded again.\n\nBased on the default startup or an attached .blend file (as simple as possible).\n\n1. open blender 2.8\n2. collapse the header menu\n \n3. save startup file so we save the header configuration\n\n4. restart blender\n5.the Header menu is expanded again after a restart, other headers like the editor menu stay collapsed after a restart.\n\n", "Edge quality of uv export\nWindows 10\nGtx 1060\n\nBlender 2.8 - February 28 ( d7d180bd3d8d )\n\ndifference looks huge between blender 2.79 and blender 2.8\n\nPlease note: check image in full view.\n\n\n\n\n\nthanks.", "Crash on forcing laptop to sleep when rendering animation\nOperating system: Windows-10-10.0.19043-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1650 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 512.78\nCpu AMD Ryzen 7 4800h\n\nBroken: version: 3.3.1\nWorked: Maybe never? (able to reproduce in 2.83)\n\nblender crash when I wake up my laptop , I lost my unsaved file\n\n- Open .blend file\n- Start render animation\n- Force laptop to sleep {nav Start -> Power -> Sleep}\n[test1.blend](test1.blend)\n[test1.crash.txt](test1.crash.txt)\n[blender_debug_output.txt](blender_debug_output.txt)\n", "Mode locked and crash if objects become disabled in viewport while not in object mode\nOperating system: Linux-5.15.0-2-amd64-x86_64-with-glibc2.33 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 470.86\n\nBroken: version: 3.1.0 Alpha\nWorked: didn't crash but showed an error message in 2.91.2\n\nObjects that are in a mode which is not Edit Mode nor Object Mode, will lock the mode setting (so the mode options are greyed out and unusable) if they get disabled in the viewport. Hiding will not cause the this bug, it needs to be disabled in viewport using that display icon in the outliner, not the eye.\n\nThe solution for this is probably to make the disable in viewport action switch the mode to object mode if the active objects gets removed (much like delete from outliner). Another solution would be to still allow the mode switch.\n\nMuch worse though! Blender crashes sometimes when it gets into this state. I can reliably crash blender as described below.\n\nGo into weightpaint mode of the default cube and disable it in the viewport using the outliner (you need to enable the option first). Now you are stuck and have to enable it in the viewport first to be able to edit anything.\n\nNow restart blender etc. and try this:\n1. Go into Vertex Paint Mode of the default cube\n2. Hit Tab to go into Edit Mode\n3. Disable the cube in the viewport using the outliner as above\n4. Hit Tab to switch out of Edit Mode -> Crash", " EXCEPTION_ACCESS_VIOLATION when entering Edit Mode\nOperating system: Windows 10 1903 (18362.239)\nGraphics card: Intel HD Graphics 4000\n\n***Broken:***\nBlender 2.80 (sub 75) / 2019-07-29 09:44 AM Windows / Stable\nBlender 2.81 (sub 1), branch: master, commit date: 2019-08-13 11:31, hash: 6f9cbbc8ec4f\n***Worked:***\nBlender 2.79 (copied mesh over through copybuffer.blend)\n\n\nWhen entering edit mode of a specific object Blender crashes with EXCEPTION_ACCESS_VIOLATION. Logs and example .blend provided.\n\nThe issue shouldn't be in my GPU, since the same issue persists even with Software Rendering.\n\n1. Open up example file\n[window_broken.blend](window_broken.blend)\n\n2. Select `window_panes` and enter Edit Mode\n3. Get EXCEPTION_ACCESS_VIOLATION\n\n**Logs (2.80 `blender_factory_startup`)**\n[blender_system_info.txt](blender_system_info.txt)\n[blender_debug_output.txt](blender_debug_output.txt)\n\n**Logs (2.81 `blender_factory_startup`)**\n[blender_system_info.txt](blender_system_info.txt)\n\n[blender_debug_output.txt](blender_debug_output.txt)", "Momentary interface freezes while EEVEE rendering causes progress bar malfunction\nOperating system: Windows 10, 64 bit\nGraphics card: Intel HD Graphics\n\nBroken: 2.80, 2.81\n\n**Rendering in EEVEE has been causing momentary complete interface freezes for each frame**, which caused the progress bar on the frame render to malfunction and jump very quickly from one value to another, thus giving momentarily true but useless information (note that the progress bar never moved to an incorrect position, it just froze). One thing that assures me this is a bug is **minimizing all Blender windows causes the progress bar visible on the minimized icon to work properly** (but unfortunately one cannot see what frame it's on).\n\nOpen the default blend file.\nRender the animation. The interface on 2.81 has frozen for each frame." ]

[ "2.8 rc crashes when clicking on Menu \"Shading\"\nOperating system: Windows-7-6.1.7601-SP1 64 Bits\nGraphics card: Intel(R) HD Graphics 4600 Intel 4.2.0 - Build 10.18.10.3574\n\n\nBroken: version: 2.80 release candidate\nWorked: (optional)\n\n\nblender crashes on me (Lenovo Y70/70 with GeForce GTX 860M, latest drivers 26.21.14.3136)\n\nStart blender rc 2.8 \n\n\nNothing loaded\njust click Menu Item \"Shading\" - pooof, crash" ]

[ "Eevee incorrect motion blur Mantaflow liquid\n\nOperating system: Windows 7 Pro 64-bit\nGraphics card: Nvidia GeForce GTX 1060 6-gig\n\nBroken: 2.90.0 Alpha - 2020-07-03 - hash cad98923d006\nWorked: ---\n\n\nEevee motion blur not correctly applied to mesh of Mantaflow liquid sim.\nCube in scene is correctly blurred.\nMantaflow appears to show only start/end positions with some sort of strange (zoomed?) artifacts in between.\n\n\n\n**Steps to reproduce**\n[Eevee motion blur bug.blend](Eevee_motion_blur_bug.blend)\n\n- Open file\n- Bake simulation\n- Render", "Rigid-Bodies Colliding Improperly\n[Basic Setup with bug 2.blend](Basic_Setup_with_bug_2.blend)\nOperating system: Windows-10-10.0.17134 64 Bits\nGraphics card: GeForce GTX 950M/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 436.48\n\nBroken: version: 2.81 (sub 13)\n\nSome of the Rigid-Body objects in the included .blend go flying when animation is played, as though they were intersecting each other, but they are not.\n\nOpen the included .blend.\nPlay animation. The two objects named \"Suspension.FL will go flying off in opposite directions.", "Hair particles won't follow instaced position under 2 level nested collections.\nOperating system: Linux-5.11.0-7620-generic-x86_64-with-glibc2.31 64 Bits\nGraphics card: NVIDIA GeForce RTX 3080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 470.57.02\n\nBroken: version: 2.93.3\n\nHappens when you have an object with hair as instanced collection (A) inside another collection (B). Then, when you instance (B) the hair particle from (A) will not follow the relative position correctly.\n\n\n- Open .blend file\n- Select \"Cube\"\n- Object -> Quick Effects -> Quick Fur\n- {key Shift A} -> Collection Instance -> Collection\n- Move the instanced collection little away\n- {key M} -> New Collection -> Ok\n- {key Shift A} -> Collection Instance -> Collection 2\n- Move the instanced Collection 2 little away\n- You will notice the hair particle with a wrong position in relation to the original emitter Cube.\n\n1 min video here:\n[bugOrFeature.mp4](bugOrFeature.mp4)\n\nblend 2.93 here:\n[#91008.blend](T91008.blend)\n\n---\n\nNote that the concept works under 2.7 using Instaced Groups:\n\nblend 2.7 here:\n[ConceptWorks_b27.blend](ConceptWorks_b27.blend)\n", "Particle instance modifier can crash when create from Children is checked and Normal is not\nOSX 10.11.6, xeon w3680 6core 12thread \n[system-info.txt](system-info.txt)\n2.69.7 5bd8ac9\nalso Nightly 2.79 34fe3f9c069\n\n\nUsing a particle instance modifier with Children checked and Normal not checked and then attempting to change certain parameters of the particle system such as in the velocity or children tabs causes a crash. Enabling Normal in the particle instance modifier prevents this from happening.\nCreate a Plane with a hair type particle system.\nCreate a cube with a particle instance modifier and set the object to the Plane with the particle system.\nCheck the box for Children and uncheck the box for Normal.\nGo to the particle system on the Plane and check Advanced. \nDrag any of the values in Velocity or Children and soon it will crash.\n[Crash_ParticleInstance.zip](Crash_ParticleInstance.zip)", "Use after free: Baking Rigid Body physics causes instant crash\nOperating system: macOS 10.15.7\nGraphics card: AMD Radeon Pro 5500M 4 GB / Intel UHD Graphics 630 1536 MB\n\nBroken: 2.92.0, branch: master, commit date: 2021-02-24 16:25, 2.93.0 Beta, branch: master, commit date: 2021-04-28 22:25, \nbroken: f7b22fc3d27113715e8726c69ab3264e1d487439\nWorked: \n\nBaking the Rigid Body world physics causes Blender to crash either instantly, or after baking less than 100 frames (it varies). Interestingly, a friend with the PC version of 2.92 says the project works fine, no crashes whatsoever.\n\nOpen the project. Make sure Allow Negative Frames is on under Preferences/Animation. Go to the Rigid Body World and hit bake.\n\n[physics test b.blend](physics_test_b.blend)\nAsan report: [T88113_asan_report.txt](T88113_asan_report.txt)", "Particles from multiple systems disappear when all Physics Type are None.\nOperating system: Windows 10\nGraphics card: RTX 2070\n\nBroken: 2.81a\n\nWhen selecting a Particle System, all particles of the object disappear.\nImages showing before and after:\n\n\n\n- In the default cube add two particle systems.\n- Set the `Frame End` of both to 1 for better observation.\n- In the `Physics` tab of both choose `Physics Type` `None`.\n- Select any of the particle systems, note that the particles disappear.\n\nAttached file contains an example of multiple particle system on one object:\n[LP_grass.blend](LP_grass.blend)\n\nThis has to be the first file you open with the application.\nIf you open blender and enable the eevee material preview for example with the default cube and open the example file afterwards it works.", "Generic Rigid Body Constraint Bug\nOperating system: Windows 10\nGraphics card: Radeon RX 470\n\nBroken: 2.83.0\nWorked: Unknown if it ever worked\n\nAngle limit not only doesn't work but causes weird neverending twitching back and forth\nUPDATE:\nTwitching happens for both cases when both rigid bodies are active and when one of them is passive.\n[M0DJ38oWQo.mp4](M0DJ38oWQo.mp4)\n\n1. Add 2 rigid bodies, 1 active, 1 passive animated\n2. Position them active on top, passive on the bottom\n3. Add an empty in between them, add a rigid body constraint to it, select the 2 rigid bodies in its settings, uncheck disable collisions.\n4. Change the joint type to Generic, limit location to 0-0 on each axis.\n5. Play. Now it works, once you select the passive RB and move it it should behave normal.\n6. Undo. Check angle limit on Z and choose some angle range (I had it set to -30 to 30).\n7. Play. Move the passive RB and watch it breaking.\n", "Mantaflow - Enabling \"Noise\" breaks fast-moving smoke\nOperating system: Windows 10\nGraphics card: Nvidia GeForce GTX 650 Ti\nProcessor: Intel® Core™ i7-3770\n\nBroken: 2.83.0\nWorked: Never\n\nEnabling noise on fast-moving smoke produces a \"grainy\" result that doesn't match original simulation.\n[C - High Speed Noise - Demo.mp4](C_-_High_Speed_Noise_-_Demo.mp4)\n\nAlso, this problem is less profound when the cache type is set to \"Final\", though still clearly visible.\n\n**Exact steps to reproduce the error**\n1. In any smoke simulation, dramatically increase the velocity of the smoke, either in the emitter settings, or with a force-field.\n2. Bake Data\n3. Bake Noise\n\n[Noise on High-Speed Smoke.blend](Noise_on_High-Speed_Smoke.blend)", "Animated particle emitter refuses transforms within simulation range\nOperating system: Win8.1x64\nGraphics card: 3xgtx580\n\nBroken: 2.79, 2.82\n\nAn animated object with a particle system it is not properly transformed in the frames within the simulation range.\nIn **Blender 2.79** the object jumps back to it's animated location on confirmation on hotkey or widget transformation.\nIn **Blender 2.82** the object doesn't even transform, but its particles does.\n\nTemporarily setting particle system Subframes parameter to 0 seems to be a workaround, eliminating the behavior.\n\nHere is a gif showing the bug in 2.82 and 2.79:\n\n\n- Open attached file\n- (G)grab move the emitter\n[particle-subframes.blend](particle-subframes.blend)\n\nIn blender 2.82 you must change the visibility of objects. If you want to skip this step, use this file:\n[particle-subframes-283b.blend](particle-subframes-283b.blend)", " Mantaflow: Adaptive Domain breaks Dissolve if 'slow' is checked\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 441.28\n\nBroken: version: 2.82 (sub 6)\nWorked: (optional)\n\n[Adaptive domain seems to completely ignore the 'dissolve' setting on the Domain object, or at least alter it massively.]\n\n[Bake the attached .blend\nHere's the results-\n\nThe left hand Domain has dissolve enabled but not Adaptive. The right hand one is a duplicate of the left (and the left hand emitter), nothing at all changed apart from enabling Adaptive domain.\nDoesn't look to me like the 'dissolve' parameter is doing terribly much at all on the right hand side, when 'Adaptive Domain' is enabled.]\n[Adaptive_Dissolve.blend](Adaptive_Dissolve.blend)\n\nI discovered this issue half way through a 400 frame render, when looking at the rendered stills; so I thought stuff-it! Carry on...\nSo here is what this issue does to an actual sim I wasted several hours on creating/tweaking, then the best part of a half a day rendering-\n[Domain_Jitter_Dissolve.mp4](Domain_Jitter_Dissolve.mp4)\n\nIt looks worse than the .blend I uploaded as it has noise baked as well, and this seems to generally expand the volume filled by the smoke from what I've experienced, therefore exacerbating the jitter where the smoke hits the Adaptive Domain boundary. \nAs you can see it's quite ugly when the boundary of the smoke gets close to the camera, and as I hope you can see the jitter at the edges of the smoke look suspiciously like the jitter in the size of the domain seen in the viewport capture I made (obviously I matched the speed in Davinci, my workstation could only sustain 2FPS playback).\nIf anyone cares for my two-pennys-worth, whatever coded-wizardry controls the size of the Adaptive Domain, need's to implement some sort of 'damping' for the domain size changes, at least several (10-15?) frames worth. As well as taking 'dissolve' into account.\n\n", "Tablet Pressure not working in New Window\nOperating system: Windows 10 20H1\nGraphics card: Nvidia 2070 8G\nDrawing Tablet: Huion Kamvas 16 Pro 2019\n[system-info.txt](system-info.txt)\n\nBroken: 2.92.0 and 2.93.6\n\nPen pressure not working on newly created window. Very repeatable. \n\n\n - Open Blender\n - Open Project\n - Create new Window from Main View\n - Move New window to Kamvas 16 Tablet\n - Use tablet pen to paint the mask. \n - **Pressure does not work.**\n - Verify by toggling Pen pressure button in radius and strength fields.\n - Use tablet pen again to verify\n - **Pressure still not working.**\n - Close The window on the Kamvas Tablet. \n - Move Main Window to Kamvas Tablet. \n - Use tablet pen to paint the mask. \n # Pressure is working.\n\n\nExpected behavior would be that the pressure works in the newly created window. The steps outlined above as well as the attached video evidence prove that the Tablet Pen is working In blender just not on the newly created window. \n[2021-11-23 03-52-41.mkv](2021-11-23_03-52-41.mkv)\n\n", "Particle system: late update of object`s tracking axis after «object`s rotation» switching.\nOperating system: Windows-7-6.1.7601-SP1 64 Bits\nGraphics card: GeForce GTX 660 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 442.59\n\nBroken: version: 2.83 (sub 13)\nWorked: Apparently never (2.79 has the same problem)\n\nAfter switching on/off object`s rotation for particles, their rotation updates wrong and I need to switch something at initial object to update rotation of particles.\n[2020-04-22_16-08-56.mp4](2020-04-22_16-08-56.mp4)\n\n- Make particle system, that emits object.\n- Switch object`s tracking axis to +Z (relations)\n- Switch Object`s rotation in the particle system render tab.\n- Rotation is wrong and not changing until object`s transform not updated.\nor\n- Open attached file\n- Move the \"Cone\" object to update the particles\n- In the Cube particle settings, enable and disable `Object Rotation`\n[particle system rotation tracking no update.blend](particle_system_rotation_tracking_no_update.blend)", "Rigid body: Animated body collision geometry have unapplied scale for the first frame.\nOperating system: macOS-12.6.5-x86_64-i386-64bit 64 Bits\nGraphics card: Intel(R) Iris(TM) Plus Graphics 645 Intel Inc. 4.1 INTEL-18.8.6\n\nBroken: version: 3.5.1\n\n\nIf an rigid body object has a scale that's not 1 , when physics simulation begins, the collision on the first frame will be calculated as if the object (collision geometry) is scaled back to 1, but subsequent frames will have correct scale.\n\n-----\n\nI have been following this tutorial\n\nwatch?v=2Dnks7C2q9U\n\nand as soon as I set the 'handle' object 'Animated' and press 'Play' on timeline, the bullet physics immediately kicks-in,\n\n\n\nThis is the file I'm using:\n\n[20230618_rigid_body_tipping_over_stacks_of_cubes_animated_wrong_activation_0001.blend](attachment)\n\n", "Rigid body bugs\nOperating system: macOS-10.15.7-x86_64-i386-64bit 64 Bits\nGraphics card: AMD Radeon Pro 575 OpenGL Engine ATI Technologies Inc. 4.1 ATI-3.10.21\n\nBroken: version: 2.93.1\n\nSimulation somehow broken for small (less than 1 cm) or light (less then 100 grams) objects. In first example below object stay still between bouncing and small shivering. And that is just one example. Another examples are – reaching speed of light instantly (one frame), partial simulation when one object (or some objects) acts like it suppose to, but others act completely inadequate, two or three similar parts in one simulation (example 2). I tried different options but nothing helped. \"Substeps per frame\" and \"Solver iterations\" seems not working because there are no actual changes with different values or simulation just goes nuts.\n\n[r_body-bug.blend](r_body-bug.blend)\n[r_body-bug2.blend](r_body-bug2.blend)\n", "[Mantaflow] Generating a huge number of particles as substeps increase\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: AMD Radeon HD 7560D ATI Technologies Inc. 4.5.13399 Core Profile Context 15.201.1151.1008\n\nBroken: version: 2.90.1\n\nThe number of particles increases as the number of substeps increases.\n\nBake the liquid from this scene.\n[substeps_bug.blend](substeps_bug.blend)\nYou will see the following:\n\n" ]

[ "Mantaflow Gas Emitters Unaffected by Particle Velocity\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 466.77\n\nBroken: version: 2.93.1\nWorked: Before Mantaflow\n\nSmoke inflow ignores particle velocity, fast moving particles generate motionless smoke. Prior to mantaflow, the old gas smoke/fire solver DID take particle velocity into account.\nThis feature was very useful for making complex looking explosions, and made it much simpler to guide gas simulations.\n\n\nOpen this file, press space.\n\n[SmokeParticleVelocityBug.blend](SmokeParticleVelocityBug.blend)", "Mantaflow: emit smoke from particles can't inherit velocity.\nOperating system: windows 10\nGraphics card:\n\nBroken: 2.92\nWorked: 2.83\n\nEmit smoke from particles can't inherit velocity.\n[volume speed bug.blend](volume_speed_bug.blend)\n\n\n2.90.0\n\n\n2.91\n" ]

[ "Outliner Floating window Drag & Drop to Geometry Nodes editor not working\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 536.23\n\nBroken: version: 3.6.0\nWorked: I don't know if it ever worked\n\nDragging and dropping objects from the outliner to the geometry node editor is not working\n\nFloat outliner to be its own window\ndrag and drop an object from the node editor to the geometry nodes\ndoesn't work.\n\n", "Create geometry nodes modifier when selecting an existing one in the editor\nSelecting an already existing geometry nodes tree in the editor on an object without a modifier should automatically create a nodes modifier. This is in parallel to the behaviour of selecting a material without a material slot.\n", "Make Instances Real doesn't work for geometry instances\nOperating system: Windows-10-10.0.18363-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 750/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 471.41\n\nBroken: version: 3.0.0\n\nWhen I want to make instances real on object that has geometry nodes, blender create duplicate of whole mesh, togather with instances, on each place (point) where single instance mesh should be. I have attached screenshots and test scene that reproduces the bug. \n\nOpen .blend file I sent you, hover with mouse over 4 instaced spheres, and pres ctrl+A... select make instances real. You will see that geometry nodes instances are not created properly (whole geometry is duplicated on each geometry node point).\n\nSimilar thing happens when instancing collections, and trying to make them real - the parent relationship/hiararchy is not working. So there is a bug there also.[geom_nodes_bug.zip](geom_nodes_bug.zip)\n\n\n\n\n", "Geometry Nodes: Mesh Boolean-Union node anamoly when boolean does not alter input mesh\nBroken: Current main.\nWorked: Never.\n\nI was trying to see if the simulation node group would get rid of internal geometry by incrementally using the **Mesh Boolean-Union** node in the **Simulation node group** but it **seems to delete some outer geometry instead.** \n\n## Without Simulation Nodes:\n| Operating as expected without Mesh Boolean Union node enabled (**creates internal geometry see red arrow**) | Operating as expected with Mesh Boolean Union node enabled (**no internal geometry**) |\n| -- | -- |\n|  |  |\n\n## With Simulation Nodes:\n\n| Description | Show |\n| -- | -- |\n| Operating as expected without Mesh Boolean Union node enabled (creates internal geometry see red arrow) |  |\n| Not Operating as expected with Mesh Boolean Union node enabled (no internal geometry but missing outer geometry) |  |\n| It should look something like this (no internal geometry but it should have \"external\" geometry also see red arrow |  |\n\nThe reason I'm trying to use the boolean mesh-union node within the simulation node group is because on some large meshes the boolean mesh-union node will fail or crash blender. \n105652#issuecomment-931240\n\nMy logic is if you have to finish an entire pie in one sitting (think of a pie eating contest) it's easier to cut them up into little pieces and eat them one at a time. **So I'm incrementally running the boolean mesh-union node on smaller pieces in the simulation node group to get rid of internal geometry therefore it's less likely to crash / fail like it does trying to do this all at once on larger meshes.**\n\nSee attached blend file: [cube repeat.blend](https://projects.blender.orgattachment)", "Geometry Nodes: Simulation and Repeat nodes is work though they are muted\nBroken: version: 4.0.0 Alpha\nWorked: Never.\n\n\n\n1. Geometry Nodes tree:\n a. Add Simulations nodes.\n b. Mute both.\n2. Play animation.\n\n", "Unexpected behavior when applying Geometry Nodes modifier on top of the Multires.\nThis is a more comprehensive report on a bug that was recently posted and closed [HERE ](T97603)\n\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 472.84\n\nBroken: version: 3.0.1+\n\nApplying Geometry Nodes on top of the Multires modifier does not give expected results. Multires is an exception from the \"Applying a modifier will apply it as if it was first in the stack\" rule, when a modifier is applied on top of the Multires, the mesh changes will be applied to the currently visible subdivision level. Quick video below.\n[Multires GNodes Bug.mp4](Multires_GNodes_Bug.mp4)\nAnd the Example .blend file > [Multires GN Bug.blend](Multires_GN_Bug.blend)\n\nHow should applying the Geometry Nodes behave ? When the Geometry Nodes modifier doesn't change the vertex count and simply deforms the mesh, it should behave just like any other modifier, see example video with the Shrinkwrap modifier. \nSo what when Geometry Nodes does change the vertex count ? I say we keep the same behavior that solidify modifier has, it skips Multires and applies it as it would be first in the stack. While of course this will bring broken results, this is a already an existing behavior so i think it would be completely fine.\n\nSeries of images presenting how applying a Solidify modifier on top of Multires behaves. (Intended behavior)\n\n\n\n\n\n\n\n", "Geometry Node: Filter out based on subtype (modifier/tool)\nWe need to filter out the nodegroups based on subtype (tool/modifier):\n\n* Template in the Geometry Nodes editor header\n* Template in the Geometry Nodes modifier\n\nNote that the Node Group node template can still take both type of node groups.\n\n", "Geometry Nodes Mesh Boolean unexpected difference result\nOperating system: macOS-13.1-arm64-arm-64bit 64 Bits\nGraphics card: Apple M1 Apple 4.1 Metal - 83\n\nBroken: version: 3.5.0 Alpha\nWorked: none\n\nGeometry Nodes Mesh Boolean Difference does not work as expected.\nNode setup:\n\n\nUnexpected boolean difference:\n\n\n- Open the attached .blend file\n- Press G to move the selected Cube.001 and the small cubes will flicker.\n# At certain positions of Cube.001 some of the small cubes will disappear when they should not.\nThanks!\n[test06.blend](test06.blend)\n", "Assert duplicating a collection in a linked scene\nOperating system: Linux-5.13.0-0.rc6.45.fc35.x86_64-x86_64-with-glibc2.34.9000 64 Bits\nGraphics card: NVIDIA GeForce GTX 970M/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 465.31\n\nBroken: version: 3.1.0 Alpha\n\nAssert duplicating a collection in a linked scene\n\nfrom the default startup:\n- 1) `Scene` > `Linked Copy`\n- 2) select the collection `Collection` in the Outliner\n- 3) context menu > `Duplicate Collection` or `Duplicate Linked` (does not matter which)\n\n```\nBLI_assert failed: source/blender/blenkernel/intern/collection.c:184, collection_owner_get(), at '(id->tag & LIB_TAG_NO_MAIN) == 0'\n```\n\nIf the file is saved (and reloaded) before step 3), the above assert does not happen", "Unify Image Texture node in geometry and shader nodes\nGeometry nodes added a new Image Texture node where the image datablock is a socket.\n\nWe should unify these two nodes. There are some design decision to make regarding how to deal with image users and the frame offset socket, see discussion in {[D12827](D12827).}. ", "Rip edge-move for curves\nCurrently with meshes you can press Alt+D to add a vertex at one side of the verts edges and drag it *(based on the mouse cursor location)*.\n\nThis same behavior could be added to curves.\n\n- the basic behavior with single vertices selected is simple\n- with multiple segments are selected this becomes more involved (see how this works for meshes).", "Inconsistent Limitations of Instances in Geometry Nodes\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 512.15\n\nBroken: version: 3.2.0 Alpha\n\nIt's not clear what operations can or can't be done to an instance, eg. Extrude and Scale Element, Dual Mesh, Triangulate etc all work fine but Set Position doesn't. \n\nSorry for the complex file but I wanted to leave in all the nodes I'd been through before realising that I was working an Instance still.\nThe Set Position node was the first one in that line that gave me any indication that I was still on an instance.\n\n[12 Marching Squares.blend](12_Marching_Squares.blend)\n", "Geometry Node Crash\nOperating system: macOS-11.6-x86_64-i386-64bit 64 Bits\nGraphics card: Metal API AMD Radeon HD - FirePro D500 1.2\n\nBroken: version: 3.6.2\n\nGeometry Node crash when I try to add a node to the default Cube.\n| Node Setup: | Crash Log: |\n| --- | --- |\n|||\n\n- open blender\n- open the geometry node workspace\n- With the default cube selected, add a new Geo Node setup\n- add a node using the menu or Shift A\n\nBlender crashes.\n\nIf I delete the cube, add a plane, and add the geo node to the plane, blender does not crash.\n\n", "Outliner: Unexpected behavior when duplicating object hierarchies\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA RTX A4500/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 512.15\n\nBroken: version: 3.4.1\n\n**may not be a bug** - duplicate a parent object (i.e. an object that has child objects). The duplication automatically selects the new object in the 3D viewport but this selection differs from that shown in the outliner. This means dragging to, say, a new collection does not work as expected (only the parent object is moved leaving the child behind).\n\nThe set up (this is the attached blend file):-\n1. Create two cubes (or any meshes) and parent one to the other to form a hierarchy.\n2. Add the object(s) to a collection.\n3. Create another top level collection which is empty.\n\nThe aim:-\nTo duplicate the object(s) (not the collection) and copy the duplicate to the second (currently empty) collection.\n\nThe steps:-\n1. Right click the parent object in the outliner and choose \"Select Hierarchy\". Note that the outliner shows the selected objects with a blue background bar (as expected).\n\n2. In the 3D viewport, use Shift or Alt D (doesn't matter which) to create new duplicate objects. Note that the 3D viewport shows the duplicate parent and child selected. The outliner shows a new object selected (with blue bar) but the tree is collapsed so we can't see the child.\n\n3. In the outliner, click and drag the selected parent object to the empty collection. Note that only the parent moves, the child is left behind. This child has then to be dragged separately.\n\nAfter step 2, every visual cue leads the user to believe that the duplicate hierarchy is selected. And this is the intuitive behaviour expected. But expanding the outliner tree shows the child object is not, in fact, selected (no blue bar). So an additional step of having to reselect the hierarchy is required before step 3.\n\nThe main issue is that with lots of objects and detailed hierarchies, just forgetting that additional \"select hierarchy\" step before dragging means that child objects can be scattered all over the place away from their parent and the user is left picking through the large list trying to sort it out.\n\nThis may not be a bug, it may be working as designed. But I cannot think of a use case where the current functionality would be useful so I'm assuming it's a small selection bug.\n\n", "Curve Deform Node\nNOTE: The design below is outdated. A curve deform node should be a field node now, with a vector input and output, and a curve input. Arguably it should just be a node group too.\n\n\n| {[F10152887](image.png) size=full}| {[F10152878](image.png) size=full} | {[F10152882](image.png) size=full} |\n| -- | -- | -- |\n\n\nThis is task is for a port of the \"Curve\" modifier to a node.\n{[F10150643](image.png), size=50%}\n\n\n**Details**\n* Any geometry component with a position attribute is supported by this node.\n* Only the first spline of the curve is used for deformation.\n* Two input modes modes are available. \n**The position mode uses point positions for the curve parameters.*** The position input is provided for convenience. In most common cases, a single axis of the position is used as a parameter for the curve.\n***The position is mapped to the curve based on length. So \"x=5m\" means, transform by traveling 5m along the spline. This is how the modifier works.** The attribute input mode uses any float attribute between zero and one, as a factor along the length of the curve.\n** For convenience, a \"Float\" option is available for the attribute when the whole geometry should have the same parameter.\n* Extrapolation should be handled by a separate node that can extrapolate a spline.\n \n\n**Further Questions**\n* Is an \"influence\" attribute input really necessary? The same thing could be achieved by mixing a copied position attribute before starting the operation.\n** The answer to this question will apply to other operations as well.\n* Which of the three \"Curve Deform\" settings are necessary in this node? \"Use Radius\" is likely not necessary, the attribute could just be filled beforehand.\n\n**Related Nodes**\n* A \"Curve Sample\" node could do a similar operation, but instead of deforming positions, it would sample an attribute along the length of a curve based on the same mapping controls.\n** This could be useful to have more flexible control of attributes with curves.\n\n\n**Implementation**\nThe implementation should be straightforward. Sampling arbitrary points along the curve is an natural extension of the uniform sampling that is already implemented.\nCurve parameters should be sorted before, and then changed into evaluated point indices and factors.\nThen, interpolation between the transform of neighboring evaluated points should be quite simple.\n" ]

[ "Geometry nodes: keyframes being deleted if you delete the same node in a node tree duplicate\nOperating system: Linux-5.4.0-91-generic-x86_64-with-glibc2.31 64 Bits\nGraphics card: GeForce GTX 1080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 460.91.03\n\nBroken: version: 3.0.0, 3.1\n\nMake a keyframe on a texture node parameter, duplicate the node tree and delete this node in the duplicate. The keyframes in the original node tree node will be deleted\n[2022-01-01_21-22-12.mp4](2022-01-01_21-22-12.mp4)\n\nIn the file attached.\nCopy current node tree.\nDelete the node.\nGo back in the original node tree.\n[gn_key_deleted.blend](gn_key_deleted.blend)" ]

[ "Cycles doesn't render volumes when clip end is inside the volume\nOperating system: Windows-10-10.0.19042-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 471.11\n\nBroken: version: 3.1.0 Alpha\n\nWhen clip end is set to a value inside a volume, the volume is not renderered at all.\nRelated to [#92491 ](T92491) so an alpha build is needed.\n\n- Open project\n- Press F12 (Volume isn't rendered)\n- Set clip end to something high (e.g. 100)\n- Press F12 again (Volume is rendered)\n\n[volume.blend](volume.blend)\n\n", "Greasepencil z fighting alpha issue\nHi,\nnot too sure if this is known issue or any fix but I am getting this alpha issue from certain angle when I have 2 different grease object. \n\n", "Dithering is not applied for Texture Baking\nDebian amd64, Radeon HD with fglrx\n\n\nBroken: current GIT, also many previous versions were affected in case of BI Baking\n\n\nDithering is not applied for Texture Baking, resulting in banding artefacts.\nCycles Baking and BI Baking are affected.\n\nAlso when saving a 32 bit float image as 8 bit PNG, no dithering is applied (despite Dithering was set to 1.0, and \"Save as Render\" was activated).\n\nA .blend file is attached. Click on \"Bake\" and wait a few seconds. After baking is finished you should see the banding artefacts in the dark-gray regions.\n\nThanks\n", "No \"affect alpha\" in brush settings in the Image Editor\nOperating system: Linux-5.3.0-40-generic-x86_64-with-debian-buster-sid 64 Bits\nGraphics card: GeForce GTX 980 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 435.21\n\nBroken: version: 2.82 (sub 7)\n(as well as the latest alpha: version: 2.83 (sub 5))\nWorked: 2.80 (Checkbox existed in Image Editor brush settings, but had NO effect on brush behavior)\n\nIn the 3D Viewport as well as the Properties Editor while in Texture Paint mode, all brushes offer the \"affect alpha\" option in the \"Advanced\" section of the Brush Settings. In the Image Editor that option is missing from the \"Advanced\" section for no apparent reason.\n\n\nOpen Blender with factory settings. Switch to \"Texture Paint\" tab. Compare the \"Advanced\" section of the active brush's tool settings in the Image Editor to that in the 3D Viewport or Properties Editor.\n\nSurely this has to be a well-known issue and exists only due to technical limitations of some kind.\nWhat should I tell my students why this inconvenience exists and when it will be addressed?", "VSE: Some \"Select Grouped\" operators don't work as expected\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.94\n\nBroken: version: 4.0.0 Alpha\nWorked: (unknown)\n\n\nSome of the operators in the VSE's \"Select -> Grouped\" menu have wrong/confusing names or don't behave as expected.\n\n---\n\n`Select -> Grouped -> Effect Type`\n\nDespite the name and the tooltip, this operator does not check the effect type. It only checks whether the strip is an effect or not. Maybe it could be renamed to \"Is Effect\"? Similarly, the vague \"Global Type\" could be renamed to \"Is Sound\".\n\n---\n\n`Select -> Grouped -> Data`\n\nWhen run on a file-based strip, it selects all strips pointing to the same folder rather than to the same file. This may be correct behavior for image sequences, but not for everything else (movies, sounds, single images).\n\nIn the screenshot below, the expectation is that it only adds the second strip to the selection, not the third.\n\n\n---\n\n`Select -> Grouped -> Effect`\n\nGoing by the tooltip (\"Shared effects\"), you'd expect that this operator selects content strips that share an effect strip with the active strip. Instead, it finds the effect *types* linked to the active strip, then selects any other content strips linked to that same effect type.\n\n(So really, this should have been the \"Effect Type\" operator above)\n\nIn the screenshot below, I'd expect that it only adds \"2.png\" to the selection. Instead, it adds all the strips that are affected by an \"Add\" effect.\n\n\n---\n\n`Select -> Grouped -> Effect/Linked`\n\nApart from the tooltip not being very clear, this operator doesn't work at all (simply deselects everything) unless \"Extend\" is enabled in the \"Adjust Last Operation\" panel.", "Make channel selection consistent with outliiner\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3050 Laptop GPU/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 532.03\n\nBroken: version: 4.0.0 Alpha\n\nChannel selection are either cleared or not detected when selecting with Shift-ctrl-LMB\n\n- open attached file\n- In outliner `ctrl-click` on two objects to select them\n- then `shift-ctrl-click` on third object (selects third object and doesn't clear previous selection)\n- Now perform same with animation-channels\n\n", "VSE Channel headers TODO\nImprovements for channel headers in general:\n- - [ ] 2-way `v2d` scroll synchronization\n- - [ ] Investigate why changing RNA property doesn't update timeline region\n- - [ ] Cleanup loading code for `SeqTimelineChannel`, looks like it could be optimized a bit more.\n- - [ ] Use better integrated drawing engine like list-views\n\nOther related tasks:\n - Decide on [D14263: VSE: Limit timeline view height](D14263)", "Current value 0 matches no enum in error with creating a new keymap from Industry Compatible or Blender 27x keymap\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 1060 6GB/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 461.09\n\nBroken: version: 3.1.0 Alpha\n\nWhen you create a new keymap from the industry compatible keymap or the Blender 2.7x keymap, then you get errors in the console about this and that operator does not match current value 0 in enum ...\n\nRelated task: T92754\n\n\n\n- Open Blender\n- Open Preferences\n- Switch to Industry compatible Keymap\n- Create a new keymap\n- Watch the console\n- Repeat with the Blender 27x keymap as the base ...\n\n\n\n\n\n\n\nEDIT, the params.action_mouse parameter could be the cause.\n\n\n\n\n\n", "The alpha channel set in Viewport Display settings for an object's material appears as if it's set to 0 when a background image is behind it\nOperating system: Linux x64 (Debian-based)\nGraphics card: NVIDIA Corporation GP107 [GeForce GTX 1050 Ti] (rev a)\nNvidia drivers, version: 440.31\nKernel: 4.19.0-6-amd64\n\nBroken: v2.81a and v2.80, downloaded from blender.org as a tar.gz\n\nWhen you change the Viewport Display color for an object to have an alpha channel < 1.0 but > 0.0, then manipulate the scene so that there's a reference or background image behind that geometry, the overlapping area will appear completely transparent (alpha 0.0).\n\n\n1. New scene\n2. Add a cube: Shift a -> Mesh -> Cube\n3. Add an image: Shift a -> Image -> Reference (or background)\n4. Select an image\n5. Position the image behind the cube in the viewport. Select the cube.\n6. In the properties panel, select Material Properties\n7. In the Material Properties panel, select Viewport Display\n8. Change the Color to have an alpha between 0.0 and 1.0, exclusive\n\n**Observed behavior**\nLike the screenshots, the overlapping area allows the background image to completely shine through.\n\nQQxLCpv\n\n**Expected behavior**\nThe object will become uniformly translucent and allow the background/reference image to be partially visible, not make the image fully transparent in the overlap.\n\n", "Data changes in pre and post scene evaluation handlers not recognised by Cycles, plus occasional ACCESS VIOLATION and inconsistent scene properties.\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 512.96\n\nBroken: version: 3.4.0 Alpha\nWorked: (Unsure)\n\nCycles is not using the evaluated scene following scene modifications via handlers that fire before scene evaluation, and some data refers to the previous frame whilst other data refers to the frame currently being rendered..\n\n\nHi, we're trying to get Blender fully integrated into our studio, but we're running into quite a lot of issues with handlers. It seems that Cycles is only partially recognising changes made to the scene in the pre_render, pre_frame, and post_frame handlers. \n\nIn the below zip file there are 3 directories, each containing a blend file with a simple script to demonstrate the issue. The script changes the file output node's base path, and it also changes blender own render.filepath in the output properties panel. The new file name includes the scene camera name (which changes on frame 1 and 3 using markers) and the frame number when the handler updated the filepath/base_path properties, followed by the __0001 added by Blender. For example frame one should read: left cam_updated_path_on_frame_1__0001.png\n\n**Steps to reproduce problem**\n\n\n\n - Unzip [cycles handler bug.zip](cycles_handler_bug.zip)\n - open each of the blend files in place (use 'load UI' as I've set up the workspace for the easiest testing)\n - notice that the blender output is set to 'not set yet', as is the file output node in the compositor.\n - choose to render the animation\n # In the folder where the current blend file resides there will now be two additional folders, one for blender's output, and another for the file output node's results.\n\n\n**Results**\n\n**pre_render handler**\n\nAs expected the camera is incorrect because the pre_render handler happens before the scene has been evaluated. Although strangely some of the information is for the unevaluated frame that the playhead has just moved to (the frame number) and some is for the frame the playhead was on prior to beginning rendering (the camera name which was active on frame 4 before the playhead moved to the beginning of the animation ready for rendering in this example).\n\nBlender's own image sequence output does use the new path which was assigned whilst in the first frames pre_render handler, but on the first frame, the camera name is incorrect and the frame number is correct:\n\n\nThe file output node images however show that cycles didn't recognise that the base_path property had been updated until the next frame. We can see here that the file ending in __002.exr was updated on frame 1, but not used until frame 2, and this continues all the way through.\n\n\n\n**pre_frame handler**\n\nStrangely, although this handler also fires prior to the scene being evaluated (directly after the pre_render handler), this time Blender's own output shows the camera name is now correct, but again, the updated file output node is still not being used until the following frame:\n\nblenders output:\n\n\nfile output node's output:\n\n\n\n**post_frame handler**\n \nBlender's output again is correct, but the file output node's new path is not being recognised:\n\nBlender's output:\n\nFile output node's output:\n\n\n\nThe above can be worked around with code by grabbing the camera from the marker list instead of the active scene camera, but one really major issue for us is that even when locking the viewport prior to rendering, if we update the file output node's base_path, we start experiencing occasional Access Violations and instant crashes, this happens even if the file output node's path is not visible to UI. (perhaps this is related to the reason why the base path isn't being recognised until the next frame?)\n\n\n", "\"Reset to Default Value\" sets alpha incorrectly in [Material Properties > Viewport Display > Color]\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 511.09\n\nBroken: version: 3.5.0\n\nRight-clicking \"Reset to Default Value\" on [Material Properties > Viewport Display > Color] sets both the color's \"value\" and \"alpha\" bars to 0.8, when the alpha one specifically should reset to 1. Alpha resets incorrectly.\n\nRight-click \"Reset to Default Value\" on [Material Properties > Viewport Display > Color] on any material.\n\n", "When IME is On, passing key inputs excluding caret operation to IME. (Windows)\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: Intel(R) Iris(TM) Graphics 5100 Intel 4.3.0 - Build 20.19.15.4531\n\nBroken: version: 3.0.0 Alpha\n\nIn Translation set \"Japanese\", Interface is \"checked, with input box,\nWhen IME(Roma-ji) opend input puts a first letter to input box. (ex. 「あ」->「aあ」, 「か」->「kか」)\n\n\n\nI don't know how to post my modified branch's source yet.\nSo I paste the modifed hunks.\n----\n```GHOST_SystemWin32.cpp:1222\n```\n /* passing key inputs excluding caret operation (move, select) and deleting to IME */\n if ((!ctrl_pressed && !alt_pressed) &&\n (raw.data.keyboard.VKey != VK_BACK\n && raw.data.keyboard.VKey != VK_RETURN\n && raw.data.keyboard.VKey != VK_SHIFT\n && raw.data.keyboard.VKey != VK_CONTROL\n && raw.data.keyboard.VKey != VK_ESCAPE\n && raw.data.keyboard.VKey != VK_END\n && raw.data.keyboard.VKey != VK_HOME\n && raw.data.keyboard.VKey != VK_LEFT\n && raw.data.keyboard.VKey != VK_UP\n && raw.data.keyboard.VKey != VK_RIGHT\n && raw.data.keyboard.VKey != VK_DOWN\n && raw.data.keyboard.VKey != VK_DELETE)) {\n if (window->getImeInput()->GetOpenStatus(window->getHWND())) {\n return NULL;\n }\n }\n```\n```\n```\nGHOST_ImeWin32.h:271\n/* Retrieve the open status of IMM. */\nbool GetOpenStatus(HWND window_handle);\n```\n```\nGHOST_ImeWin32.cpp:385\nbool GHOST_ImeWin32::GetOpenStatus(HWND window_handle)\n{\n```\nbool result = false;\nHIMC imm_context = ::ImmGetContext(window_handle);\nif (imm_context) {\n result = ::ImmGetOpenStatus(imm_context);\n}\nreturn result;\n```\n}\n```\n----\n", "VSE image transformation in preview\nThis is design / TODO task to outline core concepts of this feature\n\nBasic functionality:\n - Feature will be intrgrated within existing transform system (should reuse pivot point settings probably)\n - Uses established keys to do location, rotation, scale - g, r, s\n - Operates on all strips, except of sound strip.\n - Uses selection from timeline\n - To disambiguate selection, outline and possibly overlay should be drawn (should be further clarified, as there are many ways to do this)\n - 2D cursor should be implemented (this may be problematic, because cursor shouldn't be visible normally)\n\nOther features:\n - Gizmos - probably best would be to have these accessible when appropriate tool is active. Having all visible at once could cause issues with \ndrawing\n - Gizmo display settings\n - Snapping\n - Selection of image rectangles in preview, sync to timeline \n - Implement global/local modes (shear support)\n\nOther changes, not quite related to this feature, but would unify design:\n - Move transform effect properties to strip transform properties\n - Move text effect positon properties to strip transform properties (multiline text stacking could be done using snapping?)\n - Animation - there should be at least `I` keyboard shortcut witk keyframing options similar to 3D editor.\n", "Keyframes with stereoscopy button doesn't work in animation, individual frame works\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 442.92\n\nBroken: version: 2.90 (sub 0)\nWorked: never, tested 2.75a (when Stereoscopy was introduced) and 2.79\n\nWhole animation is render with stereoscopy ignoring on/off animation keyframes. \n\nwith stereoscopy \"on\" insert a keyframe on it, move few frames and turn off stereoscopy and insert a keyframe. While rendering animation (my setup is FFmpeg with mp4 coding) all frames are rendered as stereoscopic. I attached an example file with render result. I was expecting to have half of this animation as stereoscopic and half normal. Originally I discovered it in a situation where I was swapping cameras (binding to video marker). \n\nOpen file, render frames 39-40-41 individually.\nin viewport change set frame to 39\nRender animation (5 frames)\nin viewport, set frame to 41, \nRender animation \nnotice difference in the two. \n[T76353Stereoscopy.blend](T76353Stereoscopy.blend)\n\n[bug reportstereoscopy.blend0001-0250.mkv](bug_reportstereoscopy.blend0001-0250.mkv)\n", "Cycles Image Texture Alpha conversion inconsistency with EEVEE and Image Editor\nOperating system: Windows 10 Pro 64-bit (10.0, Build 18363)\nGraphics card: NVIDIA GeForce 920M (Laptop)\n\nBroken: v2.83.13 Beta, 2020-04-21\n\nIf an image is packed, it will get displayed differently in Cycles and EEVEE. It changes in EEVEE when\nAlpha Mode is changed from Straight to Premultiplied to match Cycles, but it does not change in Cycles\nby changing that option. If the image is unpacked Cycles will look like EEVEE with Straight Alpha Mode\nand again it will not be changed by the Alpha Mode option like EEVEE is.\n\n1. Open the provided file.\n2. Select the Image Texture Node\n3. Go into the right panel in the Shader Editor (N-panel) into the Item tab and Properties subsection\n4. Render Image\n5. try changing Alpha Mode and rerendering.\n6. try unpacking the image and rerendering.\n7. try changing the Alpha Mode on unpacked image and rerender.\n[packed_image_cycles_display_inconsistency.blend](packed_image_cycles_display_inconsistency.blend)\n" ]

[ "FFmpeg unpremultiply 2.79\nWIn10 GTX980Ti\n\n2.79bd8ac9\n\nWhen OpenGL render, alpha processing when saving with FFmpeg differs between 2.78 and 2.79.\nIt is not Unpremultiply 2.79.\n\nAdd Background images > Draw Front > Opacity to lower > OpenGL render Animation(FFmpeg Video/2.78 H.264) save\n" ]

[ "Scene update bug\nSystem Information\nwindows7 64\nASUS ATI Radeon HD6870\n\nBlender Version\nBroken: (All)\nWorked: (None)\n\n\n\n\n\n\n\n\nopen this file, select plane and try to select for this plane any mesh - all this meshes don't have users, but was saved, try to select material - materials was saved, but don't have users, \nwe can revert, close and open again many times - materails and meshes will be saved.\ntry to add gruop instance - we will have all this groups is empty.", "Dae parsing issue\nOperating system:Windows 10\nGraphics card: NVidia gforce GTX660\n\n 2.79\n\n\nIt is a common error, when trying to upload an object to Second-life or Opensim, to be unable to upload a model,created in Blender, due to a \"DAE parsing issue\"\nThis object is one that is effected.\n\n[lowpol.blend](lowpol.blend)\nExport as .DAE file and upload to Opensim or Second-life", "AMD HIP with 7900XTX crashes using Cycles in Blender 3.6\nOperating system: Ubuntu 22.04.2\nGraphics card: RX 7900 XTX\nROCM/HIP is version 5.5.1\n\nBroken: version: 3.6.0, branch: blender-v3.6-release, commit date: 2023-06-27 08:08, hash: c7fc78b81ecb, type: release\nWorked: version: 3.5.0, branch: blender-v3.5-release, commit date: 2023-03-29 02:56, hash: 1be25cfff18b, type: release\n\nWhenever I try to render anything (default cube included) in 3.6 with Cycles using GPU compute I get the following error in the console:\n`:0:rocdevice.cpp :2788: 4112443167 us: 28547: [tid:0x7f38b247d600] Callback: Queue 0x7f3793600000 aborting with error : HSA_STATUS_ERROR_INVALID_ISA: The instruction set architecture is invalid. code: 0x100f\nAborted (core dumped)`\nDownloaded 3.5 and it works just fine there. Using CPU in 3.6 also works. Can't seem to find the coredump anywhere (did ulimit -c unlimited and even disable apport but in none of the directories where I would expect a coredump I got one). \n\nTo reproduce:\n1.Open Blender 3.6\n2.Change to Cycles with GPU compute\n3.Press F12\n\n", "Adding Clearcoat freezes Blender with error \"Execution of the command buffer was aborted due to an error during execution\"\nOperating system: macOS-13.3-x86_64-i386-64bit 64 Bits\nGraphics card: Metal API AMD Radeon RX 5700 XT 1.2\n\nBroken: version: 3.5.0\nWorked: Unknown\n\nAdding Clearcoat to a Surface freezes Blender, with Console logs showing only an \"Execution of the command buffer was aborted due to an error during execution. <private>\" error.\n\nOnly appears to happen when using Metal, using Blender in OpenGL mode doesn't have this issue. The issue is also present on brand-new projects with just the default cube.\n\n1. Create new file (Eevee renderer)\n2. Change the Clearcoat amount of the default cube from the Materials tab.\n3. Blender crashes usually within seconds, with the default cube I've found it freezes more easily if you use the sliding mechanism to change the Clearcoatness.\n\n", "Blender 3.6.3 SegFault Crash at startup on MacOS Ventura 13.5.2\nOperating system: MacOS Ventura 13.5.2 (22G91)\nGraphics card: Intel Iris Plus Graphics 650 1536 Mo\n\nBroken: 3.6.3 (3.6.3 2023-09-21)\nWorked: None tried on this machine\n\nCrash on startup with `Segmentation fault: 11`\n\nEither:\n- Double-click on Blender application icon\nOr:\n- command line : `/Applications/Blender.app/Contents/MacOS/Blender` with or without option `--gpu-backend`\n\n```\nThread 0 Crashed:: Dispatch queue: com.apple.main-thread\n0 Blender \t 0x109d2903d (anonymous namespace)::Local::Local() + 29\n1 Blender \t 0x109d28fe9 memory_usage_init + 25\n2 Blender \t 0x109d26f09 MEM_init_memleak_detection + 9\n3 Blender \t 0x101e098c3 main + 227\n4 dyld \t 0x7ff815a5241f start + 1903\n```", "MacOS ARM64 builds have issues saving WEBP images from 32bpp buffers\nOperating system: MacOS ARM64 buildbot\n\nBroken: 3.6 main 7699c7407d08e79580c36554206f14310f0f56c9\nWorked: n/a\n\nThe newly added `imbuf_save` is failing for certain scenarios but only on MacOS ARM64 builds.\n\nThis is a tracking bug so we can investigate the issue and re-enable the tests.\n\n- Uncomment the 2 WebP tests in `tests/python/bl_imbuf_save.py`\n- Run and observe large deviations in expected pixel values:\n\nThe first failure seems entirely incorrect while the second exceeds our 1% failure threshold @ 0.016.\n\n```\nwebp-rgb-70__from__rgba32.webp\n Mean error = 0.0058668\n RMS error = 0.01072\n Peak SNR = 39.3961\n Max error = 0.0862745 @ (77, 45, G) values are 0.04705883, 0.043137256, 0.0627451 vs 0.13333334, 0.12941177, 0.14901961\n 9791 pixels (59.8%) over 1e-06\n 2354 pixels (14.4%) over 0.016\n\n\nwebp-rgba-70__from__rgba32.webp\n Mean error = 0.00128407\n RMS error = 0.00366359\n Peak SNR = 48.7219\n Max error = 0.054902 @ (112, 62, R) values are 0.5058824, 0.25882354, 0.50980395, 1 vs 0.56078434, 0.3137255, 0.5647059, 1\n 5139 pixels (31.4%) over 1e-06\n 312 pixels (1.9%) over 0.016\nFAILURE\n```\n\n", "Dashed geometry nodes field link too hard to see on macOS\nOperating system: macOS-13.1-x86_64-i386-64bit 64 Bits\nGraphics card: AMD Radeon Pro 575 OpenGL Engine ATI Technologies Inc. 4.1 ATI-4.9.51\n\nBroken: version: 3.5.0 Alpha\n\nBreak line (I don't know word) see picture\n\nOnly for Mac. Do not use Windows.\n\n\n\n\nSee my picture on my iMac there.\n\n\nBrowser to show picture.\n0q6NTJs\n\n\n\nTheme on Nodes \"Dashed Lines Opacity\"\n", "Crash when selecting object on Mac\nOperating system: macOS-12.5.1-x86_64-i386-64bit 64 Bits\nGraphics card: AMD Radeon Pro 5700 XT OpenGL Engine ATI Technologies Inc. 4.1 ATI-4.8.101\n\nBroken: version: 3.3.0\n\nSystem crash trying to open einar_hair.blend file\n\nDownload einar_hair.blend fle from ?asset=6072\nSelect any object.\n\nBlender will crash. Also hair is not visible. It's possible, that OS will crash as well. ", "Cycles GPU Out of memory in viewport Metal Backend\nOperating system: macOS-13.4.1-x86_64-i386-64bit 64 Bits\nGraphics card: Metal API AMD Radeon Pro 560X 1.2\nAdditional Hardware : eGPU Razer Core X with AMD Radeon 5700 XT\n\nBroken: version: 3.6.0\nWorked: version 3.3.8 \n\nGPU Out of Memory in cycles with a simple scene one object and one Sun Light\n\nLoad Scene \nSwitch to render mode \nNavigate in the scene in rendered mode changing the view, after some pan and zoom the error gpu memory error is showed and blender crash. \n\nSame scene in blender 3.3.8 is working fine.\n\nBlender 3.6.0 is using 11.55 GB of Memory\nBlender 3.3.8 is using 566 MB of Memory with the same scene in rendered mode.\n\n", "Asset Browser: Material preview with black background when Cycles is disabled\nBroken: version: 3.0.0, 3.4.0\nWorked: Never\n\nWhen a scene was saved with Cycles but I disable the Cycles add-on the material previews get a black background.\n\n* Disable Cycles add-on.\n* Click on any (e.g., the Yellow) material asset and update its preview.\n[black_preview.blend](black_preview.blend)\n", "Array/Skin-Modifier-Combo causes crash\nOperating system: Linux Mint 19\nGraphics card: GT630\n\nBroken:\n2.80 (sub 39), branch: blender2.8, commit date: 2018-12-20 23:46, hash: 7a26e930a8c0\n\n\nPlease try to toggle the \"visibility\" of the skin modifier in my simple file. It crashes instantly and repeatingly:\n[Procedural leaves feathered1.blend](Procedural_leaves_feathered1.blend)\n", "Cycles rendered 3DViewport massive performance drop, using lights with node groups. --Intel MacOS--\nOperating system: macOS-13.4-x86_64-i386-64bit 64 Bits\nGraphics card: Metal API AMD Radeon Pro 5700 XT 1.2\n\nBroken: version: 4.0.0 Alpha\n\nSee screenrecording video! \nWhen switching to a light with nodegroups in Cycles rendered 3DView, performance drops a big time. (~50%).\n\n1. Open the performance.blend file with Intel MacOS blender 4.0, switch to cycles rendered 3DView, select the cube and rotate the view.\n2. Then switch to the caustic light and try to rotate the view again.", "Image Editor: Viewing a value buffer could use incorrect channel selector\nThis is a long outstanding issue in the image editor that should be fixed.\nWhen rendering a value based AOV and switching in the render result to the AOV would render incorrect as the wrong channel selector is selected (Still Color + Alpha).\n\nThis is also visible when reading single channel images or OpenEXR files.\n\n**Solution**\nAdd an option to view `Value` option to the channel selector and change RNA get/set/itemf + add support of this to the image engine.", "Image sequence render just stops on Cycles on MacOS Metal\nOperating system: macOS-13.3.1-arm64-arm-64bit 64 Bits\nGraphics card: Metal API Apple M1 Ultra 1.2\n\nBroken: version: 3.5.0\nWorked: none, tried 2.93, 3.3\n\nafter about 100 or so frames render just stops, Blender is still responsive, its not frozen or crashed, render bar just sits at %0. I'm doing motion tracking to get some kitbashed 3d elements into the scene. Plays fine in Eevee in the viewport. This is the error I get: CommandBuffer Failed: cycles_metal_shader_eval_displace. Attached ram usage for last 24 hours, I still have plenty of ram!\n\nI've closed the file and retried, several times, never got a render all the way through.\n", "Weird VSE crash\nOperating system: Win11\nGraphics card: RX580\n\nBroken: (example: 2.80, edbf15d3c044, master, 2018-11-28, as found on the splash screen) blender 4.0.0 alpha\nWorked: (newest version of Blender that worked as expected) no clue\n\nBlender crashes on opening file. Was trying to make an alternative caption that gets around the Multiply color issue #112267 but came across a new bug.\n\nBased on the default startup or an attached .blend file (as simple as possible).\nOpen the file I've attached\n\n" ]

[ "Eevee ambient occlusion is incorrect on M1 macMini\nOperating system: macOS-11.2.3-arm64-arm-64bit 64 Bits\nGraphics card: Apple M1 Apple 4.1 Metal - 71.0.7\n\nBroken version: 2.93.0 Beta\nNote: Earlier versions I tested (running with Rosetta 2) all have this bug.\n\nEevee ambient occlusion is incorrect on M1 macMini.\n\nFrom default scene:\n- add a plane\n- scale it 10 times\n- move it down 1 unit\n- switch to material preview or rendered viewport shading\n- enable ambient occlusion\n- to best see the issue, set its parameters to [ 1.0, 1.0, 0.0, false, false ]\n- issue already visible\n- render image to still see it\nNote: Every file, every combination of settings produce this issue to some extent.", "Different AO results between apple M1 and intel\nOperating system: macOS-11.4-arm64-arm-64bit 64 Bits\nGraphics card: Apple M1 Apple 4.1 Metal - 71.6.4\n\nBroken: version: 2.93.1\n\nDifferent AO results between apple M1 and Intel machines.\nOpening the same file in rendered on EEVEE on an M1 machine gives different results from an Intel one.\nSee attached image for visual reference \nPlease note that both Rosetta and native ARM are consistent in this result.\n\nSet view in rendered. The effect becomes move visible changing ambient occlusion distance and factor values.\nBased on attached .blend file [Apple_M1_AO_bug.blend](Apple_M1_AO_bug.blend)\n" ]

[ "Macro operators [e.g. Loop cut and slide]: \"adjust last operation\" fails completely [when called from the menu -- F9 is fine]\nBroken: 2.8x\nWorked: 2.79b\n\nThis might fall into the same boat as anything related to #45964, although it does seem to be \"new\" in 2.8x.\nAfter performing a loop cut and slide operator, you can adjust the properties in the operator panel (albeit with caveats from #45964). However, if you instead open the Adjust Last Operation panel, it will list a different operator name (Loop Cut, not Loop Cut and Slide), and while you can change properties there, those changes have no effect, even though they are reflected by the operator panel.\nThis didn't use to happen in 2.79b, though IIUC the \"Adjust...\" panel is different in 2.8x from the \"Redo Last\" of the 2.7x era.\n\n1. Load the default scene.\n2. Go into edit mode on the cube.\n3. Add a loop cut, e.g. via the shortcut or using the active tool\n4. Expand the operator panel in the bottom left and adjust properties. Note how those adjustments do have an effect on the result.\n5. Go Edit -> Adjust Last Operation\n6. Note the different operator name in that newly opened panel (Loop Cut vs. Loop Cut and Slide).\n6. Try adjusting any properties in the newly opened panel. Note how the operator panel in the bottom left picks up these changes, but at the same time they have no effect on the operation.\n\n[Demo video ](eAlL08M8AZA)", "weird viewport visual glitch\nOperating system: Windows 11 64 bits\nGraphics card: NVIDIA GeForce RTX 3060 Ti Gaming X Trio\n\nBroken: version: 3.3.1\nWorked: cant bother sorry\n\nits like the position of the viewport has an offset relative to the orientation of the camera, and the viewport has a set size and cant change.\n\nfirst go to the layout workspace and have an animation playing (must have an animation playing for the bug to happen), then create a new viewport by sliding it like im doing in the video. then change the viewport into rendering mode like im doing in the video and when you resize the viewport, it has a set size and cant be changed and the position of all parts have an offset according to the position and orientation of the camera.\n\n", "Use draw manager for clip editor drawing\nCould be stalled to 2.93 as there isn't an real issue here as far as we are aware of. The compositor issues are more tangable.", "Custom Properties Improved UI\nCurrent UI:\n\n\n\n- [ ] Allow reordering of custom properties in ways that make the most sense to an artist\n- [ ] Allow editing from the sidebar\n- [ ] Decrease the space between properties\n- [ ] Using an 'X' icon to delete\n- [ ] User and edit icon instead of \"Edit\"\n- [ ] Better location of icons\n- [ ] Group properties by the selected object/bone (see image)\n- [ ] Add a drop-down to select the data type\n- [ ] Allow the use of checkboxes to represent booleans\n\n", "Adjusting multiple properties at the same time doesn't work for all property types\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.55\n\nBroken: version: 2.90.1\n\nMost properties in blender can be changed for all selected objects by holding ALT while changing it. However, this doesn't work for all properties. For instance, the 'Rig Type' property in the Rigify add-on doesn't work. We can make it work by right click and selecting 'Copy to selected' but it should be possible with the ALT shortcut for consistency.\n\n**Steps to reproduce**\nWhile in Pose Mode select multiple bones\ngo into Bone Properties -> Rigify Type -> Rig type\ntry to change the Rig Type for all of them by holding ALT", "Channel numbers becomes unreadable with extreme zoom\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: Intel(R) UHD Graphics 600 Intel 4.5.0 - Build 27.20.100.8476\n\nBroken: version: 2.91.0 Alpha\n\n\n{[F8909669](image.png), size=full}\n\n\n", "File Browser and Preference Windows open at unusable position/scale.\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 441.66\n\nBroken: version: 2.82 (sub 7)\nWorked: 2.8\n\nOn a multi-monitor set up the File Browser and Preferences windows open up on the \"wrong\" monitor and insanely small. They require being resized *every single time* they are opened - does not remember last position or scaling.\n\nScreenshot of both monitors. The main monitor is a 4k and the secondary, on which I regularly run Blender, is a Cintiq 22HD (1080).\n\n\n\n\n\nOn a multi-monitor display, especially one with a 4k as main, open the File Browser and/or Preferences windows.\n\n", "emulate numpad disables modifier keys on smaller keyboards.\nOperating system: Darwin-20.1.0-x86_64-i386-64bit 64 Bits\nGraphics card: AMD Radeon Pro 575 OpenGL Engine ATI Technologies Inc. 4.1 ATI-4.0.36\n\nBroken: version: 2.91.0 Alpha Also, 2.83.5\n\nShortcut CTRL-1 for mac not work (Subdivision Set) when emulate numpad is enabled.\nNo modifier key (ctrl, command) takes effect when used with number keys.\n\n- Turn off these four shortcuts for desktop switching in system preferences: \n\n- Turn on emulate numpad.\n- Press 1, change the view to some arbitrary view, press command 1, change the view to some another arbitrary view, press ctrl 1. \n\nMacOS 11 Beta\nBlender 2.91 Alpha\nSeptember 14, 23:31:19- 26a827f25448\nmacOS 64 bit\n\nPush the Control + 1 not work for Mac make Subdivision Set\nsubdivision_surface.html#keyboard-shortcuts\n\nYoutube: 9f-bVIP0xDE?t=158\nTime: (2:40) \n\"CTRL - 1\"\n\nFor Mac only Not for Windows.", "MeshDeformModifier Crashing\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2080 with Max-Q Design/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 528.49\n\nBroken: version: 3.6.0 Alpha\nWorked: -\n\nAfter binding, with the Dynamic attribute driver value set to 1 in driver editor, it crashed.\n\n1. Add a driver to use_dynamic_bind.\n2. Click Bind\n3. Set the driver value to 1\n\n\n\n\n", "Multi accessors don't appear to be supported on collada import\nOperating system: Win 10\nGraphics card: RX 480\n\nBroken: Not working in 2.77, 2.8, 2.93 or an alpha version of 3.0 I had.\n\nWhen importing Collada with multi accessors per array (I've attached an image showing the difference), I get 'Sax FWL (Error): Positios stride is not three.'\n\nImport attached file called crate_bad.\n\n\n[crate_bad.dae](crate_bad.dae)\n[crate_good.dae](crate_good.dae)", "Decrement and Increment arrow buttons in Transform section of Details pane both operate in same direction (VM session))\nOperating system: Windows 10\nGraphics card: NVIDIA RTX A4500\n\nBroken: 3.5.0, 1b325cfff18b, blender-v3.5-release, 2023-03-29 02:56)\nWorked: \n\nDecrement and Increment Arrow Buttons in Details Pane both operate in same direction\n\nOpen a new .blend file.\nSelect the Default Cube.\nDecrement or Increment any of the transforms via the clickable arrow keys in the Transform section of the Details pane.\nThe increments and decrements for any of the numbers for rotation, scale or location are very unreliable as to what direction they operate in. Some sort of sign error?\n\n", "Knife Tool produces bad geometry in certain cases (disconnected edges/vertices, overlapping faces)\nif i use tools knife i see this bug.\n\n\n\n", "Changing Resolution Preset doesn't allow to Undo\nBroken: 3.6.0\n\nChanging Resolution Preset and then trying to undo doesn't work, like if the action is not registered.\n\n\n- Open Blender\n- Click on the Format panel on the right icon to change preset\n- Select a preset\n- Undo\n\n", "AutoMerge Keyframes only deletes unselected keyframes (scaling muliple selected to zero or snapping to the playhead fails)\n\n\nLinux Ubuntu & GeForce GTX 580\n\n\nBroken: master\n\nWith \"AutoMerge Keyframes\" toggled on it simply doesn't do that, whether you're scaling keyframes or moving them around.\n\n[auto_merge.blend](auto_merge.blend)\n\nMake an armature, go to pose mode and set keyframes on the bone. Scroll forward a few frames, move the bone around (and rotate) and key it again. Do this a couple of time so you have some keys to work with.\n\nIf you select all the keys in the dopesheet editor and scale them down so some of them land on the same frame, blender doesn't automerge them. The keys seem to still exist on the same frame, as you can see in the graph editor. Even if you select all the keys and move them by one frame (which should trigger the automerge) it still doesn't merge the keyframes.\n", "Multi-Object Properties Editing\nIn Blender, we want to make it easier to set values for multiple objects, bones, strips, keys and other items at once. \n\nCurrently you can do this in limited circumstances, but you have to right-click on a value after you've changed it and pick Copy to Selected from the menu or hold Alt while editing values. This not only slow, but also very hidden, and the Alt-key conflicts with Emulate 3 Button Mouse. Lastly, it isn't clear at all to users which values work for multi-editing, and which values don't.\n\nHere's how we would like to solve it:\n\n\n## Selection vs Active\n\nThe concept of ***active*** in Blender means that we only show properties and only affect the active object or item. However, if we want to better support multi-item editing, we must start by making it so the Properties reflects the ***selection*** rather than the ***active*** item, of which there can only be one. Going forward, by default the Properties will be set to reflect the selection instead. This could be controlled in the Properties like so:\n\n{[F8363009](Screenshot_2020-02-22_at_17.34.54.png), size=full}\n*(The search box is added here to reflect how the Properties will be updated when that item is also added, but is a separate feature)*\n\nIf set to ***Active***, the Properties will continue to function as it does today, only showing properties for the active item.\n\n## When there are multiple selected items\n\nWhen the user has selected multiple items, a few things change:\n\nInstead of showing the name of the object at the top, Blender tells the user how many items are selected:\n\n\nIn the list of tabs, only properties sections are shown that are common among the selected items. For example, if you select a text and a mesh object, we don't show Modifier, Mesh or Curve Properties, since these properties only exist on a subset of the selected objects:\n\n\nThis rule also goes for individual panels. Here, both a Sun and Spot light are selected, and we only see properties that are shared between them. A note at the bottom clarifies that you are only seeing the shared subset:\n\n{[F8364914](Screenshot_2020-02-23_at_18.49.33.png), size=full}\n\n## When values differ\nWhen multiple items are selected, Blender will compare values to check if they are the same, or if they are different. If they differ, we change the way properties are displayed, like so:\n\n{[F8363037](Screenshot_2020-02-22_at_17.52.51.png), size=full}\n\n\n\n## Manipulating values across multiple items\n\n\n\nFor simply setting a value to be the same. you would simply type a new value, hit Return, and that value is then propagated to the selection.\n\nFor manipulating numbers and sliders relatively, you can just drag on them (like Alt-dragging today), or type the character '=', so the user can type '= -3' to subtract 3 from the value on all selected items." ]

[ "Changing multiple object dimensions doesn't work\nOS: Windows 7\nGraphics card: ATI Radeon HD 5700 Series\n\n2.78a 4bb1e22\nWorked: (optional)\n\nChanging X,Y and/or Z parameters in Dimensions doesn't work\n\nIn 3D View, I{[F377283](changing-parameters-bug-2.78.wmv)} created an object, in Object Mode, Properties tab \"N\", under Transform / Dimensions, I clicked and dragged over X,Y parameters to highlight, entered a number then clicked Enter, the parameters haven't changed, I tried to change single parameter once at a time, it worked. I repeated the same steps for Transform / Scale, it worked, no problem, the problem is only in Dimensions. \n\nThank you", "Multi-Number Button Editing doesn't work on Y value of the objects Dimensions on the Transform Properties Panel \nOperating system: Windows 8.1 x64\nGraphics card: GeForce GTX 560\n\nBroken: 2.70 2525f9c = official RC2\nBroken: 2.70 211f08d win64\nBroken: 930765f win64-vc12 (most recent experimental x64 build)\nBroken: 2.70 930765f win32\n\nTested in another computer = yes\nSystem information of the second computer = the same of the one mentioned above (Windows 8.1, GTX 560)\n\nUsing the new interface feature \"Multi-Number Button Editing\" on the Dimensions values of the Transform Properties Panel (shortcut=N) doesn't never change the Y value, and the X value sometimes doesn't change, sometimes changes a lot less than the Z value (the only value that changes normally).\n\nDownload 2.70 RC2 or most recent builds from download/\nSelect the default cube\nPress N to open the Transform Properties Panel\nClick and hold the X **dimension** value, drag the mouse down to select Y and Z, then move the mouse left or right (this is the new Multi-Number Button Editing feature)\nRelease when the value is different than 2.000\nThe Y value doens't change at all, while the X value changes a little bit, or sometimes nothing. The Z value changes normally.\n\nAlso, apparently the \"Lock\" button next to the values in the Transform Properties Panel ignores the Multi-Number Button Editing. A suggestion would be that it actually locked the value when using this feature, so that you could lock a middle one, like Y, while changing X and Z simultaneouly.", "Can't scale on all axes at the same time.\nOperating system: Windows 10\nGraphics card: Nvidia GTX 1050 TI\n\nBroken: 2.8, dafaa6f22881, 2018-12-22.\n\nWhen I scale any object using the Dimensions field in the Properties menu (the N-menu) on one axis then click and drag that value to the other 2 axes, it only gets copied to 1 of them. If I try this same process using the Location, Rotation, and Scale fields it works on all 3 as intended. \n\nHere's a video demonstration of the problem: incompatiblerelievedkentrosaurus", "Failed to edit multiple dimensions prop\nWindows 8.1 x64\nRenderer: GeForce GTX 860M/PCIe/SSE2\nVendor: NVIDIA Corporation\nVersion: 4.4.0\n\nBroken: 2.72.1 6c2ce7a\n\nFailed to edit multiple dimensions prop at once.\n\nSelect default cube and edit more than one dimensions prop at once.\n\nHere is the GIF animation to illustrate the issue:\n", "Multi-Number Button Editing Fails in Lattice Tab\nWin8.1 64bit | GTX680\n\nBroken: Blender 2.72.2 5c6e333\n\nWhen experiencing multi-number button editing in Lattice's data tab, it can toggle them all, but fails to change all three values by typing numbers.\n\nI know it is not critical. Just point it out. Because it allows ths feature, and you can even change values by dragging towards left or right. The only problem is that values cannot be changed by keyboard. :)\n\n", "Multiple dimension editing does not work correctly. \nWin7 64bit, GTX 560 viewport, GTX 970 render\n\nBroken: 2.75 official\n\nIn the \"N\" side panel under Transform there is Dimensions x, y and z. You can't edit more than one of them at the same time. The Scale works as expected.\n\nSelect an object\nClick and drag over Dimension X, Y, Z\nEnter a new value, Enter, nothing changes\nClick and drag again, this time drag the mouse also to the side - only one of the Dimensions changes\n", "Multi-Number Button Editing doesn't work for dimensions fields\nOS: Mac OS X 10.8.5 \nVideo: NVIDIA GeForce 9400 256 MB\n\nBroken: 2.69.11 f469cbe, or testbuild1 2.69.11 4789793\nWorked: none\n\nIn 3D View and Object Mode -> in Preperties panel Dimensions fields do not change correctly if I try to select all 3 fields (using multi-number button editing) and drag left or right or insert a number manually. \n\n", "Multi button editing doesn't work on object dimensions.\nArchlinux\nGTX 460\n\nBroken: testbuild 2 (517094a) and 2.70\n\nUsing the multi button editing (click and vertical drag on vertically arranged number buttons) on the object dimensions (in the properties region) will not change their values.\n\n\n1. Open Blender (2.70 or testbuild 2)\n2. Click and drag down on the X object dimension in *3D view > properties region (N) > Transform > Dimensions* to select the Y and Z axis dimensions as well\n3. Enter a value\n4. Press Enter\n5. The dimensions revert to the original values\n", "Transform Dimensions only scale UP\nWindows 10 and NVIDIA GeForce GTX 970M\n\nBroken: (2.79b) fbdc9f9d68b\n(could not figure out which build was for build-comparison testing, as they aren't well labeled for an outside user, apologies)\nUnable to test on another computer.\nDisabled my non-default addons, and problem still occurred.\n\nN->Transform->Dimensions only multiplies scale.\nI have a model that was imported (FBX) for parts (broke shape keys was replacing with backup) and when I tried to scale the dimensions under the transform tab. If I move one of the Dimensions meters, (say X) then the other two (Y and Z) will start multiplying their scale. I ended up with a model that was dimensions ~(200 200 0.04) after dragging the Z scale to the LEFT.\n\nIf you scale the skeleton, there is no problem. If you scale the Mesh, it goes haywire.\nRegular cubes and other basic shapes do not encounter this problem. (also may be responsible for Shapekeys going haywire as well, but unable to isolate that other problem's trigger so far)\n\nOpen the Object Mode's N menu and attempt to Scale the object Mesh by selecting the Mesh and changing values in the Dimensions option.\n\nDownscaled file with the same problem present.\n[RonSampleExport.blend](RonSampleExport.blend)\n\n", "Modifying multiple parameters not right on Dimensions\nwindows 10 64bit, GTX650\n\nBroken: 2.78a e8299c8\nWorked: (optional)\n\nModifying multiple parameters not right on dimensions\n\nJust open blender, selece the default cub, than press \"N\" to expand the Properties panel, Press LMB select muti parameters on the Dimensions(xyz) and drag it, you can find it not work right, and input by the keyboard have the same problem , sorry for my bad english: )\n\nthere is a gif to show it:\n\n\n", "changing all 3 dimensions at the same time doesn't work\nAll\n\nBroken: 2.70 and latest SVN\nWorked: (optional)\n\nChanging multiple axis at the same time with the new click-drag function works for Scale but not dimensions.\n\nDefault scene, Click-Drag mouse on \"Dimensions\" in the N Panel to change (x,y,z) at the same time\nEnter 2 for example\nResult: Doesn't work, 1.000 stays.\nSliding left-right only changes x and z not y.", "Multi-Button Drag: not working properly for Object Dimensions\nWindows 7\n\nBroken: 2014-02-11\n\nMulti-Button Drag - D270 - doesn't work properly in N panel for Dimensions\n\nIf you drag over all 3, only X and Z change, Y remains unchanged. As mouse button is released, the first selected property will reset, only the last hovered is kept (and scale changed)", "Click and drag across dimensions in X Y Z doesn't work.\nWin 7\n\n2.74\n\n\nIt is a very minor bug, but a bug nonetheless.\n\nWhen setting a value for different fields in blender, it is possible to click and drag across multiple values and then input identical value for all of them, but this does not work for Dimensions in N menu in 3D View for some reason. When I click and drag across the dimensions in X Y Z axes and input the value, nothing happens. This works for Scale values which are right above Dimensions for example.\n\nThe fields for dimensions values I am talking about are marked blue, the value I'm inputting for all of them is 55.\n", "Multi field entry in Dimensions not working\nWin 8.1 AMD64 Radeon\n\nAffects: any version of Blender that has multi field entry.\n\n\"Multi field entry'\" is what I am calling that very nice recent feature that allows you to drag down over more than one numerical entry field and change them simultaneously, by dragging or entering a value.\n\nThis works correctly in every field I have tried except for dimensions in the 3Dview N properties region. This behaviour is not affected by having transforms unapplied or applied. \n\nTo reproduce, first use the scale field to see how it should work. Mouse over all three and start ragging. The three fields update, keeping the offset if one was present, and the object in the window changes in all three axes. \n\nNow in the dimensions box mouse and drag over all three . Only two fields change, and only the Z has any effect in the window. (Watch the scale fields to see what is really happening.) Now drag over two fields. Only one has any effect. There is also a 'snapback' type effect that happens when you select more than one field and enter numerically instead of dragging. the fields take your entry until you press enter, and then go back to what they were before. \n\nNote that dragging or entering values in any single dimension field works correctly and as expected.", "Multiple change of dimensions doesn't work properly\nCentOS, Debian, Nvidia Titan\n\nBroken: 2.77 official (hash 22a2853 2016-03-18 12:34)\nWorked: don't know\n\nMultiple change of dimensions by drag-selecting and side-pulling or manual enter doesn't work properly.\n\n1. Try to multiple select fields by dragging, then enter some values by keyboard. In case the three values were equal it won't change them at all.\nIn case they are different, it'll make x and z equal x, but not the one you've entered. The same behavior if you just hit enter without entering anything.\n2. Try to multiple select fields by dragging, then drag to the side. It'll show you the change in x and z, but leave the chages only in z after you release the button. It happens in any cases.\n\nI know there were entries of this bug previously (not sure about exact behavior), but they marked as closed. So i guess it's broken once again.\nAll other fields like location, scale, etc are working.\n[dimensions_set_bug.blend](dimensions_set_bug.blend)\n", "drag editing on Dimensions only effects one scale\nWindows 7 x64,gtx 580\n\nBroken: 01645cc (buildbot)\n\nDrag editing on the Dimensions only effects the last one selected.\n\n\n1. open new scene\n2. click on cube\n3. open properties tab\n4. select and drag from the x dimension to the z dimension\n5. only the x and z dimension values change, and only the z scale changes." ]

[ "Exact Boolean (Union & Self Intersections -- to remove inside geometry) bad performance/memory footprint\nOperating system: Linux-5.15.0-67-generic-x86_64-with-glibc2.31 64 Bits\nGraphics card: Mesa DRI Intel(R) HD Graphics 4600 (HSW GT2) Intel Open Source Technology Center 4.5 (Core Profile) Mesa 21.2.6\n\nBroken: version: 3.5.0 Beta\n\nI have a group of nodes that gets rid of internal geometry by creating a \"mold\" of an object.\n\nBut it crashes on some meshes and not on other meshes and the crash doesn't give any real information why.\n\nIs there a work around for this?\nNodes works with certain meshes and crashes with others.\n\nWorking mesh and blend file.\n\n**Left side shows internal / overlapping geometry and Right side is free of this after using node network.**\n\n\n\n[doesn't crash.blend](attachment)\n\n\nCashes when I **click the switch node to enable it** (**see red arrow**).\nCrashes in 3.4.1 and 3.5.\n\n\n\n\n\n[this one crashes.blend](attachment)\n\nCrash text files attached below\n\n\nIs there a work around for this?\n\n", "driver on an NLA animated time strip isn't purple and cannot be removed\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 1070 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 460.79\n\nBroken: version: 2.93.0 Alpha\n+ version: 2.92.0\nWorked: never\n\nWhen adding a driver on an NLA animated time strip value, the value does not turn purple, and cannot be remove.\nDon't know if it's related or not, but you also cannot delete/clear keyframes on that particular value aswell. Should i do another bug report ?\n\n(see video attached)\nanimate the default cube (example : revolution on Z axis, from frame 1 to 250)\ncreate an empty, add a custom property, animate that property\ncopy the custom property as a new driver\ngo to the NLA, push down the cube action, go to it's strip option and check \"animated time strips\"\npaste the driver on the \"animated time strips\" value\n[bugreport_drivercolor.mp4](bugreport_drivercolor.mp4)", "Random crashes with NVidia + GPU Subdivision (sometimes when activating X-Ray or Wireframe overlay)\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\n| Graphics card: | Driver : | User:\n| --- | --- | ---\n| NVIDIA GeForce RTX 3090/PCIe/SSE2 | 4.5.0 NVIDIA 516.94 | @GabrielMoro\n| NVIDIA GeForce GTX 1050 Ti/PCIe/SSE2 | 4.5.0 NVIDIA 512.15 | @STYXtheMUTT\n| NVIDIA GeForce RTX 3070 Ti/PCIe/SSE2 | 4.5.0 NVIDIA 516.94 | @Jojolix\n|NVIDIA GeForce RTX 2070 SUPER/PCIe/SSE2|NVIDIA 516.59|@MassimilianoPuliero\n|NVIDIA GeForce RTX 3090|NVIDIA 516.94|@The5\n\nBroken: version: 3.2.2\n\nOn some seemingly random occasions, Blender crashes with the following stack:\n```lines=10\nStack trace:\nblender.exe :0x00007FF77C1458B0 BKE_subdiv_free\nblender.exe :0x00007FF77C147880 BKE_subsurf_modifier_subdiv_descriptor_ensure\nblender.exe :0x00007FF77BFFC3D0 <lambda_c000e58b9a8caf72c54e0e02a46effdb>::operator()\ntbb.dll :0x00007FFDBAF64FD0 tbb::interface7::internal::isolate_within_arena\nblender.exe :0x00007FF77BFFC6F0 BKE_mesh_wrapper_ensure_subdivision\nblender.exe :0x00007FF77CB38FA0 raycast_obj_fn\nblender.exe :0x00007FF77CB375E0 iter_snap_objects\nblender.exe :0x00007FF77CB3E530 transform_snap_context_project_view3d_mixed_impl\nblender.exe :0x00007FF77CB3F0E0 ED_transform_snap_object_project_view3d\nblender.exe :0x00007FF77CB5E7C0 applyFaceProject\nblender.exe :0x00007FF77CB5EE00 applySnappingIndividual\nblender.exe :0x00007FF77CB72F70 recalcData_objects\nblender.exe :0x00007FF77CB81010 applyTranslation\nblender.exe :0x00007FF77CB46BE0 transformApply\nblender.exe :0x00007FF77CB41920 transform_modal\nblender.exe :0x00007FF77C229D30 wm_macro_modal\nblender.exe :0x00007FF77C20F9E0 wm_handler_operator_call\nblender.exe :0x00007FF77C210BD0 wm_handlers_do_intern\nblender.exe :0x00007FF77C210000 wm_handlers_do\nblender.exe :0x00007FF77C216590 wm_event_do_handlers\nblender.exe :0x00007FF77C1FC380 WM_main\nblender.exe :0x00007FF77B2F12A0 main\nblender.exe :0x00007FF7814128B0 __scrt_common_main_seh\nKERNEL32.DLL :0x00007FFDC7B67020 BaseThreadInitThunk\nntdll.dll :0x00007FFDC85A2680 RtlUserThreadStart\n```\n```lines=10\nStack trace:\nblender.exe :0x00007FF7A50C5B10 BKE_subdiv_update_from_mesh\nblender.exe :0x00007FF7A50C7880 BKE_subsurf_modifier_subdiv_descriptor_ensure\nblender.exe :0x00007FF7A52A1110 draw_subdiv_create_requested_buffers\nblender.exe :0x00007FF7A52A2080 DRW_create_subdivision\nblender.exe :0x00007FF7A52D1870 DRW_mesh_batch_cache_create_requested\nblender.exe :0x00007FF7A52C6000 drw_batch_cache_generate_requested\nblender.exe :0x00007FF7A529F1A0 drw_engines_cache_populate\nblender.exe :0x00007FF7A529B8C0 DRW_draw_render_loop_ex\nblender.exe :0x00007FF7A529C9B0 DRW_draw_view\nblender.exe :0x00007FF7A5D3FA00 view3d_main_region_draw\nblender.exe :0x00007FF7A5551070 ED_region_do_draw\nblender.exe :0x00007FF7A51A3700 wm_draw_window_offscreen\nblender.exe :0x00007FF7A51A3560 wm_draw_window\nblender.exe :0x00007FF7A51A2FB0 wm_draw_update\nblender.exe :0x00007FF7A517C380 WM_main\nblender.exe :0x00007FF7A42712A0 main\nblender.exe :0x00007FF7AA3928B0 __scrt_common_main_seh\nKERNEL32.DLL :0x00007FFD1C4A54D0 BaseThreadInitThunk\nntdll.dll :0x00007FFD1D524830 RtlUserThreadStart\n```\n\nOccasions observed:\n- try to snap objects on faces (the object that shall be snapped to is usually a complex mesh with modifiers like subdivisions, mirror, solidify, bevel and booleans)\n- toggle the shading mode between shaded and wireframe or between X-Ray and Wireframe Overlay.\n[intro.crash.txt](intro.crash.txt)\n\n\n- open attached file\n- disable and enable X-Ray multiple times\nIt happens specially when working fast.\n\n[BLENDER_Youre_Alive_v67.blend](BLENDER_Youre_Alive_v67.blend)\n\n", "Crash When Adding Hair Subdivision upon Viewer node\nOperating system: Windows-10-10.0.17134-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 970M/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 466.11\n\nBroken: version: 3.4.0 Alpha\nWorked: Never\n\nCaused by: c55d38f00b\n\n\n\nAdding hair subdivision will cause crash immediately when you are using viewer node to visualize field values in the viewport.\nNote no crash will occur if the viewer node is inactive.\n\n[Hair Crash.crash.txt](Hair_Crash.crash.txt)\n\n1. select default cube\n2. shift+A to add empty hair\n3. in the node tree add viewer node.\n4. Crash if adding subdivision\n[Hair Crash.blend](Hair_Crash.blend)", "Crash when rendering animation\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA RTX A4500/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 512.15\n\nBroken: version: 3.4.1\n\nBlender crash when rendering an animation\n\nIn an attempt to simplify my rendering of multiple scenes, I added an additional scene to a blend file with no objects itself. Its purpose is to be a place to put multiple *Render Layer* nodes in its compositor for all the other scenes in the file. In other words I hit \"Render\" when on that special scene and all other scenes are rendered in one go. Running this render as a single image works fine. Running render animation crashes after one frame.\n\nMy actual file has 7 scenes to be rendered. But to help demonstrate and reproduce this, I have simplified the file to just two scenes:-\n\"*Scene 1*\" is a simple scene containing a cube and a camera. Rendering this scene directly, works fine as still image or animation.\n\"*Render all*\" is a second scene that has no objects of its own (just a camera because my render farm seems to insist upon that). This scene is set to not be rendered (\"*Use for Rendering*\" unticked). However it's compositor has a *Render Layers* object which looks at *Scene 1* and which is connected to a *File output* node.\n\n\n[crash-test3.blend](crash-test3.blend)\n\nOpen file and render animation.", "Crash handling collections in a timer\nOperating system: Mac OS Monterey version 12.0.1\nGraphics card: Radeon Pro 555X 4 GB\n\nBroken 3.0.1\nbroken blender-3.2.0-alpha+master.d3b1cce4000b-darwin.x86_64-release.dmg \n\nBuilding Nodes add-on causes Blender to crash consistently when editing nodes in Building Style Editor window. I've been in communication with the author off the add-on (user name Durman on GitHub), and he has provided the script and instructions to reproduce the crash below.\nCrash log [blender.crash.txt](blender.crash.txt)\nAsan report [F12872713](T95861_asan.txt)\n\n[#95861.blend](T95861.blend)\n\n1. Open the attached file\n4. Push execute button of the text editor\n5. Wait until the crush\n\n", "Quick Smoke crash guide February 3, 2023\nOperating system: macOS-13.2-x86_64-i386-64bit 64 Bits\nGraphics card: Metal API AMD Radeon Pro 575 1.2\n\nBroken: version: 3.5.0 Alpha\n\nBake ALL button within Sample 128 from Frame 1 to 220 cause CRASH GUIDE\n\nBake ALL button within Sample 128 from Frame 1 to 220 cause CRASH GUIDE\n\n[image.png](image.png)\n[image.png](image.png)\n[Blender-2023-02-03-113746.ips](Blender-2023-02-03-113746.ips)\nTest File: [3.5.zip](3.5.zip)\n```\nCall Stack: Main thread\n__psynch_cvwait\n_pthread_cond_wait\ntake_gil\nPyEval_RestoreThread\nPyGILState_Ensure\npy_timer_execute\nBLI_timer_execute\nwm_event_do_notifiers\nWM_main\nmain\nstart```", "Decrease SharedCache overhead\n## Background\nAdding more uses of `SharedCache` to mesh/curves runtime data might\nbe a performance concern in the long run if we continue to add more.\nTo test this, I made a loop of the worst case for the shared cache concept.\n```\nSCOPED_TIMER_AVERAGED(__func__);\nfor ([[maybe_unused]] const int i : IndexRange(5000)) {\n Mesh *mesh = BKE_mesh_new_nomain(2, 1, 0, 0, 0);\n Mesh *copy = BKE_mesh_copy_for_eval(mesh, true);\n BKE_id_free(nullptr, mesh);\n BKE_id_free(nullptr, copy);\n}\n```\nThis is the worst case because none of the caches are used and the mesh\nis tiny, just one edge, so the cache allocation is more likely to stick out.\nAnd we allocate 10000 meshes.\n\nI ran this with current master, and then again after adding 40 shared caches\n( `SharedCache<Array<int>>`) to the mesh runtime struct.\n| | master | 40 extra shared caches |\n| -- | -- | -- |\n| Time to create, duplicate, free 5000 1-edge meshes | 17.4ms | 21.5ms |\n\nIt's clear that there is some overhead, but given the extreme situation, not _that_ much.\n\n## Solutions\n\n- Avoid extra allocation when copying/moving from an existing `CurvesGeometry`/`Mesh`\n- Use some kind of free-list instead of using the allocator directly. Doesn't really solve the scalability issue, but can help anyway. Not really worth doing for now though.\n- Combine multiple shared caches into one.\n- Only allocating the shared cache pointer when a copy is done (protected by a separate mutex).\n- Only allocate the shared cache eagerly when there is certain minimum number of points.\n", "Crash when loading a geometry nodes asset into itself\nOperating system: Win 11\nGraphics card: 2070\n\nBroken: 3.4, 4.0\nWorked: none\n\nWhen working on an geometry nodes asset in an asset library - if you search for the name and adding it to the graph it causes a crash. \n\nDrag n drop from the asset browser behaves correctly here - so I think its just the node search code which is missing the check for self referencing.\n\n1. Download the blend file.\n2. Put into a folder.\n3. Add folder as asset library in prefs -> file paths -> asset libraries.\n4. Open the blend file from that folder.\n5. Go to the geometry node graph (should be open already).\n6. `Shift` + `A` to add new nodes.\n6. Search \"noop\" and press enter.\n\n", "Compositor: Continously add/remove for cryptomatte.\nSource: 24223\n\nPROBLEM\n\nSometimes when using Cryptomatte the objects to isolate and divide into various passes can be a lot, and the process becomes tedious because you can only Add or Remove one object per time, then go back to the Cryptomatte node and click the + or - icon again, slowing a lot the process. Instead, it would be way more handy being able to select multiple specific objects.\n\nSOLUTION\n\nHave Add and Remove to work with multiple clicks, and then press Esc to go out of the selection mode when needed.\n\nNOTE: although categorized as good first issue, it might require more changes than expected as UI code is shared with color picker. Of course the color picker workflow should not change.", "Grease Pencil: Line Art › Light Contour causes crash\nOperating system: macOS-11.6.5-x86_64-i386-64bit 64 Bits\nGraphics card: AMD Radeon R9 M295X OpenGL Engine ATI Technologies Inc. 4.1 ATI-4.6.21\n\nBroken: version: 3.6.0 Alpha\nAlso reproduced with 3.5 and 3.4.1\n\nWhile using the Grease Pencil Line Art modifier, enabling the Light Contour feature causes a crash. The problem does not appear with a very simple scene, so a certain amount of geometry may be needed, hence the 6mb sample file. While trying to reproduce with smaller scenes, I sometimes had to toggle the Light Contour checkbox on and off a few times to trigger the crash, and the number of toggles actually varied from one launch to the next.\n\n1. Open provided file, GP crash.blend\n2. Select Grease Pencil object and expand its Line Art modifier\n3. Click Light Contour checkbox to enable it\n\nThis reliably causes a crash on my Intel Mac\n\n", "Quick favorites: Add previously removed behavior causes duplication\nOperating system: windows 10\nGraphics card: RTX 3090\n\nBroken: Current main.\nWorked: N/A.\n\nIf you remove one item from Quick favorites, the next time you will be able to add it again.\nThis cause duplication:\n\n\n1. Add some items to Quick Favorites.\n2. Remove some items.\n3. Sometimes you will able to add previously existing items back to the list again. (see attached video examples)", "GPencil: Randomize brush's 'Position' attribute doesn't work when camera is too close to Grease Pencil object\nOperating system: Win10 Home 64\nGraphics card: rtx3070\n\nBlender Version\nBroken: 2.93, 3.1.0\nWorked: N/A\n\nExact steps to reproduce the error:\n\n- Open file\n- try to use Randomize brush\n- Strength and Thickness work, but not Position \n- Switch to the other camera that's farther away from the GP object, and randomizing the position works fine.\n\n[randomize_not_working_close_NEW.blend](randomize_not_working_close_NEW.blend)\n\n\nwatch?v=4pTrEm5HqVU", "Crash on Changing Repeat value to -1 in Corrective Smooth modifier\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2080 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 496.49\n\nBroken: version: 2.93.5\n\nNot Responding when I change the Repeat value to -1\n\n{[F11810617](image.png), size=full}\n\n", "Mixing sequencer and preview regions causes logical problems\nWith the addition of many editing operations to the sequencer preview,\nthere are an increasing number of logical problems with parts of the\nUI that operate on the area, not the region.\nMainly menu items and the tool-system.\n\nNote that this is only a problem with both are enabled at once.\n\nTo pick some examples:\n\n- Delete from a menu (should this delete the selection in the preview or the sequencer?)... same for select-all... and in the future copy/paste.\n- Transform (Move for e.g. is a different operation for the sequencer and preview).\n- Box select from the menu currently can't run in both sequencer and preview regions.\n\nIf the code-base is left in it's current state I'm concerned users will run into glitches/bugs (finding the mixed preview/sequencer to be broken).\nFor the most part these won't be things that can be resolved in isolation, so I think it's worth considering a solution now.\n\n----\n\n**Supporting Mixed Preview/Sequencer**\n\nTechnical solutions for this are possible but impractical in my opinion, for example the operator_context could support multiple items:\n\n```\n layout.operator_context = {'INVOKE_REGION_PREVIEW', 'INVOKE_REGION_WIN`}\n```\n\nFor the sake of argument we could have 2x menu items for Delete, Select-All... etc.\n\nOther options include introducing an \"Active Region\" concept or having a header for each region.\n\nThis amounts supporting functionality in separate regions that areas already handle.\n\n---\n\n**Proposed Solutions**\n\nNote that I don't have a strong opinion on the exact solution,\nonly that it should not over complicate the current code-base\n*(unless we wish to support larger changes to Blender's windowing in general).*\n\nWhen mixing preview/sequencer:\n\n* Remove the preview region interactive behavior.\n\n```\nThis would make it view-only, any interactions, menu items, tools etc\nwould only apply to the sequencer.\n```\n\n* Remove support for mixing preview/sequencer at once.\n\n```\nWe could handle this with versioning (splitting the area).\n```\n\n" ]

[ "Regression: Crash when recursively nesting NLA meta strips\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 497.09\n\nBroken: version 3.2.0 stable, version: 3.3.0 Alpha\nWorked: 2.93.9 (I don't have other versions available to test right now)\nCrash caused by 63cf0d0890\n\nWhen creating a meta strip in NLA, which contains another meta strip, Blender crashes.\n\nFrom this file:\n[bug_recursivemetanlacrash.blend](bug_recursivemetanlacrash.blend)\nCombine the first two strips into a meta (ctrl+G), then combine the resulting meta with the third strip (ctrl+G) and observe a crash.\n\n```\nError : EXCEPTION_ACCESS_VIOLATION\nAddress : 0x00007FF707E62A24\nModule : blender.exe\nThread : 00003218\n```", "Regression: Crash after moving NLA action strips\nOperating system: Linux-5.17.9-1-MANJARO-x86_64-with-glibc2.35 64 Bits\nGraphics card: AMD VERDE (LLVM 13.0.1, DRM 2.50, 5.17.9-1-MANJARO) AMD 4.5 (Core Profile) Mesa 22.0.4\n\nBroken: version: 3.2.0, 3.3.0 Alpha\nWorked: 2.91.2, 2.93.0, 3.1.2\nCaused by 63cf0d0890\n\nIf you try to move NLA strips, Blender crashes, no matter if they are linked or local.\nMy initial assumption was that the crash requires them to be linked, but that was false. So while I attached both files, \"... destination\" isn't strictly required for the crash, but if you try it there, it also crashes.\n\n\n [only in case of \"CRASH NLA - Source.blend\": push the action down, duplicate it at least two times on the same NLA track, distance does not matter]\n\n- Open `CRASH NLA - destination.blend`\n- select all three action strips,\n- Move them with mouse drag or {key G} key\n- Blender crashes.\n\n[CRASH NLA - Source.blend](CRASH_NLA_-_Source.blend)\n[CRASH NLA - destination.blend](CRASH_NLA_-_destination.blend)\n\n```lines=10\nBlender 3.2.0, Commit date: 2022-06-08 10:22, Hash e05e1e369187\nbpy.ops.action.push_down() # Operator\nbpy.ops.nla.duplicate(linked=False, mode='TRANSLATION') # Operator\nbpy.ops.transform.transform(mode='TRANSLATION', value=(20.5877, -105.42, 0, 0), orient_axis='Z', orient_type='GLOBAL', orient_matrix=((1, 0, 0), (0, 1, 0), (0, 0, 1)), orient_matrix_type='GLOBAL', mirror=False, use_proportional_edit=Fa>\nbpy.ops.nla.duplicate(linked=False, mode='TRANSLATION') # Operator\nbpy.ops.transform.transform(mode='TRANSLATION', value=(31.6585, -54.216, 0, 0), orient_axis='Z', orient_type='GLOBAL', orient_matrix=((1, 0, 0), (0, 1, 0), (0, 0, 1)), orient_matrix_type='GLOBAL', mirror=False, use_proportional_edit=Fa>\nbpy.ops.transform.transform(mode='TRANSLATION', value=(-11.0708, 5.02, 0, 0), orient_axis='Z', orient_type='GLOBAL', orient_matrix=((1, 0, 0), (0, 1, 0), (0, 0, 1)), orient_matrix_type='GLOBAL', mirror=False, use_proportional_edit=Fals>\nbpy.ops.nla.select_box(tweak=True, xmin=46, xmax=476, ymin=79, ymax=191, mode='SET') # Operator\n\nbacktrace\n./blender(BLI_system_backtrace+0x20) [0xc1491e0]\n./blender() [0x11d8e2a]\n/usr/lib/libc.so.6(+0x3e8e0) [0x7ff8ed08f8e0]\n./blender() [0x12760bf]\n./blender() [0x12760f2]\n./blender(BKE_nla_tracks_copy_from_adt+0x53) [0x12767c3]\n./blender(BKE_animdata_copy+0x57) [0x158c467]\n./blender(BKE_libblock_copy_ex+0x1b3) [0x1243883]\n./blender(BKE_id_copy_ex+0x73) [0x1243903]\n./blender(_ZN7blender3deg34deg_update_copy_on_write_datablockEPKNS0_9DepsgraphEPKNS0_6IDNodeE+0x179) [0x1913399]\n./blender() [0x1912382]\n./blender() [0x1912552]\n./blender() [0xc14cf5d]\n./blender() [0x1795125]\n./blender() [0x17953db]\n./blender(BLI_task_pool_work_and_wait+0xaa) [0xc14cd6a]\n./blender(_ZN7blender3deg23deg_evaluate_on_refreshEPNS0_9DepsgraphE+0x1c2) [0x19128e2]\n./blender() [0x12ee344]\n./blender(wm_event_do_depsgraph+0xab) [0x179faab]\n./blender(wm_event_do_refresh_wm_and_depsgraph+0x87) [0x179fbe7]\n./blender(wm_event_do_notifiers+0x3f0) [0x17a71a0]\n./blender(WM_main+0x28) [0x179aa38]\n./blender(main+0x322) [0x10d47e2]\n/usr/lib/libc.so.6(+0x29290) [0x7ff8ed07a290]\n/usr/lib/libc.so.6(__libc_start_main+0x8a) [0x7ff8ed07a34a]\n./blender() [0x11d570c]\n\nPython backtrace\n```" ]

[ "Pinning Bones/Constraint Tabs in Properties Editor doesn't work\nBroken: 2.76 - master (1283c66)\nWorked: Unknown\n\nThe context pinning functionality in the Properties Editor doesn't work for Bones and Constraints. There may be a few others too that I haven't noticed yet, but these are the biggest ones.\n", "Linked group breaks Object Mapping in Texture Coordinates\nWindows 10\n\nBroken: 2.79b official release\n\nLinked group breaks Object Mapping in Texture Coordinates, when:\n- a texture (on a plane) uses another object (empty) as texture coordinates\n- all objects are in the same group\n\nFurther explanation in the blend file.\n\n1) Link the group \"object_mapping\" from the blend file\n2) move the group in any direction\nDespite the gradient texture on the plane using the empty's coordinates, this is completely ignored once linked as a group\n[object_mapping.blend](object_mapping.blend)", "Custom property of armature object not up-date visual value with pose bone transorm value\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 980 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 461.40\n\nBroken: version: 2.92.0\n\nArmature object custom property which driven by the armature bone transform value,not change value as visual untill select the property in properties objcet panel\n\n\nWhen I move mouse curser on the property area, it suddenly change value. but when I rotate bone along with X , it not up-date value.\n\n\nThough value seems up-date correctly (internal) , but when use bone driver to control many property,\nlike driven propA drive other propB, propB drive shape keys ,, etc,,current behavor make really difficult to check pose bone driver working or not. \n\n- Open attched file\n- In the 3D View Sidebar, change the bone rotation X value. (This value drive the custom property float value label as \"prop\". The driver expression only use one variable var = transform channel bone X rotation value).\n it not up-date visual value in object property panell >>custom properties prop value. , until select the prop value area. \nBut In driver editor view prop value actually up-date (change with bone X rotation value)\n\n[bonedriver.zip](bonedriver.zip)\n\n\n \n\n", "blender's UI repeatedly freezes when editing a material while the material settings are open.\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3090/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 471.11\n\nBroken: version: 2.93.2 Release Candidate\n\nWhen I load any .blend file containing at least 1 material that has image texture maps assigned to it, and I open the material settings, blender's UI will freeze for a while when I change any value or property related to said material, either in the shader editor or the material settings. \n\nThe whole process will repeat whenever you change any value/property in the materials settings or shader editor. The time it freezes depends on what formats and resolutions the textures are. In the below example with 16k textures, it should freeze for about 10 seconds on a fast computer, but if you replace the textures with 8 or 16k PNGs (which are available from the same site where the blend is hosted), the freeze time increases significantly: 120 seconds or more. \n\nThe only way to get it to stop repeatedly freezing is to close the material settings. \nIt will keep happening in render preview or in solid mode, it also happens regardless of whether the material preview in the material properties is enabled or disabled. \n\nThis also happens with 8k and 4k textures, but it's less noticeable.\n\n\n1. download and open clay_floor_001.blend\n2. select the sphere.\n3. open the material settings.\n4. change any property or value in the material properties or in the shader editor\n5. wait for a moment and blender's UI should freeze completely for 10-25 seconds. \n6. after it unfreezes, change any value/property in the materials properties or shader editor again, and the process should repeat.\n\ndownload the blend hosted here: clay_floor_001\nNote: you can also download png versions of the textures for this blend by clicking the little menu button to the top right of the site.\n", "Multi-Object (Weight) Painting\nMulti-object weight painting would be extremely useful when rigging characters with clothing pieces that are separate objects. If a character has a shirt and a jacket, they both have to be weight painted identically, but separately.\n\nThe objects could be combined, but that would mean they must share a modifier stack, which is not always an option. We could also paint one then transfer the weights, but that only works between very similar objects. So we are stuck with meticulously weight painting both clothing pieces, while making sure the weights match between them.\n\nWith multi-object weight paint, the same brush strokes could affect both meshes, so they would be a lot easier to keep in sync.\n\nThe bare minimum functionality that I would consider useful:\n- Be able to select an armature then several meshes that are rigged to it, and enter weight paint mode\n- Selecting a bone activates the relevant vertex group on all meshes\n- Brush strokes affect all meshes\n- Can switch to multi-object edit mode\n- Geometry masking features work\n\nIn my personal workflow, supporting multiple armatures is not important.\nIf anyone thinks this is important, please comment with a use case example.\n\nAdditional functionality that would be nice to have:\n- Indicating active object in the 3D View. So far this hasn't been necessary, since it was always just one object.\n- Changing the active object without leaving weight paint mode. Symmetry settings are stored on the mesh, so this would be needed to be able to access those.\n- All of the above for Vertex and Texture Paint :)\n\nI tried to keep wall of text to a minimum, I hope this is not too bad @ideasman42 :)", "Hair particles won't follow instaced position under 2 level nested collections.\nOperating system: Linux-5.11.0-7620-generic-x86_64-with-glibc2.31 64 Bits\nGraphics card: NVIDIA GeForce RTX 3080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 470.57.02\n\nBroken: version: 2.93.3\n\nHappens when you have an object with hair as instanced collection (A) inside another collection (B). Then, when you instance (B) the hair particle from (A) will not follow the relative position correctly.\n\n\n- Open .blend file\n- Select \"Cube\"\n- Object -> Quick Effects -> Quick Fur\n- {key Shift A} -> Collection Instance -> Collection\n- Move the instanced collection little away\n- {key M} -> New Collection -> Ok\n- {key Shift A} -> Collection Instance -> Collection 2\n- Move the instanced Collection 2 little away\n- You will notice the hair particle with a wrong position in relation to the original emitter Cube.\n\n1 min video here:\n[bugOrFeature.mp4](bugOrFeature.mp4)\n\nblend 2.93 here:\n[#91008.blend](T91008.blend)\n\n---\n\nNote that the concept works under 2.7 using Instaced Groups:\n\nblend 2.7 here:\n[ConceptWorks_b27.blend](ConceptWorks_b27.blend)\n", "Object properties In-front doesn't work in viewport shading\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1060 with Max-Q Design/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 445.87\n\nBroken: version: 2.82 (sub 7)\nWorked: Never\n\nObject properties In-front doesn't work in viewport shading\nAll the other options works in Object properties/viewport display, except for In-front \n\n\n- Open attached file\n- Go to Viewport Shading\n[lookdev_in_front_bug.blend](lookdev_in_front_bug.blend)\n", "History filled with \"Double click to rename\" when user changed index of a UIList\n\n\nBroken: version: 3.2.0\nWorked: Never\n\n- open Blender\n- add particle systems\n- change particle systems index\n- see how your history got vanished by useless flags\n\nGifs: \n\n\nI believe in order to fix this bug,\na `SKIP_UNDO` flag need to be implemented at a property level \nexample: \n`my_uilist_idx : bpy.props.IntProperty(options={'SKIP_UNDO'},)`\n\nThis issue is not only linked with UILists indexes, but also with any properties used strictly GUI\n\nsee the blender dev thread where we talk about the general issue of using properties for interface\n4\n\n\n\n", "Linked Mesh Editing Doesn't Work As Expected When Using Data Transfer Modifier\nOperating system: Windows-10-10.0.17763-SP0 64 Bits\nGraphics card: GeForce GTX 970/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 442.19\n\nBroken: version: 2.82 (sub 7)\nWorked: 2.81 (sub 16)\n\nObject instances doesn't work as in blender 2.81\n\n\nIn blender 2.81, after entering the edit mode, all instances become active. You can see changes in real time.\nScreen below:\n\n\nIn Blender 2.82 it work ok till some time. I couldn't determine at what moment it breaks...\n\n\nChanges appears only after You enter to object mode.\n\n\n\nHere is blend file to test\n[instances_issue.blend](instances_issue.blend)\n\n\n", "Clearing a CollectionProperty of ID pointers will dereference duplicate IDs below zero\n## Simplified Report\n\n- Execute the first script below in a new scene with a Cube.\n- The first script will add the cube 10 times to the collection property.\n- Save the file and reopen.\n- Execute the second script.\n- The second script will clear the collection.\n- The cube will get dereferenced to a user count below zero. With this assert:\n - `ERROR (bke.lib_id): source/blender/blenkernel/intern/lib_id.c:322 id_us_min: ID user decrement error: OBCube (from '[Main]'): 0 <= 0`\n- Saving the file again and reopening will remove the cube since it has no users. Even though it was linked to the scene.\n\n```lang=python\nimport bpy\n\nclass TestPropertyGroup(bpy.types.PropertyGroup):\n```\n object: bpy.props.PointerProperty(type=bpy.types.Object)\n```\n\nbpy.utils.register_class(TestPropertyGroup)\nbpy.types.ViewLayer.test_list = bpy.props.CollectionProperty(type = TestPropertyGroup)\n\nfor i in range(0):\n```\n item = bpy.context.window.view_layer.test_list.add()\n item.object = bpy.data.objects[\"Cube\"]\n```\n```\n\n```lang=python\nimport bpy\n\nclass TestPropertyGroup(bpy.types.PropertyGroup):\n```\n object: bpy.props.PointerProperty(type=bpy.types.Object)\n```\n\nbpy.utils.register_class(TestPropertyGroup)\nbpy.types.ViewLayer.test_list = bpy.props.CollectionProperty(type = TestPropertyGroup)\n\nbpy.context.view_layer.test_list.clear()\n```\n\n## Original Report\n\n\nOperating system: Linux-5.15.0-46-generic-x86_64-with-glibc2.31 64 Bits\nGraphics card: NVIDIA GeForce RTX 2080 with Max-Q Design/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 470.141.03\n\nBroken: version: 3.3.0\nWorked: version 3.1.2\n\n`ERROR (bke.lib_id): source/blender/blenkernel/intern/lib_id.c:322 id_us_min: ID user decrement error: OBCube (from '[Main]'): 0 <= 0` occurs when calling `.clear()` on a CollectionProperty attached to a ViewLayer. In the attached example, the CollectionProperty is a PropertyGroup containing a PointerProperty to bpy.types.Object.\n\n\n1. register properties and operators with provided script\n2. run `bpy.ops.view_layer.test_add_objects_operator()` to add selected objects to the view layer property\n3. save and close file\n4. open file and register properties and operators\n5. bpy.context.view_layer.test_attribute_list.clear() to get error\n\nstart with provided 'untitled.blend' to just run steps 4 and 5\n{F13493946}\n", "Cannot to make single user in object properties (if object is used by multiple collections)\nOperating system:\nGraphics card:\n\nBroken:\n\n\nI made a linked copy of an object (not object data!) in another collection by dragging the object from one window of the outliner to another.\nNow in object properties showing amount users equals 2:\n\n\n\nif you click on button \"make a single copy\" then nothing will happen. I am not sure but seems this UI inconsistency\n\n\nOpen the file and try to make single user to click in object property window:\n\n[make-single-copy.blend](make-single-copy.blend)", "Fade Geometry state changes selectability\nOperating system: Windows-10-10.0.18362 64 Bits\nGraphics card: GeForce GTX 1060 3GB/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 431.86\n\nBroken: version: 2.83 (sub 10)\n\nWith Fade Geometry enabled, other objects can not be selected while in pose mode.\nWith Fade Geometry disabled, other objects can be selected while in pose mode.\n\n[#70907.blend](T70907.blend)\n\n- Open file\n- Try clicking on bone and cube - selected item and interaction mode should be changed\n- With the bone selected and in pose mode: in Viewport Overlays, check Fade Geometry\n- Repeat step 2 - once in pose mode, cube can not be selected\n\n\n", "IDProps mismatch (pyRNA-defined vs. custom props) can confuse liboverride system in Collection property case\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 398.36\n\nBroken: version: 2.93.5\n\nOverride libraries linking seem to have an issue when linking collection property registered by an addon or a script.\nIf I add items to a collection property of an object in the source file, and then link the object using override libraries in a new file. I can't delete this item in the collection property, that's ok. **The problem is when I go back to the source file and clear the collection property,** so it doesn't have items anymore, then going back again to the linked file,**the object still has this property item,** even if it was deleted in the source file, and it still can't be removed.\n\nThe rar file includes a source file, a file with the linked collection, and a test addon which just register a property, you can also just run the script from the source file, but better to test with the addon.\n- Install the addon or run the script in order to register a collection property called test_proprety on the box object.\n- Open the file override_properties_to_link.blend\n- Select cube and Write in the console C.object.test_property.items() or just press the arrow up in the console\n- There should be one item assigned - bpy.data.objects['Cube'].test_property[0])\n- Try to clear using C.object.test_property.clear() and then check the items again. The item is still there for me and can't be removed, also when using remove(0).\n- Go to the source file and check for the items again using C.object.test_property.items(). It should be empty.\n\nHow to repeat the error?\n- In the source file add items to the object property using C.object.test_property.add()\n- Start a new clean file, link the collection from the source file.\n- Make the cube object a library override.\n- The object should have the item in the C.object.test_property from the source file. Save the file.\n- Go back to the source file and clear the items using C.object.test_property.clear() and save the source again.\n- Open the saved file with the linked source. Run C.object.test_property.items() in the console, and it will still have the item that was cleared from the source.\n\n\n\n[Override_property_Bug.rar](Override_property_Bug.rar)", "Stronger binding of Active and Selected\n**Context**\n\nFor many things that can be *Selected*, there is also a state called *Active*. There can be 0..∞ *Selected* things, and 0..1 *Active* things. This applies to objects, bones, fcurves, fcurve keys, mesh elements, and more. These will just be referred to as *thing* in this task to avoid having to repeat that entire list.\n\nThe *Selected* and *Active* states are independent, so there are four states a thing can be in:\n\n| | Not Selected | Selected |\n| -- | -- | -- |\n| Not Active | x | x |\n| Active | x | x |\n\n**Problem**\n\nThere are two states possible that can cause issues:\n1. **There are *Selected* things, but none of them is *Active*:** a sidebar/numerical/properties panel will not show information, and generally be empty without explanation why.\n2. **There is an *Active* thing, but it is not *Selected*:** this is not shown in the UI, as there is no separate color for this situation; all *Active* things are shown the same way regardless of them being *Selected* or not. This causes confusion and unexpected behaviour, and also doesn't allow for an actual clear use of this possibility.\n\n**Related Reports**\n\nThese reports are examples of confusing situations where the disconnect between *Active* and *Selected*. The list is not exhaustive, and may be expanded later.\n\n- #14429 (selected bone is not active)\n- #65219 (Objects don't become active object if they are selected by a box)\n- #68647 (objects selected using box select cannot be moved to collection if there is no active object)\n- #81922 (The Graph Editor doesn't show any curve of the Armature (if in Pose mode, but armature unselected, with box/circle/lasso tools))\n\n**Proposal**\n\nThis design task proposes linking *Active* and *Selected* such that:\n1. If there are *Selected* things, there is always one of those that is the *Active* thing.\n2. If a thing is *Active*, it is also always *Selected*.\n\n**Hurdles to Overcome**\n\nRule 1. above requires that one of the selected things is marked active. There is not always a unique, obvious choice here, for example when box-selecting multiple things at the same time. In this case the choice should at least be consistent so that users know what to expect after a while. The thing closest to the mouse cursor (when using the mouse to select) could be a reasonable choice.\n\nRule 2. above can be dealt with in various ways:\n- Create a function that returns the *Active* thing, and make it return nil when that thing is not also *Selected*. This is what was implemented for the Active Keyframe (983ad4210b). This has the downside that now nothing is active, breaking rule 1 and forcing the selection-changing code to pick another thing to become *Active*.\n- Refuse deselection of the *Active* thing.\n- ... something more elegant.\n\n**Goal of this Design Task**\n\nThe goal of this design task is to answer the following questions:\n- Is there an agreement that the current situation causes issues, and that having a stronger correlation between *Active* and *Selected* is a good thing?\n- Is there an agreement that the proposed two rules are the way to go?\n- How do we implement these rules in a way that makes sense to artists?\n- Can we implement this in a way that can be reused by all the affected parts? Some unified selection/activation API that can be implemented for different data types?\n", "Asset Browser - Current File doesn't place objects in active collection\nOperating system: Windows 11\nGraphics card: RTX 4070ti\n\nBroken: 3.5.0\nWorked: Supposedly never (tested 3.0-3.6)\n\nWhen adding an object into the scene from the Current File category in the Asset Browser, objects are placed into the same collection as the assets they originate from instead of the collection you are currently working in.\n\n- Move default cube into a new collection\n- Mark as asset\n- Ensure the scene collection is active, not the cube collection\n- Drag cube asset from Asset Browser Current File category into the scene\n- It will appear in the same collection as the asset reference object\n" ]

[ "Library override fails to map drivers in materials to overridden objects, once the linking file is reloaded\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 466.27\n\nBroken: version: 2.92.0\n\nWhen adding a driver to a material that points to an armatures bone, it works as expected. But when the object and armature are linked and then added as a library override, then the material drivers point to the overridden object what is correct. But when the file is reloaded, blender fails to reapply the mapping from the linked to the overridden objects. This makes it impossible to control materials with armatures if the new lib overrides system is used.\n\n\n 1. create an armature and a material on any object with a driver that points to the armatures bone\n 2. Save the blend \n 3. Create a new blend, link the previous blend and create a library override of the linked collection\n 4. Save the second blend and reload\n 5. try to modify the driven property (eg y loc)\n\nDemonstration: \n[2021-05-13_19-33-30.mp4](2021-05-13_19-33-30.mp4)\n\nFiles:\n[linking issue.zip](linking_issue.zip)", "Library Overrides: Shader drivers stop working after save and reload\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.71\n\nBroken: version: 2.92.0 Alpha\n\nShader drivers seem to stop working after a file with a library override is saved and reloaded. This can be fixed by running Resync Library Override Hierarchy on the collection every time the file is opened.\n\n- File: [effect_plane_override_bug.blend](effect_plane_override_bug.blend)\n- Link the CH-plane collection to a fresh file and Make Library Override.\n- Enter Material Preview in the 3D viewport.\n- On the armature's only bone, find the \"transparency\" custom property and change it. It works.\n- Save your file and reload it. The transparency custom property no longer works.\n- Resync the collection and it works again.\n\nCC @mont29 \n\n**TEMPORARY WORKAROUND:**\n\nFor now it is possible to work around this by relaying the information through an intermediate custom property on the material:\n\n* Add a custom property to the *material* datablock and move the currently existing driver to it.\n* Use a driver that copies the value of the newly added material custom property into the node tree.\n\nFor some more details and an example file see comments below.\n", "Driver vanishing with Ctrl + Z\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 1080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 461.40\n\nBroken: version: 2.93.0 Alpha\n\n[Driver is not working anymore when pressing ctrl+Z]\n\n[Hi, I put a custom propertie on an object to driver a value in his material. The Object is linked in another scene and I made it libray overidden. The driver is not working, so I copy the value of the custom prperties as new driver, then I paste it in the the value field in my shader. It works again. I continue my animation, but every time I pressed Ctrl +Z, the driver is not working anymore.]\n[2021-02-18_10-28-46.mp4](2021-02-18_10-28-46.mp4)\n\n" ]

[ "Some particles do not appear in the render when the 'Vector' pass is enabled\nOperating system: Linux-5.19.17-2-MANJARO-x86_64-with-glibc2.36 64 Bits\nGraphics card: NVIDIA GeForce RTX 2060 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 470.141.03\n\nBroken: 3.2.2 - 3.3.1\nWorked: 2.79\n\nParticles in Render don't match particles in viewport. Some particles do not appear. When they move very fast the issue is more visible.\nIt happens when we enable the pass `Vector`\n\n- Open attached file or:\n - In the default scene (with the Cube), add a Plane\n - Add a ParticleSettings toe the plane\n - In {nav Particle Properties > Render} set `Render As` to `Object`\n - For the `Instance Object` choose the Cube\n - In {nav Particle Properties > Field Weights} set `Gravity` to `0.0`\n - Optionally bake for cache\n - Set `Render Engine` to `Cycles`\n - In {nav View Layer Properties > Passes > Data} enable `Vector`\n - Move Cube out of camera view\n- Choose a frame in the middle and render\nNote that the particles closest to the plane disappear\n\n[particles displaced.blend](particles_displaced.blend)\n\n", "Cannot set the Color attribute to display an object through Geometry Nodes\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2070 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 535.98\n\nBroken: version: 3.5.1\nWorked: version: 3.4.1\n\nbehavior changed in 6514bb05ea\n\nUsing the \"Store Named Attribute\" node setting the Attribute \"Col\" in geometry nodes, no longer sets the Vertexcolor of an object procedurally. When the Curve is converted to mesh and then the created \"Col\" Vertexcolor attribute is clicked, the colors show up.\n\nIn 3.4 and below this would override the vertexcolor attribute in \"realtime\" without converting the curve to mesh.\n\nOn a Curve Object, add a geometry node\n- create \"Store Named Attribute\"\n- set the \"Name\" to \"Col\"\n- change the color\n- Set the Viewport shading to color \"Attribute\" (seeattached Images) to view the result\n\nAttached file with example setup that used to overwrite the vertexcolor in 3.4.1, but no longer does in 3.5.1\n\n", "Adding Node Group inputs unhides newly added inputs on nodes with hidden node sockets\nBroken: version: 3.0\nWorked: never\n\nWhen using multiple group inputs in one GN node tree, and hiding unused node sockets to keep the tree visually clean, adding a new input socket into Group Input node will make this new socket appear in all existing Group Input nodes which are set to hide unused node sockets, so the user has to go through all of them and cycle the \"hide unused node socket\" toggle to properly hide the unused node sockets again.\n\n- Open .blend file\n- Hide unused sockets of any one Group Input node ({nav Nodes > Toggle Hidden node Sockets})\n# Add a new input socket from sidebar.\nResult: The newly added input has appeared on all instances of Group Input node, despite not being used on them and the nodes being set to \"hide unused node sockets\".\nExpected: Newly added input to a Group Input node will not appear on nodes with \"hide unused node sockets\" enabled, because it is unused on them.\n\nTest File:\n[#94413.blend](T94413.blend)\n[2021-12-27 14-29-03.mp4](2021-12-27_14-29-03.mp4)\n", "Geometry Nodes: 3D Viewport Overlay Statistics don't respect Viewer Node\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 471.11\n\nBroken: version: 3.4.0 Alpha\n\n\nThe geometry output by the preview node is not taken into account by the overlay scene statistics:\n[2022-09-30 11-35-43.mp4](2022-09-30_11-35-43.mp4)\n", "Make Instances Real doesn't work for geometry instances\nOperating system: Windows-10-10.0.18363-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 750/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 471.41\n\nBroken: version: 3.0.0\n\nWhen I want to make instances real on object that has geometry nodes, blender create duplicate of whole mesh, togather with instances, on each place (point) where single instance mesh should be. I have attached screenshots and test scene that reproduces the bug. \n\nOpen .blend file I sent you, hover with mouse over 4 instaced spheres, and pres ctrl+A... select make instances real. You will see that geometry nodes instances are not created properly (whole geometry is duplicated on each geometry node point).\n\nSimilar thing happens when instancing collections, and trying to make them real - the parent relationship/hiararchy is not working. So there is a bug there also.[geom_nodes_bug.zip](geom_nodes_bug.zip)\n\n\n\n\n", "Workbench: Cull before batch creation during viewport animation rendering\n\n## Current Situation\n\nWhen drawing an object (eg sending draw calls to the GPU) the object boundary sphere is tested against the active view). When the test fails the object isn't send to the GPU. Before this happens the GPUBatches were already generated.\n\nWhen doing animation playback this could lead to the situation that the GPU Batches are calculated, but not used.\n\n## Target Situation\n\nWhen using the Overlay engine/Workbench engine we could test the objects to the default view (e.g viewport). When this fails we could skip this object at all. This feature would be disabled when the shadows are turned on and only enabled when doing an image/animation render.\n\nA consistent framerate is more important to animators then having some frames faster.\n", "Regression: Visual bug when using the object gizmos\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 471.68\n\nBroken: version: 3.0.0 Alpha\nWorked: 2.93\nCaused by 917a972b56\n\nDifferently from using the transform tools in the toolbar, when activating the object gizmos that stay visible also with the select tools, a strange bug occurs that the e.g. moving the object on an axis keeps a grey axis where it was before, the gizmo is always present (differently again from using the transform tools), and if you switch from this to the transform tools to the selection with gizmo again, it messes the origin of the object.\n[2021-09-20 14-59-52.mp4](2021-09-20_14-59-52.mp4)\n\n\nI'm pretty sure that this is caused by the recent change of keeping the viewport gizmo visible when doing these operations.\n\nUse the select tool in the toolbar, activate on or more object gizmos that stay always visible, do a transform operation.\n[Based on the default startup]\n\n", "Hidden objects still affect viewport performance when they are hidden as collection\nOperating system: Windows-10-10.0.19043-SP0 64 Bits\nGraphics card: Radeon RX 570 Series ATI Technologies Inc. 4.5.0 Core Profile Context 22.7.1.220725\n\nBroken: version: 3.4.0 Alpha\nWorked: n/a\n\nAfter ac20970bc208 hidden objects do not affect viewport performance but they still do when hiding the collection that they are belong to.\n\n(I've wondered that this might not be a bug but a complication from the fact an object can belong to multiple collections. If then I guess this can be marked as a known limitation or something, because I think many users would expect this to work with visibility of collections.)\n\n* Use file from the patch: [F13435936](slow_object.blend)\n* Hide \"Collection\" instead of the object itself\n* Play animation, the object still affect viewport performance\n", "EXCEPTION_ACCESS_VIOLATION error when trying to change object viewport color with the viewport compositing enabled in depsgraph_update_post\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 680/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 466.27\n\nBroken: version: 3.5.0\nWorked: 3.4\n\nThis error occurs only if the viewport compositor is enabled - rendered mode of the viewport and Compositor = Always.\nWhen trying to change the viewport color of the object by script on depsgraph update event, this error occurs.\n\nThe short script example:\n```\nimport bpy\n\ndef on_depsgraph_update_post(scene, depsgraph):\n if depsgraph.id_type_updated('SCENE'):\n for obj in depsgraph.updates:\n if isinstance(obj.id, bpy.types.Object):\n print('change object color')\n bpy.data.objects[obj.id.name].color = (1, 0, 0, 0)\n\nbpy.app.handlers.depsgraph_update_post.append(\n on_depsgraph_update_post\n)\n```\nAfter executing this script in the Text editor, switching to the \"Rendered\" mode of the 3D Viewport and enabling viewport compositing - when adding or copying the object - Blender throws an error.\n\nThis occurs only if the viewport compositing is enabled, if it is not used - code works as expected.\n\nError occurs on 3.5 and 3.6 alpha.\n\n1. I attached .blend file with the example\n2. Open .blend file\n3. Switch the viewport to the \"Rendered\" mode (z - Rendered)\n4. Execute script in the Text editor\n5. Try to copy cube (shift + d)\n6. Error throws\n\n", "Links persist invisibly after their source sockets are removed\nBroken: Probably forever (e.g. 3.0, 3.1)\nWorked: N/A\n\nIf the output socket of a node is removed, the input socket it was connected to does not show a button to input a value. It acts like it's still connected.\n\n**Detailed replication**\n1. Go into the Shader Editor\n2. Create the *Voronoi Texture* node and a *Math* node\n3. Set the *Voronoi Texture* node to 4D, and plug its new W output into the *Math* node\n4. Set the *Voronoi Texture* node to 3D. Observe that the input on the *Math* node still acts like it's connected\n5. Click and drag from that input on the *Math* node. A bogus wire will appear from the *Voronoi Texture* node, coming from where the W output used to be.\n\nYou can then plug this wire into another socket. It will disappear again once you let go. If you set the *Voronoi Texture* node back to 4D, the wires will re-appear.\n\nVideo example:\n\n[XwBMq60aMI.mp4](XwBMq60aMI.mp4)\n\n**Another example:**\n\n{[F13079601](free-socket.blend), size=full}\n\nIn this case you can even save the file in this state where you can't see or edit the value of the node until you disconnect it.\n\n", "Visual geometry to Mesh does not duplicate a linked object\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 970/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 471.96\n\nBroken: version: 3.2.0 Alpha\n\nWhen applied to a linked object, Visual geometry to Mesh does it on the shared mesh.\n\nApply the subdiv modifier to the original cube. Link copy it. Use Visual geometry to Mesh on the linked object. Now select the original object and go into edit mode. You should see that the the subdiv modifier has been applied.\n\nThere is also a side issue to this. If you put a Decimate modifier (something visible like planar 20 degrees) on the original object, selected the linked one and go into edit mode, you'll see that in the viewport the modifiers on the original object don't show their effect.", "Retopology: Active Object Geometry gets Occluded by Semi-Transparent Reference Object in Edit Mode\nOperating system: Linux Ubuntu 16.04\nGraphics card: Nvidia GeForce MX130\n\nBroken: 2.83.0\nWorked: 2.82.a\n\nActive mesh is occluded by high poly or another reference mesh in Edit mode. This is: I want to edit mesh A. I need mesh B to get a reference. So I make B to have a semi-transparent material for Viewport view. I select A and go to Edit mode to edit it. However, its edges and vertexes are occluded by B. We can see a comparison of 2.83.0 (first image) with 2.82a (second image): version: 2.83.0 does not show active object geometry so that we can not edit it. On the other hand 2.82a does show required geometry.\n\n \n\n1. Open Blender 2.83.0\n2. Create a Sphere mesh in the world origin so that it fits inside the default cube.\n3. Assign a semi-transparent viewport display material to the Cube so that we can see the Sphere inside. Change color also to have a better view.\n4. Select Sphere and go to Edit mode.\n5. Press A to select all vertexes.\n6. We can see that Edges and Vertexes are occluded by the Cube so that we can not edit the Sphere properly.\n\n\n[bug-report.blend](bug-report.blend)\n", "GPencil: A layer used as mask for 2 or more other different layers makes the GP object partially unselectable in the viewport\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 970/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 442.74\n\nBroken: version: 2.83.0 Beta\nWorked: /\n\nA layer (let's call it layer A) is masking more than one layer (B and C, for example; so B and C have in their \"mask(s) list\" layer A in it). Doing it makes the grease pencil object unselectable in the viewport unless you click on the shapes that have been drawn on layer A (or, at the precise point where the origin of the grease pencil is located, but only if the origin is displayed in the viewport).\n\nStrangely, saving with \"ctrl+S\" makes everything works as expected, but only until the file is closed. Opening it again and the problem is back.\n\nIn a concrete case scenario, it is a problem since if you use a layer as an invisible mask to hide strokes that go beyond certain limits, it makes it random to select the GP object (or forces to use the outliner).\n\n- Create a GP monkey\n- Create a new layer (layer A) that will mask the 2 already existing layers (\"Lines\" and \"Fills\")\n- In \"Lines\" and \"Fills\" layer, check the \"mask(s)\" option and add \"Layer A\" in each list\n- Back in \"Layer A\", draw a filled shape anywhere\nNow, in object mode you can only select the GP monkey by clicking on the filled form in layer A. You can invert the mask layer in the mask list of the layer named \"Fills\": It seems that it doesn't matter whether the masked parts are visible or not.\n\nBased on the attached .blend file:\n- Open file\n- Try to select the monkey by clicking on the part supposed to act as a mask (the \"selectable\" part in the screenshot)\n- Try to select the monkey by clicking in any other part of the monkey.\n\n{[F8563683](select_bug.png), size=full}\n\n[mask_selection_bug.blend](mask_selection_bug.blend)", "Viewport display of multiple Dupli-Collections when different draw-types are used : wireframe might disappear (only selected are garuanteed to draw)\nOperating system: Windows 10 / Ubuntu 20.04\nGraphics card: GTX 1080 / RTX 3080\n\nBroken: 2d9d08677e a2d59b2dac \nBroken: all versions >= 2.8\nWorked: 2.79\n\nViewport display of Dupli-Collections is inconsistant.\nDigging deeper into it, it looks to be when there are different draw types on the same dupli-collection\n\n1. Open attached blend file\n2. select the empties in the middle of the screen which do not look like they have anything attached\n3. notice how they randomly appear / disappear upon selection.\n4. note that one of instancer is set to `Textured` draw type, the other to `Wire`, if all are set to `Wire` the issue is gone\n\nThis has happened in quite a few of our production scenes.[bug.blend](bug.blend)\n\nNOTE: Might boil down to be the same as #101822 (Collection instance set to Display As Wire doesn't show in Solid mode when not selected) or #86732 (Instanced wire objects are hidden in solid mode instead wire appearance)\n\n", "Z-fight with invisible face in edit mode\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 1660 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 457.51\n\nBroken: version: 3.0.0 Alpha\nWorked: \n\nI have 2 cubes. Bigger one is set viewport display as wire. Both share the same XZ plane. In Object mode shading is fine, but if I enter in Edit mode with smaller cube artifacts are appear.\n\n[ladder_01.blend](ladder_01.blend) \n\n" ]

[ "Geometry Nodes does not show instances of collections in \"Render image\" ok, while both cycles and eevee render those correctly in Viewport\nOperating system: Ubuntu 20.04.4 LTS\nGraphics card: GeForce GTX 1650\n\n**Blender Version** 3.1.2\nBroken: 3.1.2 daily build April 15, 01:58:06, cc66d1020c3b as well as earlier versions\n\n\nInstances made with geometry nodes don't show collections in *Render image*, unless the collection itself is rendered in the image, too. Why this is serious bug and not a feature is explained below. Also note that both cycles and eevee do render the geometry nodes correctly in Viewport, although not in the final Render image.\n\n\n1. Start a new general blender file\n2. Add UV sphere, and move it slightly away from the cube, so that you now see both the cube and the sphere\n3. Make a new collection, and move the cube into that collection\n4. Go to Geometry Nodes workspace, select the sphere, and click \"+ New\". You now have \"Group Input\" ----> \"Group Output\" for the sphere shown in the Geometry Nodes editor\n\n5. Drag the new collection containing the cube into the Nodes editor. It is shown as \"Collection Info\" box within the editor.\n6. Connect the Geometry output from the \"Collection Info\" to the geometry input of the \"Group Output\". The sphere changes into a cube, and so you see two cubes.\n7. Hide the collection containing the cube both from Viewport and from Renders by clicking the respective eye and camera icons on the right side of the collection name\n\nNow you see just one cube in Viewport. It is where the sphere was located. **This perfect as it should be**. And it doesn't matter which kind of visual or which workspace you have in Viewport. All of them show one cube, **from the wireframes and solid modes to eevee and cycles rendered scenes - as long as eevee and cycles are rendered in Viewport**.\n\n8. Hit F12 or select Render Image from the Render menu. You don't see one cube rendered anymore, there is nothing in the image. This is both inconsistent with the same cycles (and eevee) rendering in Viewport, and more importantly, plainly wrong and highly problematic, as exemplified below. Correct behavior would be showing one cube, just like both eevee and cycles show when those are rendering for Viewport.\n\n(9. *Note that if you unhide the collection in the Renders by clicking the crossed camera icon, and then F12 to render, you shall see two cubes. That rendered image is right and ok as such,* but this behavior shows a current requirement that the collections themselves must be clicked unhidden in Renders, so that one could use instances of a collection in the Geometry nodes. This is a serious bug that is hard to circumvent.) \n\n**No - that is not a correct feature but a bug. A practical example illustrates:**\n\nMake some complicated object, for example a small spaceship, and put all of its components into a collection. So far so good.\n\nYou wish to make an animation of a fleet of 100 of similar spaceships engaged in some situation and flying in the scene. \n\nA beautiful method to achieve is this by geometry nodes. The original spaceship collection would be hidden from Viewport and from Renders, since you are not using it and not moving it around. You are using it just as model for you fleet's ships.\n\nYou make your fleet, and make it move by some logic, and everything looks great in Viewport even when rendered with cycles. But when you begin to render your final images, no spaceship will be seen, because your collection was hidden.\n\nOk, perhaps you just unhide the collection and use it as one your ships? No good, since when you rotate or move your collection, all of the geometry node instances (your 99 other space ships) move and rotate similarly as well. The collection has to remain unmoving and unrotating, at origin. And that spaceship will look pretty stupid, just standing still while all the other ships move around. (Note that even if some future versions of blender would make it possible that the collection itself and the geometry node instances could move independently from each other - so that e.g. rotating the collection would not rotate the instances - it would still be wrong to require that the original collection is visible in the scene to have those geometry node instances in the final Render image. For the reasons that has been said this far, and also because then one would need to make two separate modules for moving and rotating the spaceships in the scene. One for moving the collection and its single spaceship, and one for moving all the 99 other spaceships that are generated by geometry nodes! Right way is making all of the 100 spaceships with the geometry nodes, and making just one module/logic that move all those spaceships in a way that looks good and consistent.)\n\nAlso, you cannot hide the unmoving collection spaceship by scaling it to zero size, because all of the other spaceships would scale similarly with the collection.\n\nMaybe put the camera so that the collection is behind it? Then it wouldn't matter, even if the collection is unhidden and just standing still at the origin...? Well, this would be really clunky workaround, and even that wouldn't work well in a general case, since you might need to turn the camera to different directions, if you are for example following some particular spaceship.. So the (unmoving) collection could become visible in such cases.\n\nA functioning hack would be positioning our camera, scene and everything that is interesting say 10,000 units away from the origin, and the collection would be at origin. This way the collection would be too small and too far to be seen in Render image, even if it is unhidden. Yet realistically, this is not the right way to do it...\n\n\n\nThere has been a very short bug report that relates to part of this, but it didn't make the real problems very understandable. It was T93429\n", "Source Collection visbility will affect the final render of the links object(geo nodes) in eevee/cycles\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2080 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 512.15\n\nBroken: version: 3.2.0 Alpha\n\nWhen collection is used in GN tree and it is disabled for final render (camera icon in outliner). Then the new object is not visible in final render\nWhich is not the case when object is used instead of collection in GN tree\n\n1. create a simple object\n2. move it to a new collection\n3. turn off the collection's visibility\n4. new object with geo mod, use the collection as output\n5. render, the new object didn't show in the final render(but viewport)\n\nor \n- Open test file\n- Do final render i.e. F12 (object is not visible in final render)\n- Now connect `object info` node to group output and render (object is visible)\nTest File\n[#97535.blend](T97535.blend)\n[2022-04-22 12-55-10.mp4](2022-04-22_12-55-10.mp4)" ]

[ "Set parent (keep transform without inverse) has no effect on scale\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: GeForce GTX 750 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.71\n\nBroken: version: 3.2.0 Alpha\n\n\nSet parent (keep transform without inverse) has no effect on scale. End result is different if use Set parent (without inverse)\n\n[parent_scale.mp4](parent_scale.mp4)\n", "World Cavity Not Scale-Agnostic\nOperating system: Linux\nGraphics card: Intel\n\nBroken: 2.92, 2.7, 2.8\nWorked: Never, I think.\n\n\nBlender's viewport AO behaves differently depending on scene scale, to the effect of significantly limiting its usability at larger or smaller scales than 1BU.\n\nA scene that has been scaled up does not have the same AO patterns as the original scale, even if you also increase the \"Distance\" and/or \"Valley\" factors proportionally.\n\nThe noise behaviour is also significantly degraded at larger scales.\n\nDefault scene scale and world cavity settings:\n\n\n\nScene scaled up 100X, cavity still defaults:\n\n\n\nScene scaled up 100X, cavity distance also increased 100X:\n\n\n\nScene scaled up 100X, cavity distance and strength also increased 100X. Notice the noise:\n\n\n\nThis might seem like a feature request for a new AO algorithm, but I think it crosses over into being a bug report because the current behaviour means that the usefulness of world cavity rendering is significantly degraded for any feature sizes that are much larger or much smaller than 1BU. I.E. It's not that a totally new algorithm is necessarily required; the issue is that the current system is effectively broken outside a very narrow range of settings. Outside that range of settings, its best performance is either too weak or too noisy to be useful.\n\n\nSet up two geometrically similar scenes at different scales, then try to find world cavity settings that produce equivalent results for them. Set \"Lighting\" to \"Flat\" to make it easier to see.\n\nOr: Set up a scene where the average feature size is much larger than 1BU, and then try to find useful world cavity settings for it.\n\nOr: Open the below file that has two differently sized Suzannes in two viewports, and play around with world cavity settings.\n\n[WorldCavityFailure.blend](WorldCavityFailure.blend)", "Decals crash Eevee on AMD graphics card\nOperating system: Linux 6.2.7 Manjaro \nGraphics card: AMD 7900XT\n\nBroken: 3.3LTS, 3.4.1\n\nUsing decals will crash blender in Eevee. cycles works\n\nas an example file please see the .blend that was linked here by the decalmachine creator: 66438\n\nThe issue there is unrelated but the file can serve to show the issue I am having. The moment I switch to rendered view or materials preview blender will crash. I see no error message in console. The log files states just: \n\n```\nbpy.context.space_data.shading.type = 'RENDERED' # Property\nbpy.context.space_data.context = 'RENDER' # Property\nbpy.context.scene.render.engine = 'BLENDER_EEVEE' # Property\n \n```\n\n", "UV: Opening an image in UV Editor changes zoom\n1) Open default cube\n2) Tab to enter edit mode\n3) Open UV Editor\n4) Observe scale of unit square\n5) Open an image (Click \"open\" and choose an image file)\nBug, unit square is no longer in same location.\n\nThanks @BlenderBob for the initial report!\n\n", "Tool shelf circle select. Changing the brush radius by hotkey does not update the radius\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 1060 6GB/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.43\n\nBroken: version: 2.92.0 Alpha\n\nWith the legacy circle select tool you can hold down the mouse, and with scroll wheel or numpad + / - buttons you can change the radius of the brush.\n\nWith the circle select tool from the tool shelf the hotkeys still performs a radius change, a visible radius changes its size, but the radius does not update. It remains at the radius it was.\n\n\n\nOpen Blender, press C, hold down lmb, scroll the scroll wheel.\n\nExit circle select, activate circle select in the tool shelf. Hold down lmb, scroll the scroll wheel. Watch the result.\n\n[2020-12-04 12-36-43.mp4](2020-12-04_12-36-43.mp4)\n\n\n", "Cycles: Pointiness has artefacts on extremely small mesh size\nOperating system:\nGraphics card:\n\nBroken: version: 3.6.0 Alpha\nWorked: Never\n\nPointiness has artefacts on extremely small mesh size. \n\n| Normal size | Small size | Smaller size |\n| -- | -- | -- |\n|  |  |  |\n\n1. Create cube with beveling edges. Create small copies on vertices (as on example pictures).\n2. Set shader to cube. Shader with:\n a. Inputs: Geometry -> Pointiness.\n b. Map range: From Min: 0.5, From Max: 0.54.\n c. Material Output.\n3. Render that cube with applied different sizes.", "Improve visualisation and manipulation of lights in the viewport\nThe idea is to better control/visualize light parameters in the viewport. Currently we can only control the spot angle of spot lights, the size of area lights and the light direction. \nConcept drawings by @eyecandy for reference, not the final design.\n**Tasks**\n- [ ] visualize light color (translucent filled? opaque outline? center circle?)\n\n- [ ] register gizmo undo\n- [x] make scaling transform follow the cursor exactly\n\nAdd gizmos to\n- [ ] spot light\n - [x] radius\n \n - [x] make scalable even when radius is zero?\n - [x] blend \n - [ ] spot angle via new gizmo type instead of arrow \n- [ ] area light\n - [x] constraint scaling in X/Y directions \n - [ ] beam spread (only supported in Cycles) \n - [ ] clip end/custom distance (only in Eevee) \n- [ ] sun lamp\n - [ ] angular diameter \n- [x] point light\n - [x] size ", "Cycles: Baking to an externally loaded image fails to update the image buffer results in Cycle's Rendered Viewport Shading\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2080 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 512.59\n\nBroken: version: 3.3.0, 2.93\nWorked: probably never\n\nBaking to an externally loaded image will not changed the baked image as displayed in the Viewport Shading's Rendered Mode.\nThe newly baked image buffer is shown correctly in the image editor and (EEVEE's) Material Preview, but wrong in Cycles's render view.\n\n- Open the attached .blend file.\n- Open the attached baked image in the image texture node.\n- Use it to bake any type of pass to the loaded image.\n- Observe that the image looks unchanged in the Rendered View by connecting the image texture node to the material output.\n[#98063.blend](T98063.blend)\n\n[2022-05-12 09-12-20.mp4](2022-05-12_09-12-20.mp4)", "Volumetrics: Add Light Grid support\nVolumetrics are currently not affected by lightgrids. FIX THAT.", "Radius in place of Size at Empty viewport menu\nI{[F6596506](RIGHT.JPG)}\n\n am new to blender and not know where to post this but plz check it.It seem wrong to me\nthanks", "Affect origins indicator gizmo is not affected by unit scale\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 1080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 451.48\n\nBroken: version: 2.83.2\nWorked: never\n\nThe gizmo used to indicate affect origin feature of transform tool is enabled does not change with different unit scale. This makes the gizmo invisibly tiny when using unit scale of 0.01, which is mandatory for Unreal Engine export capability, as described here: scene_scale.mp4\n\n1. In a new .blend file, set unit scale to 0.01\n2. Create a new cube. Newly created cube should be sized appropriately for new unit scale.\n3. Select new cube, Activate move tool, and activate \"Origins\" Affect Only option.\nResult: The gizmo to indicate Affect Only Origins feature is enabled is invisibly tiny and requires extreme zoom in:\n\nExpected: The Affect Only Origins indicator gizmo scale changes appropriately with different unit scales.\n", "EEVEE/Cycles discrepancy in how image texture node's alpha is managed.\nOperating system: Windows-10-10.0.19043-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 511.79\n\nBroken: version: 3.2.0 Alpha\nWorked: ?\n\nWorkbench and EEVEE perform an unpremultipy calculation when premultiplied alpha mode is selected and the image texture node's alpha socket has a connection.\nWorkbench and EEVEE ignore the image alpha when straight alpha mode is selected and the image texture node's alpha socket has a connection.\nSee: [node_shader_tex_image cc](node_shader_tex_image.cc$125-153)\n\nIn Cycles, there is no apparent difference between the straight and premultiplied alpha modes and the alpha socket's connection status does not change how the alpha is managed.\nSee:\n[shader_nodes cpp](shader_nodes.cpp$200-250)\n[image cpp](image.cpp$471-566)\n[image_oiio cpp](image_oiio.cpp$154-204)\n\nExample:\n\nExample 2:\n\n\n**Test File**\n[Alpha_Image_Test_3.1.0.blend](Alpha_Image_Test_3.1.0.blend)\n", "Paint Mode: Brush Radius VS Brush Size\nThis is a brief task design to explain **why Blender should switch brush radius to brush size** and how this change will benefit artists.\n\nI will start introducing the **issues that texture paint mode have currently by using brush radius:**\n\n**1. Makes impossible to paint single pixels.**\n\n- As minimum brush radius is 1px (integer values), **minimum brush size is 2px**, so it's **impossible** to use current brushes **to paint single pixels as for pixel art** styles.\n\n\n**2. Spectrum of brush size values cut by half.**\n\n- Current brush radius makes it **impossible** for artist **to paint with an odd number brush size**/diameter.\n- This likely **cut the values by half** as artists can't paint with a brush with sizes of 1, 3, 5, 7, 9, 11, and so on...\nThat's a silly limitation for artists when we have an easy solution of moving to brush size/diameter.\n\n\n**3. Contra-intuitive.**\n\n- **Brush size**/diameter **gives artists better feedback on the real size of the brush** over the texture space while they don't have to think in brush radius * 2.\n\n\nBrush size is a standard over brush radius but not because it is driven by big companies but because it is the most intuitive for artists while radius may be more intuitive for mathematicians and programmers, so I think we should invest into improving UX for the end-user, specially in texture paint mode that lacks a lot of it.", "VSE: Transform settings doesn't get copied across scenes\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce RTX 2060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 457.30\n\nBroken: version: 3.0.0 Alpha\n\nTransform settings doesn't get copied across scenes\n\nCopy a strip, which doesn't match the render resolution, across scenes:\n{[F10121941](copy_missing_transform.gif),size=full}\n\n", "With enabled Modifier On Cage option crashes on entering Edit Mode if in Render Viewport Shading\nOperating system: Windows-10-10.0.19043-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3070 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 522.30\n\nBroken: version: 3.3.0\n\nIn Eevee, in Render Viewport Shading, an object having a Geometry Nodes modifier with enabled On Cage Modifier Result, that affects a UV Attribute which is used in Shader, crashes upon entering Edit Mode.\n\n - Set Viewport Shading Mode to Render\n - Enter Edit Mode ->Crash\n[displace-preview.blend](displace-preview.blend)" ]

[ "Scaled Spotlight Does Not Work In Cycles\nOperating system: Windows 10 64 bit\nGraphics card: Nvidia GTX 750 Ti\n\nBroken: 2.82a, 2.90 alpha eaf7d36d66e5\n\nWhen a Spotlight is scaled, the intended behavior is that its shape changes.\nThis works in EEVEE, but not in Cycles.\n\nIn Cycles, it stays circular.\n\n[spotlight_ellipse.blend](spotlight_ellipse.blend)\nOpen the file.\n\nCompare renders by EEVEE and Cycles.\n" ]

[ "Update animation operators for previously hidden redo panel contents\nWith the changes in [D14960](D14960), many operators that had previously hidden options are showing their redo panel.\nThis applies to the timeline editor, dopesheet (all modes) and NLA editor.\n\nI tested all operators in each editor to find out which ones need changes. The following list is complete, meaning that any operator that is not listed here works as expected when tweaked:\n\n# Dopesheet\n\n- - [ ] Transform\n - Hide W axis\n - Not sure how to do this for `transform.transform` with `TIME_TRANSLATE`\n- - [x] Box Select\n - Hide redo panel (it never shows in 3d view)\n- - [x] `select_column`\n - Hide mode selector (the modes are very distinct from each other, and don't make sense to be changed by the user)\n- - [ ] `leftright`\n - Hide the \"Check if Select Left or Right\" mode\n - Does not seem like you can hide enum entries, so this may have to be split into a separate property\n\n# Markers\n\n- - [x] Holding select (rightclick)\n - Hide redo panel\n- - [x] Move\n - Hide tweak checkbox\n- - [x] Select before/after\n - Capitalize parameter names\n- - [x] Box Select (see Dopesheet)\n\n# NLA Editor\n\n- - [x] Duplicate\n - [D15086](D15086) fixed the issues related to the duplicate operator\n- - [x] Push Down Action\n - Hide redo panel (seems to be an internal property not worth changing)\n- - [ ] Transform (see Dopesheet)\n - Additionally, the Proportional Editing property can not be toggled (probably since the NLA editor does not have this option at all), so the checkbox should be hidden\n- - [x] Box Select (see Dopesheet)\n- - [ ] `leftright` (see Dopesheet)\n\n# Graph Editor\n\nWhile not affected by the above patch, it's still worth touching up, since the original patch that added its redo panel [D6286](D6286) did not change related operators either.\n\n- - [x] Box Select\n - While we should hide redo panel for consistency, the \"Select Curves\" option may be interesting to toggle? But at the same time, it currently does not update, so needs to be fixed.\n- - [x] `select_column` (see Dopesheet)\n- - [ ] `leftright` (see Dopesheet)\n\n---\n\nA lot of operators would benefit from functional changes, now that the user can adjust their parameters not only through different menu items, but also by tweaking the settings in the redo panel.\nThis could be tracked as an additional to do task.", "Geometry Nodes fouling up motion blur\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA RTX A5000/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 526.86\n\nBroken: version: 3.1.0\nWorked: N/A\n\nGeometry Nodes applied to a spline which affect its shape over time do not blur when Motion Blur is enabled.\n\nLoad attached simple file. Render frame zero. There is no blur at all. Render frame 1, *everything* is blurred, despite only part of the spline is in motion. It would appear the entire spline is blurred according to how the base object is moving, not how the spline is deformed as a result of the applied Geometry Nodes.\n\nNote that same file loaded into 3.3.1 causes Blender to crash (GPU rendering only - it's a stack corruption bug in the nVidia driver, but 3.1 doesn't cause a problem). I previously reported that crash but it was rejected due to possibly being related to the Photographer add-on, which I have uninstalled. It still crashes.[Motion Blur Bug.blend](Motion_Blur_Bug.blend)", "Regression: Crashes when uses preview or final render modes\nOperating system: Mac Os Monterey 12.6.1\nGraphics card: AMD Radeon R9 M380 2 GB\n\nBroken: blender 3.6.2\nWorked: Blender 3.4.1\n\nBlender 3.6.2 crashes on MacOS when you change between preview or final render modes. I Think is a problem related with the Realtime Viewport compositing. I think it is not enabled at all on MacOS but maybe developers did not do it correctly and some render features are requesting missing elements.\n\nSimply make 3 or 4 render previews and then press F12 to obtain a final render (or viceversa). Blender Crashes immediately.\n\n", "Geometry nodes preview, clipping Region in the viewport its not working\nOperating system: Linux-5.15.0-58-generic-x86_64-with-glibc2.35 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 470.161.03\n\nBroken: version: 3.4.0 Alpha\nWorked: Never\n\nCaused by: c55d38f00b\n\nTo be compleatly honest i don't remember if this was a limitation of the preview node or not and i was unable to find the answer. \nSo as the title says viewport clip in viewport doesn't work in the geonodes overlay preveiw\n\n\n- Try to preview any data from geometry nodes. \n- alt+b to do a viewport clip\nshould look like in the left side of the image above, but it seems that a second geometry with the info from geo nodes in alpha blend mode its not being clipped \n\n", "Mode locked and crash if objects become disabled in viewport while not in object mode\nOperating system: Linux-5.15.0-2-amd64-x86_64-with-glibc2.33 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 470.86\n\nBroken: version: 3.1.0 Alpha\nWorked: didn't crash but showed an error message in 2.91.2\n\nObjects that are in a mode which is not Edit Mode nor Object Mode, will lock the mode setting (so the mode options are greyed out and unusable) if they get disabled in the viewport. Hiding will not cause the this bug, it needs to be disabled in viewport using that display icon in the outliner, not the eye.\n\nThe solution for this is probably to make the disable in viewport action switch the mode to object mode if the active objects gets removed (much like delete from outliner). Another solution would be to still allow the mode switch.\n\nMuch worse though! Blender crashes sometimes when it gets into this state. I can reliably crash blender as described below.\n\nGo into weightpaint mode of the default cube and disable it in the viewport using the outliner (you need to enable the option first). Now you are stuck and have to enable it in the viewport first to be able to edit anything.\n\nNow restart blender etc. and try this:\n1. Go into Vertex Paint Mode of the default cube\n2. Hit Tab to go into Edit Mode\n3. Disable the cube in the viewport using the outliner as above\n4. Hit Tab to switch out of Edit Mode -> Crash", "Can't undo the keyframe change in edit mode\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3090/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 512.15\n\nBroken: version: 3.1.2\n\nWhile in edit mode, when you hit Undo in the graph editor after moving a keyframe, the graph doesn't change.\n\n- Animate anything using a couple of keyframes.\n- Change to edit mode in viewport\n- Switch timeline to the graph editor (not necessary, you can keep timeline as it is)\n- Select a keyframe and move it\n# Hit undo\n\nResult: The graph editor stays the same\n\nExpected: The move you just made in the graph editor should be the only thing being undone!\n\nTest file with first four steps included\n[#97700.blend](T97700.blend)", "Delete Geometry node face mode behaves unexpectedly when used after Join Geometry\nOperating system: Windows-10-10.0.18363-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 471.68\n\nBroken: version: 3.2.1\nWorked: N/A (problem exists on 3.1.2)\n\n\nNot sure if this is by design, but I find that the following behaviour unexpected:\n\n- By itself, a Mesh Line output connected to Delete Geometry + Geometry Proximity node using Face mode won't delete anything.\n- However, when such a Mesh Line is joined with a Grid primitive, then Delete Geometry + Geometry Proximity in Face mode will cause all of Mesh Line to be deleted, regardless of distances.\n\n\n\n\n\n\n\n\n- Download the blend and observe Mesh Line is preserved.\n- Unmute Grid node link to Join Geometry, and observe Mesh Line is now gone.\n\n[line-deletion.blend](line-deletion.blend)\n", "Improve UX for interaction of modifier stack and node editor\nWhen a non-nodes modifier is selected as active modifier there is no indication or way to change that in the geometry nodes editor and it's very confusing for the user, especially when the properties editor is not displaying the modifiers panel.\n\nThis is related to #84927\nA potential solution has been proposed there as well (move towards interface similar to shader editor and material slots).", "Redo panel disappears during edits in Tweak tool\nOperating system: Win10 x64\nGraphics card: Nvidia 1070\n\n**Blender Version** \nblender-2.80.0-git.26d5a3625ed-windows64\n\nThe info panel for move disappears in edit mode if tweak mode is in action. This does not seem to happen if one uses \"g\" for grab instead.\n\n\n- Enable left click select, optionally select all toggles\n\n- In your space bar menu turn the select mode (W) on\n\n- Create a mesh and move into edit mode \n\n- Start grab and drag of verts and watch the move info panel at the left bottom\n\n- Now try the same thing with selecting a vertex then press \"g\" to move.\n\n", "Text editor undo system, undo step limiting behavior\nCurrently having a linear undo stack has a limitation which is most obvious with text undo.\n\nTyping a character uses a single undo step, meaning if you are working in 3D sculpting for eg, then type a short paragraph over 32 characters - you'll loose all your undo steps.\n\nProposed solution is for undo systems to support counting their undo steps separately, so text editing operations wouldn't use up undo steps for edits in other modes.\n", "Geometry Nodes: Rendered glitch with Image Texture and Scene Time\nOperating system: Linux-5.15.0-76-generic-x86_64-with-glibc2.35 64 Bits, X11 UI\nGraphics card: NVIDIA GeForce GTX 1060 6GB/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 535.54.03\n\nBroken: version: 3.6.0\nWorked: Never\n\n**Short description**\nGeometry nodes that use single images or image sequences are rendered inconsistently but usually look normal in the viewport.\nHigh-res images seem to be more unstable.\nHaving a connected/unconnected scene time node in the geo node group while using an image sequence with any number of frames also makes the rendered image/animation more unstable.\n\n|Expected render|Occasional render glitch|\n|--------------------------|--------------------------------------|\n|| |\n\n\n**Reproduction steps**\nThis is the most unstable setup:\n\n1. Create new project\n2. in the image editor window add a new image with resolution of around 1080x1080 pixels and set the source an image sequence \n3. add geometry nodes modifier to the default cube\n4. follow this node setup; in the image texture node use the created image sequence \n5. Render animation\n6. some of the rendered images will be glitched\n\nI've also tested this approach with face extrusion or instance selection instead of deleting faces. The resulting renders were also glitched. \n\nInside of this zip file is a blender project with a similar setup along with some sample images:\n[geo node groups that use images are rendered inconsistently.zip](attachment)\n\nThis reddit post seems to be describing a similar issue:\ngeometry_nodes_rendering_glitch_info_in_comments/\n\nThe workaround with disabling geo nodes modifier in the viewport mentioned there seems to be working in a simple setup. In the above project from the .zip file it is less effective.\n\n", "Crash: Custom Panel UI Toggle while Undoing\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 441.66\n\nBroken: version: 2.83 (sub 4)\nBroken: 2.82\nBroken: 2.81\n\n\nRead Access Violation when toggling custom UI button and repeatedly undoing. Crash unlikely to happen first time but will happen eventually. In the video, I toggle a few times, then hold Undo while toggling. (Video shows exact steps)\n\nSometimes it can take a bit of time before the crash happens. It might happen faster if you reload the file then repeat the toggling process.\n\n\nOpen attached .blend File\nRun Script\nGo to 3DView -> N -> Panel \"Crash\"\nToggle the button a few times\nHold Undo while continuing to toggle the button. \nCrash occurs eventually.\n\n\n[crash_simpler_vid.mp4](crash_simpler_vid.mp4)\n\n[crash_simpler1.blend](crash_simpler1.blend)\n\nBelow is a pic of where the exception is thrown (Blender version for source pic attached)\n\n\n\n\nThank you for your time and hard work.\n\n\n\n\n", "Modifier changes to support sub-panels for Geometry Nodes\nThis is a follow up of #108649\n\n", "`user_remap` fails silently when object is in edit mode\nBlender 3.6\n\nWay to reproduce:\n1) Create new empty blender project\n2) Select cube\n3) Enter edit mode\n4) Run the code below\n\nResults - mesh wasn't replaced and no errors occured. Expected - to either mesh to change or get some warning/error.\n\n```python\nimport bmesh\nimport bpy\n\nobj = bpy.context.active_object\n\ndata = bpy.data.meshes.new('test')\nbm = bmesh.new()\nbmesh.ops.create_monkey(bm)\nbm.to_mesh(data)\nbm.free()\n\nobj.data.user_remap(data)\nassert obj.data == data\n```\n\n", "Undo when sculpting changes brush type\nOperating system: Windows 10 Pro - 20H2 - 19042.867\nGraphics card: RTX 2070 SUPER - Driver version 497.09\n\nBroken: 3.0.0, f1cca3055776, master, 2021-12-02\n\nWhen sculpting, pressing undo will change the brush type and the brush settings to the previous selection.\n\nThe issue has existed for me for several years across multiple PCs, and countless versions of Blender. It has been reported here by many people but I've never seen anyone mention the following which I think could be helpful. I always work with a second window on another screen with just a UV Editor open which I think has something to do with the issue.\n\nFor example, in a new scene, start with the Draw brush active and change to the Grab brush then make three brushstrokes on the corner of the cube with it to pull it out. Pressing undo once will only remove the last Grab brushstroke as expected but also revert the active brush to the Draw brush which is not expected.\n\nAfter loading factory settings, the problem doesn't exist. What causes it to happen is if you add an additional New Main Window and set it to the UV Editing workspace and combine the UV Editor area and the 3D Viewport area to make just one large UV Editor area. Repeating the previous steps mentioned in the Sculpting workspace then causes the issue to happen.\n\nThere is some inconsistency with what happens afterwards with regards to changing the windows, workspaces and areas around. It seems various combinations can cause this to happen but the following steps should consistently replicate the issue for anyone on factory settings.\n\n - File > Defaults > Load Factory Settings\n - Window > New Main Window\n\n - On the New Main Window, select the UV Editing workspace\n - Combine the two main areas so only the UV Editor area remains, covering most of the workspace, and the 3D Viewport area is removed\n\n - On the original Main Window, select the Sculpting workspace\n - Select the Grab brush\n - Make three brushstrokes\n - Press undo once\n - Observe as one brushstroke is removed as expected but the brush type is also changed back to the default Draw brush\n\nAdditional observation:\n - Close the New Main Window that has the UV Editor open\n\n - On the original Main Window, select the Sculpting workspace\n - Select the Grab brush\n - Make three brushstrokes\n - Press undo once\n - Observe as one brushstroke is removed as expected and the brush type does not change" ]

[ "Editing redo panel properties of torus in edit mode reverts some steps\nOperating system: macOS-10.14.6-x86_64-i386-64bit 64 Bits\nGraphics card: NVIDIA GeForce GT 650M OpenGL Engine NVIDIA Corporation 4.1 NVIDIA-12.0.24 355.11.10.50.10.103\n\nBroken: version: 3.4.1\nWorked: (never worked as expected)\n\nAdding and then adjusting a Torus or any Add Mesh add-on in edit mode reverts edits made to nodes, video sequencer, and more, since entering edit mode on the mesh object.\n\n1. Go to the Geometry Nodes workspace\n2. Select the default cube\n3. Click the New button to add the Geometry Nodes modifier and create and assign a new node group\n4. In the 3D Viewport switch to Edit mode\n5. In the Node Editor, pour your heart and soul into crafting an elaborate node tree (lol, jk 👀💔)\n6. In the 3D Viewport add a Torus\n7. Tweak any setting in the Adjust Last Operation panel\n8. Observe Geometry Nodes edits reverted (all edits since entering edit mode in the 3D Viewport)\n\nNo other built-in mesh primitives (nor even primatives!) have this effect, but the Add Mesh add-ons all do.\n\nI have checked some of the editors to see what data is reverted and what is preserved. This is not yet an exhaustive review.\nThe following data is reverted:\n- Video Sequencer\n- Properties\n- every Node editor\n- Timeline\n\nThe following data is preserved:\n- Preferences\n- Text Editor\n- Python Console\n- Movie Clip Editor\n- UV Editor\n\nThe Outliner has mixed results:\n- Hierarchy expansion reverts in Blender File, Data API, and Orphan Data but not in Scenes nor View Layer.\n\nSome editors have special behavior:\n- Image editor\n - Paint mode\n - {Add torus, paint image, tweak torus} reverts the paint\n - {Paint, add, tweak} kicks out of Edit mode into Object mode\n - Switching Image Editor modes does not close the ALO panel. The following sequence of steps produces a Python error:\n - Add torus\n - Switch Image Editor modes\n - Adjust torus\n (If this merits its own tracker issue, tell me and I will write one up. This all seems so tied to one root culprit that it is not my initial inclination to do so)\n\n" ]

[ "Overlapping shadows of MNEE Caustics and filter glossy caustics produce incorrect result\n**System Information:**\nOperating system: Linux-5.17.0-1-amd64-x86_64-with-glibc2.33 64 Bits\nGraphics card: NVIDIA GeForce RTX 3060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 510.68.02\n\n**Blender Version:**\nBroken: version: 3.3.0 Alpha\nWorked: NA as this issue is probably in all builds of Blender with MNEE\n\n**Short description of error:**\nWhen the shadows of an `Shadow Caustic` light and a normal light overlap and filter glossy is set higher than 0, then the caustics produced in this overlapping area by the normal light will be rendered partially with the set filter glossy setting and partially with a filter glossy of 0.\n\nHere is some images demonstrating the issue (The file I used for testing is attached below):\n||||\n| -- | -- | -- |\n|Expected result|Actual result|Actual result - Label image|\n\n**Exact steps for others to reproduce the error:**\n1. Change the render engine to Cycles.\n2. Setup the scene for MNEE caustics by adding a caustic caster, caustic receiver, shadow caustic light, and configuring them properly.\n3. Add another light to the scene and ensure the `Shadow Caustic` option is NOT enabled.\n4. Place this new light somewhere in the scene where part of the shadow and caustics caused by the light passing through the caustic caster onto the caustic receiver overlaps with the shadow from the light with the `Shadow Caustic` option enabled.\n5. Ensure the {nav Render Properties > Light Paths > Caustics > Filter Glossy} is set to a value larger than 0.\n6. Reduce the power of the light with the `Shadow Caustic` option enabled so it has only a small impact on the final render. (Note: The power of the light must be above 0 for this to work) - This step is not necessary, but it makes it much easier to see what's happening.\n7. Render the scene.\n8. Compare that render to another render with the `Shadow Caustic` light excluded from the final render. Notice how there was an impact on the shadow and caustics produced by the non-`Shadow Caustic` light.\n\nHere is a .blend file with steps 1-6 done for you:\n\n[#97932 - Overlapping Caustic Shadows issue.blend](T97932_-_Overlapping_Caustic_Shadows_issue.blend)\n\nThe steps for testing with the file is as follows:\n1. Open the attached file.\n2. Render the scene. Notice the weird line in the caustics.\n3. Disable `MNEE Light` from the final render and render again, comparing the result to the first render you did.", "Camera Background image is appearing greyish in 3D Viewport \nOperating system: Windows 10\nGraphics card: Nvidia GTX 1080Ti\n\n**Blender Version** 3.2.1\nBroken: version: 3.2.1, branch: master, commit date: 2022-07-05 15:44, hash: a2d59b2dac9e\nWorked: \n\nCamera Background image is appearing greyish in deep black area in 3D Viewport but showing completely normal in Image editor section and renders. \nEven after turning on \"View as Render\", problem is still there.\n \nOpen the attached blend file.[Blender BG Bug Report.blend](Blender_BG_Bug_Report.blend)", "High Resolution Render support\nSome fields needs to be able to output really high resolution renders out of EEVEE.\n\nFor now, we just try to allocate the asked buffer size and live with the consequence.\n\nMost GPU have a limit of 16K² texture resolution and some really high renders could not fit GPU memory in some cases.\n\nTo workaround this issue, we need to implement a multipass rendering.\n\nThe issue with this approach is that Screen Space effects (GTAO, SSR, SSRefract, SSSubSurfaceScattering) cannot take the whole picture into account.\n\nWe can split the rendering into tiles, scanlines, checkerboard pattern but the artifacts of SS effects will likely to be different for each", "Cycles: sRGB texture compression happens after interpolation \nBroken: 2.81\n\nCycles does sRGB->linear texture conversion after texture interpolation instead of before. This leads to incorrect results.\n\nExample scene is attached. Two squares with a fine checkerboard texture, one with linear interpolation, the other with closest interpolation.\n\nExpected result: Both textures should render to an identical shade of gray. When switching the texture color space to \"linear\" instead of \"sRGB\", the result should not change as the texture contains only pure black and white pixels.\nActual result: The textures show different shades of gray. Switching the color space changes the result.\n[checkerboard.blend](checkerboard.blend)\n\n", "different generation of graphic card won't utilise together.\nOperating system: Windows 10\nGraphics card: Radeon R9 295X2 (secondary chip) + Radeon RX560 4GB\nAMD Ryzen 1700X\nMSI X370 Gaming Pro Carbon\n3 * 8GB DDR4-2400\n\nBroken: 2.82.7 to 2.83.3 <--- where x is from 0 to current stable version\nWorked: 2.8, 2.83.4 (partly worked, may end up sudden wattman reset)\n\nRadeon Version: 20.7.2\n\nafter the 3rd step, the render did not show anything, switching back to material viewport, the whole program was frozen. not sure if that is graphic card problem, but that way i could not even render anything. on 2.82.7, this also happened and the render kernels failed to work.\n\n\n1. open up a file of as shown\n2. set the preferences that would use OpenCL to render with R9 295X2 (and RX560)\n3. preview in Cycles\n4. unable to load the preview, switch back to material viewport.\n\n[zeta school 1.blend](zeta_school_1.blend)", "Regression: black seams in my Cycles bakes (caused by denoising) due to margin pixel extension darkening\nOperating system: Windows-10-10.0.19042-SP0 64 Bits\nGraphics card: GeForce RTX 2080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 461.92\n\nBroken: version: 3.0.0\nWorked: 2.93.7\n\nI get black seams when I bake textures that I never used to have before upgrading to 3.0. They occur on any corner that I UV as shown here: 4773f8c541316a64af42092a919c3e64.png\n\n[Black Seams.rar](Black_Seams.rar)\nUV a mesh with only 1 split on the corners leaving the other sides still attached. Bake a combined texture to 2048x2048 resolution with Cycles. Apply to mesh. Observe the black seams on the corners.\nNow open that same file and bake that same mesh in 2.93.7. No black seams. Re-save file in 2.93, and re-open that file in Blender 3.0, and it now bakes without a black seam too.\n\n", "Bevel improvements for \"Correct Face Attributes\"\n@mano-wii recently made a new commit {4387aff99e01} for adjusting uv's during mesh transformation. After discussing with him and @ideasman42 I was asked to add a new task with suggestion for improvement to the bevel operator so that it works well with **correct face attributes** like mesh transform does. \n\n---\n**Relevant links**\n\n#78671 (Improvements to \"Correct Face Attributes\")\n\n---\n\nSettings for **correct face attributes** can be found in the top right corner of the 3d viewport.\n \n\n\n**Bevel - bevel edge(s) currently does not work with **Correct face attributes****\n\n[bevel does not correct uv.mp4](bevel_does_not_correct_uv.mp4)\n\n**Suggested behavior for bevel**\nStep 1 - selection\n\n\nStep 2 - bevel\nCreated face inherits uv coordinate from one face and then uv coordinates are corrected. UV overlap are expected in a lot of cases.\n\n", "Image Texture shader node with Box projection outputs extreme values when mesh has unusual normals\nOperating system: Windows-10-10.0.19042-SP0 64 Bits\nGraphics card: GeForce RTX 2070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 457.30\n\nBroken: version: 3.0.0\nWorked: can't test in other versions, since they can't open .blend files saved with Blender 3.0\n\nIf you open the attached file, you will see a mesh that's black everywhere except for some *very* bright pixels (the Bloom effect makes this glitch very pronounced).\nThe material is just an Image Texture node connected to BSDF's color, using Box projection and a completely black texture. Theoretically, it should only output interpolations between the colors of the source texture, but that's definitely not the case here.\nOn its own, the mesh doesn't exhibit the glitch, but some modifiers (such as Displace or Wave) make it manifest for certain combinations of parameters. I suspect that the mesh has some unusual normals or face orientations that the modifiers amplify into even weirder normals, which in turn cause trouble in Image Texture's Box projection.\n[shiny glitch blender3.blend](shiny_glitch_blender3.blend)\n\nOpen the attached file, wait for shaders to compile.", "Viewport material is darker than the rendered version\nOperating system: Windows-8.1-6.3.9600 64 Bits\nGraphics card: Intel(R) HD Graphics 4400 Intel 4.3.0 - Build 10.18.14.5067\n\nBroken: version: 2.82 (sub 3)\nConfirmed in 2.81 release\n\n\nThe material consists \na transparent shade mixed with a translucent shade, mixed with an emission shade with\nfrenel node multiplied with a noise texture as the factor\n\nThe sample sizes are the same\n\nJust render\n\n\n[necro_core_reduced.blend](necro_core_reduced.blend)", "The bridge tool fails in a situation with multiple rows of faces facing each other\nWin10 Pro 64 bit; AMD Ryzen 2700X; Nvidia GTX 1060 6GB; 32 gigs RAM\n\nBroken: hash d1be30f (buildbot build May 4).\nWorked: never\n\nIn a situation where you have multiple rows of faces that face each other across empty space (with other areas where they face away from each other). The bridge tool (even with*loop pairs* checked) is unable to give priority to the direction the normals are facing. The result of this is that instead of getting nice bridges among the multiple rows of cubes, the tool produces faces that overlap and have zero area. As hinted, the*loop pairs* checkbox in this case does not do anything and the result is the same.\n\nOpen the .blend, invoke the bridge operator (the result is the same whether or not \"loop pairs\" is selected.\n\n**blend**\n[IffyBridge.blend](IffyBridge.blend)\n\n[IffyBridge.blend](IffyBridge.blend) (updated blend)\n\nAn additional note; I submitted this issue years ago, but it was never fixed and I couldn't find my old report. If the bug is not going to be fixed, then just close this as a todo item or with a \"won't fix\" tag.\n", "Cycles freeze on render and texture baking (with packed textures)\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 1660 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 461.72\n\nBroken: version: 2.92.0\nWorked: 2.83\n\nThis broke somewhere between b21ba5e579 and f2b5f731d5\nAssume 99436acde8\n\nBlender freezes on render\nBlender also freezes on texture bake and rendered viewport\nTried on 3 different PCs, all the same\n\nIf textures are unpacked, rendering is fine.\n\nSimply open attached .blend file and hit F12\n\n[render_test2b_delmesh3_simplified.blend](render_test2b_delmesh3_simplified.blend)", "MacOS Crease Problem Back end Metal\nOperating system: Macbook Pro 14\nGraphics card: Apple M2 Pro\n\n3.6.2\n\n\nWhen the backend setting of macos is set to metal, the color change of the edge when applying blender's vertex crease. When edge crease is used, magenta color does not come out and it is processed as black color.\n\nrecording video\n\nview?usp=sharing\n\nview?usp=sharing\n\n", "Keyed Particles don't render as expected\nOperating system: Windows-10-10.0.17763 64 Bits\nGraphics card: Intel(R) HD Graphics 4000 Intel 4.0.0 - Build 10.18.10.5069\n\nBroken: version: 2.81 (sub 16)\nWorked: (optional)\n\n[KeyedParticlesJL.blend](KeyedParticlesJL.blend)\n\n- Open file\n- Go to frame 105\n- Render\n- Compare render to viewport\n\nRender is different to viewport\n\nThat's what I expected:\n[ViewportRender.mp4](ViewportRender.mp4)\n\nThat's what I get when rendering with Eevee:\n[EeveeRender.mp4](EeveeRender.mp4)\n\nThat's what I get when renderng with Cycles:\n[CyclesRender.mp4](CyclesRender.mp4)\n", "Screen flickers / goes black/ red when starting up Blender in fullscreen via startup file.\nOperating system: macOS Big Sur 11.0.1\nGraphics card: Radeon Pro Vega 48 8GB\nProcessor: 3.6 GHz 8-Core Intel Core i9\niMac (Retina 5K, 27 inch, 2019)\nMemory: 72GB 2667 MHz DDR4\n\nBroken: Blender 2.83.9, 2.92 alpha `ba740ad2abc6`, 8dc95f61f3\n\nThe screen flickers/ goes black/ goes red for about 2-3 seconds if the startup file was saved with Blender in full screen. Opening the startup file's again after Blender becomes usable doesn't cause this.\n\n\n- Delete the 2.83 config file or start from Factory Settings\n- Start Blender and the splash screen is displayed correctly\n- Now Click the Green button at the top left hand corner to Maximise the screen to full\n- Save the Default Startup File\n- Reopen Blender and the screen turns red apart from the header bar which is displayed correctly\n- The screen stays red for about 3 seconds\n# The splash screen is now displayed correctly and I can use Blender so far without any issues\n\nRestarted the iMac and restarted Blender with the same red screen error as above\n\n**Workaround**\nPress the Green Button so that Blender is not maximised anymore\nSave the Default Startup File\nRestart Blender and the splash screen is displayed (without the red screen)\n\nThis is my first bug report, so I hope I have filed this correctly.\n\n\nKind Regards,\n\nClayton", "Diffuse Direct, Indirect Bake to Vertex Color Artefacts\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.94\n\nBroken: version: 3.5.0\nWorked: This has never worked properly (At least in my experience)\n\nWhen baking Diffuse light to vertex color, black triangles and random dark spots appear in vertex color. This effect is more apparent with:\n- Area lights.\n- Emissive surfaces.\n- Lights with very high power.\n\n1. Create geometry and give each a byte_color vertex color channel.\n2. Place area lights in scene with high power.\n3. Bake diffuse direct and indirect to color attribute.\n\nSee .blend of my project attached. In the file i've already baked the emissive light to vertex color where you can see the artefacts. Switch to cycles live render view to see the correct output. I think the tiled texture may have something to do with it. I've added a tiled texture and packed it in the example.\n\n" ]

[ "Creating black color when two objects intersect in Cycles render\nCreating black color when two objects intersect in Cycles render.\nThis only happened on Blender 3.1, the 3.0 and before versions are fine\n\nOperating system: Windows 10\nGraphics card: RTX 3070(Zotac Gaming)\n\nBroken: Blender 3.1\nWorked: older versions 3.0 and before\n\n\n**Exact steps for others to reproduce the error**[Blender bug 3.1 2022.blend](Blender_bug_3.1_2022.blend)\nBased on the default startup or an attached .blend file (as simple as possible)." ]

[ "Rip edge-move for curves\nCurrently with meshes you can press Alt+D to add a vertex at one side of the verts edges and drag it *(based on the mouse cursor location)*.\n\nThis same behavior could be added to curves.\n\n- the basic behavior with single vertices selected is simple\n- with multiple segments are selected this becomes more involved (see how this works for meshes).", "Box select shortcut is missing deselect tool tips\nOperating system: Windows 10\nGraphics card: NVIDIA GeForce GTX 1080 Ti\n\nBroken:\nBroken: 2.80, 2018-11-30\n\n\nThe circle select shortcut shows tool tips for deselect, the box select does not despite having the functionality.\n\n\n* Open blender, select new and pick \"General\".\n* Select the cube. (left click)\n* Enter edit mode. (tab)\n* Deselect all. (alt-a)\n* Enter box select mode. (b)\n* Select a vert (left click and drag)\n* Enter box select mode. (b)\n* Deselect a vert (shift left click or middle click)\n* Notice the vert has been deselected despite there being no tooltip\n\n", "\"Input Pending\" for Loop Cut and Slide when called from menu.\nOperating system: Windows 10 64 bit\nGraphics card: Nvidia GTX 1660Ti\n\nBroken: 3.3.2\nWorked: 2.93\n\n**Caused by:** da2ba402689d7ed752b0b37ad49dce1a5851cc77\n\n**Reproducing the error:**\n\n- Ctrl+R on default cube edit mode, regardless of selection mode, loop cut slide works as usual.\n- Use search menu or header->edge menu and select loop cut slide, header bar only shows \"Input Pending\", if pressed left mouse, then operator runs correctly. If pressed right mouse, then operator is cancelled.\n\n\n\n", "Weight Paint Vertex Selection Display issue\nOperating system: Linux Mint\nGraphics card: RTX 3070\n\nBroken: latest master\nWorked: never really, more on that below\n\nWhen in weight paint mode with\n* vertex selection enabled\n* show wire enabled\n\nedges will not be drawn within the selection\n\n\n**The odd part**\nClicking in an empty space will get the wires to show again, but keep the vertices selected. \nThat's a behaviour that has changed from 3.1.2 to 3.2.2\nPreviously it deselected the vertices but didn't draw the edges\n\n\n\nBased on the default startup or an attached .blend file (as simple as possible).\n* Open the attached blender file\n* you can see that edges are not drawn between selected vertices\n* click in an empty space (important to click, not click drag)\n* vertices stay selected but wires are drawn\n\n", "Node Panels Polishing\n- [ ] Support A to expand/collapse panels.\n- [ ] Support Ctrl+click to isolate panel.\n- [ ] Support dragging all panels to collapse/expand them.\n- [ ] Use the horizontal padding vertically:\n\nBefore:\n\n\n\nAfter:\n\n\n\n---\n\n- [ ] Don't space out socketes when hiding the node\n\n\n- [ ] Bug, hide panel when all its sockets are hidden with Ctrl+H\n\n", "Skin modifier does not work on meshes created from scratch during evaluation (primitives, etc.)\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 471.11\n\nBroken: version: 3.0.0 Alpha\nWorked: NA\n\nSkin Modifier does not work with Geometry Nodes Mesh Primitives, see attached blend file.\n\nOpen attached file, note that Skin Modifier has no effect.\n\n[GN_skin_modif_bug.blend](GN_skin_modif_bug.blend)\n", "'BLOCKING' operator option on multiple windows\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1050 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.59\n\nBroken: version: 3.2.0\nWorked: -\n\n'BLOCKING' operator option on multiple windows stop blocking everything else from using user input on window focus change (active window).\n\n* Enter edit mode, select Knife Tool.\n* Open new window in same session\n* Start Knife operator (because its built-in 'BLOCKING' modal operator) in first window\n* Mouse click on desktop/Blender console/taskbar (change window focus)\n* Click in second window (its already \"unlocked\")\n* Select all > Delete\n* Click on first window\n* Blender crush\n\n\n\n[blender_system_info.txt](blender_system_info.txt)\n\n[blender_debug_output.txt](blender_debug_output.txt)\n\n", "Nodes: Reroute node disappear when pressing `Ctrl H`\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1650/PCIe/SSE2 NVIDIA Corporation 4.6.0 NVIDIA 536.23\n\nBroken: version: 4.0.0 Alpha\nWorked: Never.\n\nIf the reroute is not connected to anything, then pressing `Ctrl + H` will make the reroute invisible.\n\n\n1. Add reroute in nodetree using `Shift + A`.\n2. Press `Ctrl + H`.\n\n", "Linked collection with smoke simulation cached doesn't work\nOperating system: Windows 10\nGraphics card: RTX 2070 laptop\n\nBroken: 2.82\nWorked: IDK\n\n Linked collection with smoke doesn't seems to load the cache properly or something.\n\n- Create a scene with a quick effect smoke,\n- Save the file,\n- Bake the simulation.\n- Save the file again,\n- File -> New -> General\n- File -> Link the collection of the another file\nThe domain doesn't load the cache", "Sculpting - Cloth brush - Snake Hook deformation not working\nOperating system: macOS-10.15.7-x86_64-i386-64bit 64 Bits\n\n\nBroken: version: 3.5.0 Alpha\n\nThis is the image I got after kept dragging my cloth brush around the same spot for a few times:\n\n\nand this is the file I'm using:\n[20221206_sculpt_cloth_brush_snake_hook_donot_work_0001.blend](20221206_sculpt_cloth_brush_snake_hook_donot_work_0001.blend)\n\nI personally think the option is 'wired' to a wrong block of code.\n\n", "Mesh Bevel Node\nNOTE: Design outdated. See #98674 for current design and progress. It is a development task though, so please don't add comments like \"when will this be finished?\"", "Bisect Tool Snap does not Work\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 536.67\n\nBroken: version: 3.6.1\n\nWhen using the bisect tool and dragging the yellow arrow, snapping does not work when holding the CTRL key despite it being an option in the context tool bar. With the rotate gizmo it works fine. \n\n1. Add for example Suzanne.\n2. Select the whole mesh.\n3. Use the bisect tool.\n4. Try first the rotate gizmo and hold Ctrl, snapping works fine.\n5. Now use the yellow arrow instead, no snapping at all regardless of the snapping settings.", "Particle edit select linked not working from menus\nOperating system: Windows-7-6.1.7601-SP1 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 430.86\n\n\nBroken: version: 2.80 (sub 74)\n\n\nHi, I noticed that the select linked option in hair \"Particle edit\" mode/Select/ select linked doesn't work\n\nPlease open the file press W then select linked\n\nthankyou\n\n[hair.blend](hair.blend)\n\n\n", "Shader editor: nodes moved with G key added to frames based on cursor position\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 452.06\n\nBroken: version: 2.83.4 (also with the 2.90 beta from August 25th)\nWorked: I don't know. I remember this strange behaviour for a long time (2.80?) but could not pinpont the exact behaviour until now\n\nIn shader editor, when you have a frame and move nodes outside it with the shortcut G, the node is added to the frame based on the cursor position and not on the node position.\nAs expected, when pressing G key, a user does not align the cursor to the node to be moved but presses the G key regardless of the cursor position.\n\n- Default startup file\n- In shader editor, add and join some nodes into a frame (Ctrl J)\n- Grab another node outside the frame with G with the cursor position far from the node to be moved\n- Move the cursor so that the cursor is inside the frame but the node is not and click to confirm new position\n- The node is added to the frame based on the cursor position although it is far outside the frame\n", "Cut brush - curve sculpting\nSee #96259.\n\nThis initial implementation doesn't need yet to have the Knife overlay control (similar to mesh edit knife)." ]

[ "Keyboard shortcut cut links (Ctrl + click) don't work in the Shader Editor\nOperating system: Windows 7 64 bits\nGraphics card: Nvidia GTX 750\n\n**Blender Version** \nBroken:\n2.80 (sub 35), branch: blender2.8, commit date: 2018-12-03 00:27, hash: a205493426b, type: Release\nbuild date: 02/12/2018, 17:01\n\n\nIn the Shader Editor, the keyboard shortcut cut links (Ctrl + click) don't work.\n\n\nCrtl + clic on the Shader Editor to cut the links of the nodes\n\nI add, since the first day of the Beta 2.8 the bug is present\n\nThanks.", "shotcut cut links in node edit not working with 2.80Beta\nWindows 10 Pro 64bit\nCPU : intel i7-7700\nRAM : 32 GB\nGPU : GTX 1060 6GB\n\nshortcut **Cut Links**(Ctrl + Left Mouse) in Node editor **not working** with 2.80 Beta\n\nversion download - blender-2.80.0-git.0f5b53ba4dc-windows64\n\n\nAlpha\n\n\nBeta\n\n\n" ]

[ "Cycles: improve subsurface scattering for some meshes shapes\nIn cycles-x, the Christensen-Burley dipole approximation was fully replaced by random walk.\n\nHowever for some mesh shapes results are not as good, or could be improved regardless.\n* Self-intersection meshes give dark borders. Counting back/front facing intersections could help determine inside/outside better to avoid such artifacts.\n* Sharp corners \"glow\", which may be physically correct but is not necessarily desired. It may be possible to add user control with an option somehow, using some minimum throughput\n\nIf these can't be solved, we may consider adding back the dipole approximation, but since path splitting and the memory usage is a difficult fit for the new kernels we'd really like to avoid it.", "Output from Geonodes Edge Domain to Shader behaves strangely\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: Quadro K3100M/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 426.78\n\nBroken: version: 3.1.0 Beta\nWorked: Never, as far as I know\n\nWhen retrieving data stored on the edge domain in a shader, the values are inconsistent between adjacent faces, and generally do not interpolate across faces as expected.\n\n\n\n\nOpen attached file and switch to rendered mode.\n\n[Edge Domain Issue Demo.blend](Edge_Domain_Issue_Demo.blend)\n", "Regression: GPU subdivision - Mesh does not immediately count viewport statistics\nOperating system: Windows-10-10.0.19043-SP0 64 Bits\nGraphics card: Radeon RX 570 Series ATI Technologies Inc. 4.5.14800 Core Profile Context 22.3.1 30.0.15002.1004\n\nBroken: version: 3.3.0 Alpha\nWorked: blender-3.3.0-alpha+daily.`d095fcd6b417`\n\nI didn't do rigorous bisecting but I suspected this is caused by 163f6a17e4 so I tested with a daily build before it was committed.\n\nMesh does not immediately count in viewport statistics with GPU subdivision enabled.\n\n\n\n- Open startup file.\n- Add Subdivision Surface modifier to the cube\n- See the mesh is not counted towards viewport stats\n- Deselect the cube (or do some other action related to viewport display, for example, ticking viewport display options), and see now it is counted towards viewport stats\n", "Wrong SubSurf modifier results with inconsistent normals\nOperating system: Windows 10 Enterprise Version 1803\nGraphics card: 2 x GTX1080 ti\n\nBroken:\nblender2.8, 2018-12-05\n\nFrom latest builds subsurf modifier make a wrong subdivision of geometry as displayed in the attached files.\nBefore 18/20 November the behaviour of subsurf modifier was ok.\nI am sure because I did an animation on that date and the subdivided vase was ok.\n\nBest regards,\n\nUmberto\n\n\n\n\n\n[ceramic.blend](ceramic.blend)", "Random crashes with NVidia + GPU Subdivision (sometimes when activating X-Ray or Wireframe overlay)\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\n| Graphics card: | Driver : | User:\n| --- | --- | ---\n| NVIDIA GeForce RTX 3090/PCIe/SSE2 | 4.5.0 NVIDIA 516.94 | @GabrielMoro\n| NVIDIA GeForce GTX 1050 Ti/PCIe/SSE2 | 4.5.0 NVIDIA 512.15 | @STYXtheMUTT\n| NVIDIA GeForce RTX 3070 Ti/PCIe/SSE2 | 4.5.0 NVIDIA 516.94 | @Jojolix\n|NVIDIA GeForce RTX 2070 SUPER/PCIe/SSE2|NVIDIA 516.59|@MassimilianoPuliero\n|NVIDIA GeForce RTX 3090|NVIDIA 516.94|@The5\n\nBroken: version: 3.2.2\n\nOn some seemingly random occasions, Blender crashes with the following stack:\n```lines=10\nStack trace:\nblender.exe :0x00007FF77C1458B0 BKE_subdiv_free\nblender.exe :0x00007FF77C147880 BKE_subsurf_modifier_subdiv_descriptor_ensure\nblender.exe :0x00007FF77BFFC3D0 <lambda_c000e58b9a8caf72c54e0e02a46effdb>::operator()\ntbb.dll :0x00007FFDBAF64FD0 tbb::interface7::internal::isolate_within_arena\nblender.exe :0x00007FF77BFFC6F0 BKE_mesh_wrapper_ensure_subdivision\nblender.exe :0x00007FF77CB38FA0 raycast_obj_fn\nblender.exe :0x00007FF77CB375E0 iter_snap_objects\nblender.exe :0x00007FF77CB3E530 transform_snap_context_project_view3d_mixed_impl\nblender.exe :0x00007FF77CB3F0E0 ED_transform_snap_object_project_view3d\nblender.exe :0x00007FF77CB5E7C0 applyFaceProject\nblender.exe :0x00007FF77CB5EE00 applySnappingIndividual\nblender.exe :0x00007FF77CB72F70 recalcData_objects\nblender.exe :0x00007FF77CB81010 applyTranslation\nblender.exe :0x00007FF77CB46BE0 transformApply\nblender.exe :0x00007FF77CB41920 transform_modal\nblender.exe :0x00007FF77C229D30 wm_macro_modal\nblender.exe :0x00007FF77C20F9E0 wm_handler_operator_call\nblender.exe :0x00007FF77C210BD0 wm_handlers_do_intern\nblender.exe :0x00007FF77C210000 wm_handlers_do\nblender.exe :0x00007FF77C216590 wm_event_do_handlers\nblender.exe :0x00007FF77C1FC380 WM_main\nblender.exe :0x00007FF77B2F12A0 main\nblender.exe :0x00007FF7814128B0 __scrt_common_main_seh\nKERNEL32.DLL :0x00007FFDC7B67020 BaseThreadInitThunk\nntdll.dll :0x00007FFDC85A2680 RtlUserThreadStart\n```\n```lines=10\nStack trace:\nblender.exe :0x00007FF7A50C5B10 BKE_subdiv_update_from_mesh\nblender.exe :0x00007FF7A50C7880 BKE_subsurf_modifier_subdiv_descriptor_ensure\nblender.exe :0x00007FF7A52A1110 draw_subdiv_create_requested_buffers\nblender.exe :0x00007FF7A52A2080 DRW_create_subdivision\nblender.exe :0x00007FF7A52D1870 DRW_mesh_batch_cache_create_requested\nblender.exe :0x00007FF7A52C6000 drw_batch_cache_generate_requested\nblender.exe :0x00007FF7A529F1A0 drw_engines_cache_populate\nblender.exe :0x00007FF7A529B8C0 DRW_draw_render_loop_ex\nblender.exe :0x00007FF7A529C9B0 DRW_draw_view\nblender.exe :0x00007FF7A5D3FA00 view3d_main_region_draw\nblender.exe :0x00007FF7A5551070 ED_region_do_draw\nblender.exe :0x00007FF7A51A3700 wm_draw_window_offscreen\nblender.exe :0x00007FF7A51A3560 wm_draw_window\nblender.exe :0x00007FF7A51A2FB0 wm_draw_update\nblender.exe :0x00007FF7A517C380 WM_main\nblender.exe :0x00007FF7A42712A0 main\nblender.exe :0x00007FF7AA3928B0 __scrt_common_main_seh\nKERNEL32.DLL :0x00007FFD1C4A54D0 BaseThreadInitThunk\nntdll.dll :0x00007FFD1D524830 RtlUserThreadStart\n```\n\nOccasions observed:\n- try to snap objects on faces (the object that shall be snapped to is usually a complex mesh with modifiers like subdivisions, mirror, solidify, bevel and booleans)\n- toggle the shading mode between shaded and wireframe or between X-Ray and Wireframe Overlay.\n[intro.crash.txt](intro.crash.txt)\n\n\n- open attached file\n- disable and enable X-Ray multiple times\nIt happens specially when working fast.\n\n[BLENDER_Youre_Alive_v67.blend](BLENDER_Youre_Alive_v67.blend)\n\n", "Blender crashes when I try to subdivide my model\nOperating system: Windows 11 Pro 64-bit, 12th Gen Intel(R) Core(TM) i9-12900KF (24 CPUs), ~3.2GHz\nGraphics card: NIVIDIA GeForce RTX 3070 Ti\n\nBroken: 3.6, latest version\n\n\n\n- Open Blender, Open File\n- Go to Preferences, Turns off GPU Subdivision\n- Change to CUDA and tick Graphics Card Box\n- Goes to Modifier, Click Subdivision Surface, Processing, Crashes\n\nTo give a bit more context, my current model only has 160000 vertices which I think is quite little? But still it cannot subdivide and crashes whenever I click the subdivide option in modifiers.\n\n", "Bevel modifier disconnects vertices in the UV map\n{[F8713789](2KmuzHs.png)}Operating system: Windows 10 x64\nGraphics card: GTX 970\n\nBroken: 2.83.1, 47b6c332587ca45eddc98e75b3f563e2f17a0d47, probably all 2.8+ versions\n[SM_UVTest.blend](SM_UVTest.blend)\n\nWhen using an angle or weight limit on the Bevel modifier, vertices connecting beveled and unbeveled edges are split in the UV map, producing distorsions.\n\nAdded a blend file. To reproduce the issue, apply the Bevel modifier, and select all faces to see the problematic vertex in the UV editor. \n\nSome context is here : 1241831", "stack_overflow exception while rendering with viewport render on\nOperating system: win10\nGraphics card: 2070super\n\n[blender_system_info.txt](blender_system_info.txt)\n\nBroken: 2.83.4, 2.91 6fe609d4c3\n\n\nwhen pressing f12 in render mode, blender crashes,\nwhen pressing render in solid mode, blender doesnt crash\n\n[blender_debug_output.txt](blender_debug_output.txt)\n```\nError : EXCEPTION_STACK_OVERFLOW\nAddress : 0x00007FFF42963FDB\nModule : C:\\Windows\\system32\\DriverStore\\FileRepository\\nv_dispsi.inf_amd64_6af46268c19e4980\\nvcuda64.dll\n```\n[2.blend](2.blend) \n1. switch to render mode\n2. press f12 & watch task manager.\n3. blender will close", "GPU Subdivision Causes Artifacts in Mesh Analysis Overlay\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3090 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 531.41\n\nBroken:\nversion: 3.1.1 Release Candidate\nversion: 3.3.1\nversion: 3.5.1\nversion: 3.5.1, branch: blender-v3.5-release, commit date: 2023-04-25 13:50, hash: 6ee8aa4997ee, type: release\nversion: 3.6.0 Alpha\n\nWorked: could not find.\n\nWhen GPU subdivision is enabled, the Mesh Analysis overlay shows artifacts.\n\n\nPlease follow these steps to reproduce the issue:\n1. Start Blender.\n2. Select the default cube.\n3. Apply the Subdivision Modifier (press Ctrl+1).\n4. Enter edit mode.\n5. In the viewport overlay menu, enable Mesh Analysis.\n\n**Known Workarounds**\n1. Disable GPU Subdivision.\n2. Disable \"Show in Edit Mode\" for the Subdivision modifier.\n\n", "Mirror threshold too high, values less than 0.00002 are snapped to center\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX TITAN X/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 431.60\n\nBroken: version: 2.83.3\nWorked: 2.79b\n\nBlender 2.8x has a bug that forces it to snap to the YZ plane with a value less than 0.00002.\nThe above is occurring with the snap off.\nThe bug occurs even if the vertex's X coordinate is moved to more than 0.00002.\nDuplicating the object does not eliminate the bug.\n\n3D Viewport / Edit Mode.\nCreate an object with the following vertices:\n```\nObject-1 = {\n Vertex-A x = 0.00002, y = 0.00005, z = 0.0\n Vertex-a x = -0.00002, y = 0.00005, z = 0.0\n Vertex-B x = 0.000019, y = -0.00005, z = 0.0\n Vertex-b x = -0.000019, y = -0.00005, z = 0.0\n}\n```\n3D Viewport / Edit Mode / Tool / Options / Mirror X [enable].\n3D Viewport / Edit Mode / Snap [disable].\nSelect Vertex-A and Vertex-B.\nMove Vertex-A and Vertex-B.\nBlender 2.8x has a bug that forces it to snap to the YZ plane with a value less than 0.00002.\n\n3D Viewport / Object Mode.\nDuplicate the Object-1 to obtain the Object-2.\n3D Viewport / Edit Mode.\nMirror the vertices of object 2 as if they were object 1. This causes the same problem.\n\nEnlarge the Object-2 sideways.\nMirror the vertices of Object-2. The same problem still occurs. The problem occurs even if the Vertex-B x-coordinate is greater than 0.00002 after magnification.\n\n\n[X-Mirror_bug.blend](X-Mirror_bug.blend)\nThe objects in the X-Mirror_bug.blend file are very small, so zoom in on the view to the maximum size.\n\n\nThis is a 2.79b file with the X-Mirror working correctly:\n[X-Mirror_2.79b.blend](X-Mirror_2.79b.blend)\n\n\nThe difference in x-mirror behavior between 2.8x and 2.79b is shown in the following video:\n[Blender 2.8x BUG_X-Mirror.mp4](Blender_2.8x_BUG_X-Mirror.mp4)\n\n[Blender 2.79b X-Mirror.mp4](Blender_2.79b_X-Mirror.mp4)\n\n\nIf it is unintentionally aligned to the YZ plane like 2.8x, it will not be possible to edit it (if this behavior is a specification, it will be useful if you can implement it as a function separate from the mirror, align it to the YZ plane, and set a threshold to align).\n\n**When importing model data from the outside and editing a mirror, all vertices with an X coordinate of less than 0.00002 are already implicitly aligned to the YZ plane, so editing with an X mirror is no longer possible at all. Hence, until this bug is fixed, edits involving X mirrors should use 2.79b.**\n", "Edge Sliding with subdivision surface is extremely laggy (caused by \"bad\" UVs and UV Smoothing in the modifier)\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 527.56\n\nBroken: version: 3.4.0\n\nSo I've been modelling a raptor the last week and I noticed that viewport performance in edit mode was extremely laggy with subdivision surface modifier enabled. I checked the face count and it was just 150k faces which is not a lot at all. It's definitely a modifier problem because when I apply the subd modifier the lag goes away completely even when sliding twice the amount of edges around. I thought that was maybe a hardware issue so I opened task manager and while sliding edges with subd on I get like 4 fps with severe stuttering to top it off. All that while the cpu was at 10% usage and 5% gpu usage with memory barely touched (6gb used out of 64gb). I have nothing else running in the background like an ativirus or music app, I have the latest version of windows 10, latest gpu and chipset drivers, latest bios etc... I even have gpu subdivision enabled without autosmooth to make sure it actually works properly. I even run an sfc/scannow to check for other kinds of issues but nothing changed. Subd in edit mode however is not only slow when sliding edges and vertices but also in adding loop cuts or insetting faces. It is extremely frustrating so it would be great if you could fix this. I know edit mode was not made to handle massive polycounts but 150k is again not a whole lot and definitely comfortable to run on my PC. Here is the file to see for yourselves.\n\n- Open attached .blend\n- Execute Edge slide/loop cuts/inset faces\n[Broken Subd.blend](Broken_Subd.blend)\n\n", "Knife Tool: incorrect interaction with Array modifier when the Merge \"First Last\" option is set\nOperating system: Win10\nGraphics card: nVidia 1070\n\nBroken: 2.80 (sub 44), branch: master, commit date: 2019-02-14 04:22, hash: caa8e7ff2894\nWorked: not sure\n\nWhen the Merge First Last option is set for an Array modifier, it seems like hidden, virtual geometry is picked up while using the knife tool to cut across the surface of the primary shape.\n\nThis results in an improper cut.\n\nBad in-progress cut:\n\n\nBad confirmed cut: \n\n\nGood cut -- with the First Last option turned off:\n\n\n[array-bug.blend](array-bug.blend)\n- Create a simple shape and array it in a radial fashion so that the first/last vertices of the shape are eligible for merge\n - Or just load the attached .blend\n\n- Attempt to make a cut across the main shape with the Merge First Last option set\n- Notice that the cut line is picking up some hidden edges somehow -- perhaps due to triangulation?\n- Confirm the cut and notice it's still incorrect and pretty weird", "render lagging behind tiles\nOperating system: win10\nGraphics card:rts 2070 super\n\n2.81\nwhen rendering, the picture doesnt appear for a while after the render tile already finished rendering that region, not sure if its a bug. didn't see anything like this online. here is a video\n\n\n[bug (2).blend](bug__2_.blend)\n\n- Open file and render\n\n[gnVknluHJZ.mp4](gnVknluHJZ.mp4)", "CUDA misaligned address error when rendering the image\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1660 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 536.40\n\nBroken: version: 3.6.0\n\nThe error occurs when rendering in a scene. It only occurs on the gpu and consistently occurs on optix and cuda. The scene starts rendering and gets to about 20/40 samples before the error occurs and it completely stops rendering. This happens both in viewport and render modes. Blender cannot be closed normally or go into viewport mode after the error occurs. It just crashes\n\nThe GPU is not running out of VRAM, Task manager shows blender barely uses 3GB VRAM before running into the error even though the gpu has 6GB\n\n`Misaligned address in CUDA queue copy_from_device (integrator_shade_volume integrator_queued_paths_array)`\n\n\n\n1. Open the scene\n2. Render image\n\nI have tried updating drivers, rolling back drivers, using previous versions (all the way back to 3.4)\n\nTest file: view?usp=drive_link", "Blender 3.5.0 Win11 RTX3090 Fatal Crash on activating Cycles\n@deadpin thanks for replying.\n\nDriver Version: 31.0.15.3161 (NVIDA 528.24) DCH Win1164 installed\n\nCycles now working without issue thank u.\n\n_Originally posted by @Jaymi-McManus in /blender/blender/issues/106953#issuecomment-923094_\n\n>>>>>\n\nUpdate: \n\n1. Closed the ticket too soon, after adding some simple color, specular, metallic, changes in BSDF and then navigating between layout and other tabs, fatal crash occurs again\n\n2. Updated NVIDIA drivers to 31.0.15.3161 (NVIDIA 531.61) DHC / Win11 64 the issue still persists\n\nUpdate logs attached." ]

[ "subdivision surface modifier, broken.\nOperating system: Windows-10-10.0.22000-SP0 64 Bits (it's actually windows 11)\nGraphics card: NVIDIA GeForce GTX 1650/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 512.15 (certified driver)\n\nBroken: version: 3.1.2\nWorked: 3.1 \n\nmodifying a mesh breaks display of subd modifier.\n\nadd a subd modifier to any mesh\nmodify mesh structure (extrude, cut, anything that modifies number of vertexes)\nmesh breaks\n\n[broken subdiv.mkv](broken_subdiv.mkv)", "Assigning multiple materials to object breaks subdivision in Blender 3.1.2\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 1080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.71\n\nBroken: version: 2.83.20\n\nApplying 2 Materials to an object when using mirror and subdivision causes the geometry of an object to break.\n\nBased on Default startup.\n\nCreate Sphere Primitive.\nDelete Half the faces\nSelect some Faces and Apply a Material (Default)\nSelect some different faces (with some overlap) and apply second material (Default with edited colours)\nApply Mirror Modifier so it looks like a sphere again.\nApply Subdivision Surface Modifier.\nObserve Geometry Breaks\n\nLoading file in an older version of blender (2.8) the issue cannot be recreated.\n", "Regression: GPU Subdivision with mirror/Array modifier shows artifacts\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 511.65\n\nBroken: master, 3.1, `d3b1cce400`\nWorked: `52be06301257`\nCaused by 993839ce85\n\nWhile GPU subdivision enabled, subdivision modifier shows artifacts in edit mode.\n\n\n**Steps to reproduce**\n- Open Blender.\n- Add material slot for default cube.\n- In edit mode, assign second material slot at least one mesh.\n- Add mirror modifier on the cube, then subdiv.\n\n**Details**\nThis issue happen when conditions below are met:\n- GPU subdivision is enabled.\n- Mirror modifier is in the first stack of modifiers, Subdiv is in the second.\n- 'Display modifier in edit mode' enabled. (default)\n- Two or more materials are assigned.\n\nWorkaround:\n- Move subdivision modifier to the first position in the modifier property.\n- Disable GPU Subdivision.\n\n**Files**\n[2022-03-29_13-52-43.mp4](2022-03-29_13-52-43.mp4)\n[system-info.txt](system-info.txt)\n[BF_GPU_subdiv.blend](BF_GPU_subdiv.blend)", "Mesh with modifiers and multiple materials glitches heavily in edit mode\nOperating system: Windows-10-10.0.19043-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3090/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 512.59\n\nBroken: \n- version: 3.1.2\n- Blender 3.1.2 official, Windows\n- Blender 3.2.0 beta, Windows\n\nWorked:\n- Blender 2.92 official, Windows\n- Blender 2.93 official, Windows\n- Blender 3.0 official, Windows\n\nSee this animated GIF, which views the glitching:\n\n\nWhen you try to edit the mesh with multiple materials and modifiers, it glitches a lot.\n\n- I have the latest of this date NVIDIA STUDIO drivers (downloaded in 2022-05-03)\n- Open Blender 3.1.2 official, for example\n- Load factory settings\n- Subdivide the cube with CTRL+2\n- Convert the mesh to mesh or apply the subdivision modifier\n- Delete half of the mesh in front view\n- Put mirror modifier to it\n- Subdivide again with CTRL+2\n- Define two different materials to the mesh (yellow and blue, for example)\n\nOr: Open the .blend file below, go to edit mode and simply try to move some of the faces with G.\n\nAs soon as you put the second material in, the mesh starts to glitch in edit mode. Especially when you move vertices or extrude faces.\n\nI think it's very critical to get this fixed, it looks very unreliable even though might be only cosmetic.\n\n[edit_mesh_with_multiple_materials_glitching.blend](edit_mesh_with_multiple_materials_glitching.blend)\n\n", "Glitchy subdivision modifier result\nOperating system: Windows-10-10.0.19043-SP0 64 Bits\nGraphics card: Quadro RTX 3000 with Max-Q Design/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 512.78\n\nBroken: version: 3.1.2\n\n\n\nOpen the attached file and enter edit mode.\n\n[Rough concepts.blend](Rough_concepts.blend)\n\n", "Mirror + Subdivision Surface Glitch\nOperating system: Ubuntu Linux 20.04\nGraphics card: Nvidia GeForce GTX 1060 6GB (510.54 drivers version)\n\nBroken: 3.1.2 (Snap Version)\nWorked: 3.0.1 (Snap Version)\n\nOddly enough, a weird graphics glitch appears in 3.1.2 in most cases when combining Mirror and a Subdivision Surface modifier after it and switching from object mode to edit mode. It's like the whole thing gets corrupt to the bone with random black lines appearing all over the place. I've asked someone to test this on Windows with the same Blender version and they were not able to reproduce it. Oddly enough, if there's another modifier after Subdivision Surface (solidify for example) the error won't appear.\n\n\n\n\n\nBased on the default startup or an attached .blend file (as simple as possible).\n\n1. Cut the mesh in half.\n2. Add Mirror modifier.\n3. Add Subdivision Surface modifier after it.\n4. Switch between object and edit modes a few times and see the error happening while in edit mode.", "GPU subdivision breaks when another modifier and another material slot is used in edit mode\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: Radeon (TM) RX 480 Graphics ATI Technologies Inc. 4.5.14802 Core Profile Context 22.2.3 30.0.14029.5006\nAlso broken on driver version 22.4.1 (\"Graphics card: Radeon (TM) RX 480 Graphics ATI Technologies Inc. 4.5.14800 Core Profile Context 22.4.1 30.0.15021.1001\")\n\nEarliest broken build: 3.2.0 Alpha\nLast working build: 3.2.0 Alpha\n\nAlso broken on 3.2.0 Alpha and working on Release 3.1.2.\n\nGPU subdivision produces holes or various incorrect triangles; just artifacts in general.\n\n- Add some mesh (preferably something more complex like a cylinder or monkey so the issue is more obvious)\n- Add mirror or array (though bevel also works, and other modifiers likely also break), then add subdivision below them\n- Add 2 material slots, it's not necessary to assign any materials to them at all, just assign the 2nd slot to some face/s\n- Observe breakage in edit mode only (change the select mode to get different artifacts most of the time)\n\nTurning GPU subdivision off of course fixes this.\n\nVideo: [2022-04-05_21-31-45.mp4](2022-04-05_21-31-45.mp4)\nExample file: [subdivBug.blend](subdivBug.blend)\n(Simply go into edit mode on the object)\n" ]

[ "Dopeheet (in the Grease Pencil submodule) take account the transform keys of the GP object wen you press jump to previous/next key frames.\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: AMD Radeon (TM) Pro WX 7100 Graphics ATI Technologies Inc. 4.5.14802 Core Profile Context FireGL 21.Q4 30.0.14011.2006\n\nBroken: version: 3.4.0 Alpha\n\nDopeheet (in the Grease Pencil submodule) take account the transform keys of the GP object wen you press jump to previous/next key frames.\nIn this Grease Pencil submudule only the GP drawing keys should be taken into account.\n\n**Steps to Reproduce**\n- Open .blend file\n- Select Grease pencil mode in Dope Sheet\n- Jump to next keyframe \n[#100538.blend](T100538.blend)\nHere a quick video with the issue:\n [2022-08-21 15-13-26.mp4](2022-08-21_15-13-26.mp4)\n\n\n\n", "Save Copy - Does not maintain state of packed + edited images\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.71\n\nBroken: version: 2.93.4\nWorked: Never?\n\nWhen using the 'Save Copy' function (from File menu or via op: wm.save_as_mainfile(copy=True)), packed images that have not been individually saved since their last change do not have those changes maintained in the saved copy.\n\nCreate an image, set it to 'Packed', edit it in some way. Now use the 'Save Copy' function to create a copy of the blend file. Open the new copy and your edits to the packed file will be missing.\nWorkaround is to save all images before using the Save Copy function. However the current behavior is not intuitive and makes it easy to lose changes when creating backup copies of your working file.\nThere is also the case where you may want to have your current edits ONLY in the saved copy and not save them in the current mainfile, which currently is impossible to have happen. \n\n", "After setting a shortcut for the \"Auto Keyframe\" button, there is an exception in the status display.\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 512.96\n\nBroken: version: 3.2.0\nWorked: none\n\nSee below\n\nUse right click->assign shortcut key\nCreate a shortcut for the auto keyframe button e.g. ctrl+alt+n\nThen test the shortcut in a different viewport and you will see that it looks like nothing is happening.\nBut,**The auto keyframe state is actually changed**, but the button display is not refreshed, and the button dispaly state is changed only when the mouse moves to the timeline window.\n\n", "Snap vertices with \"Closest\" is not acting as expected\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 432.00\n\nBroken: version: 2.83.3\nWorked: 2.82a at least. Didn't check any 2.83 releases. \n\nSnapping to vertex when moving with \"Closest\" works as if \"Center\" is chosen when moving several vertices at once.[cubicle.blend](cubicle.blend)\n\n- In the attached file, go to edit mode with the selected cubic shape.\n- 3 vertices already selected, try to move them down on Z axis with snap to vertices, \"Closest\".\n- The expected behavior is that the vertex at the bottom among the 3 selected vertices, should be at the same position of the vertex I'm snapping to.\n\n#46892 (Snapping closest vertex with Pivot point set to Individual Origins)\n\n", "Not possible to add keyframes to HSV values.\nOperating system: Linux-5.15.76-1-MANJARO-x86_64-with-glibc2.36 64 Bits\nGraphics card: Mesa Intel(R) Graphics (ADL GT2) Intel 4.6 (Core Profile) Mesa 22.2.1 & NVIDIA GeForce RTX 3050Ti Laptop GPU 520.56.06\n\nBroken: version: 3.3.1\n\nIt is not possible to add keyframes to the HSV values of any color picker.\n\nOpen a new file. Go to the material tab of the default cube. Right-click in any of the HSV values of the Base Color picker tool.\n\n", "Mirror select doesn't select appropriate vertices where there are overlapping verts\nOperating system: Windows 10 Pro\nGraphics card: GTX 1070 + GT710 for extra outputs\n\nBroken: 2019-03-10 22:28\nHash 6fd11a21f5c5\n\n\nCtrl-Shift-M (or using the menu to select mirrored) doesn't select all mirrored verts where there are overlapping verts.\n\n\n[63Stingray.zip](63Stingray.zip)\n\nOpen blend file, hopefully it'll already be in edit mode with the appropriate vertices selected.\n\nShould look like this;\n\n\nFrom there select mirrored from either the select menu, or Ctrl-Shift-M, doesn't matter, and enable the following settings;\n\n\n\nObserve incomplete selection;\n\n\n\nThe area's of incomplete selection are here, where the overlapping underbody verts are in the same location as the verts for the body;\n\n\n\n", "Mirror UI in does not update immediately when a shortcut is assigning \nOperating system: Windows-10-10.0.16299-SP0 64 Bits\nGraphics card: NVIDIA GeForce 940MX/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 471.11\n\nBroken: version: 3.2.0\nWorked: for sculpt mode it used to work in blender 2.80-2.92, broke in blender 2.93 and up. \n\nAssigning a shortcut to the mirror widget in any mode does not immediately trigger a ui update when pressed.\n\n- Enter sculpt or any other mode\n- Assign shortcut to the mirror x widget, v for example since its unassigned by default \nUI does not update immediately when pressing the shortcut, mouse cursor need to hover over the ui to trigger the update, this used to work in 2.80-2.92, but since 2.93 and up the ui does not refresh after pressing the shortcut. \nVideo of the issue:\n[Blender 2022-07-02 14_30_21_Trim_Trim.mp4](Blender_2022-07-02_14_30_21_Trim_Trim.mp4)", "VSE autokeying transforms in preview dont respect keyingsets\nOperating system: Linux-5.13.0-0.rc6.45.fc35.x86_64-x86_64-with-glibc2.34.9000 64 Bits\nGraphics card: NVIDIA GeForce GTX 970M/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 495.44\n\nBroken: version: 3.3.0 Alpha\nWorked: never since autokeying came in ffd1a7d8c8\n\nVSE autokeying transforms in preview dont respect keyingsets\n\n(this report was split from #98015 (VSE strip transforms do not animate correctly and persist after clearing the animation) since multiple issues came up in that report)\n\n[VSE_keyframing.blend](VSE_keyframing.blend)\n\n- open .blend\n- create a keyframes for Position X/Y on frame 1\n- for **Position X** only: {key RMB} Add To `Keying Set`\n- in Scene Properties, Keying Sets panel, verify only `transform.offset_x` is in the Keying Set\n- in the Timeline, Keying popover, verify that that Keying Set is used\n- turn ON autokeying\n- in the Timeline, autokeying popover, enable `Only Active Keying Set`\n- go to frame 10\n- translate the image in the `Preview` again and confirm the translation\n- notice keyframes are created for **Position X as well as Position Y** (only Position X should be keyed though)\n\n\n\n", "The bridge tool fails in a situation with multiple rows of faces facing each other\nWin10 Pro 64 bit; AMD Ryzen 2700X; Nvidia GTX 1060 6GB; 32 gigs RAM\n\nBroken: hash d1be30f (buildbot build May 4).\nWorked: never\n\nIn a situation where you have multiple rows of faces that face each other across empty space (with other areas where they face away from each other). The bridge tool (even with*loop pairs* checked) is unable to give priority to the direction the normals are facing. The result of this is that instead of getting nice bridges among the multiple rows of cubes, the tool produces faces that overlap and have zero area. As hinted, the*loop pairs* checkbox in this case does not do anything and the result is the same.\n\nOpen the .blend, invoke the bridge operator (the result is the same whether or not \"loop pairs\" is selected.\n\n**blend**\n[IffyBridge.blend](IffyBridge.blend)\n\n[IffyBridge.blend](IffyBridge.blend) (updated blend)\n\nAn additional note; I submitted this issue years ago, but it was never fixed and I couldn't find my old report. If the bug is not going to be fixed, then just close this as a todo item or with a \"won't fix\" tag.\n", "Gpencil: dopesheet selection issues [clicking in summary deselects corresponding ID channel, e.g. breaking \"Jump to keyframe\"]\nBroken: version: 2.92.0 Alpha\nWorked: n/a, tested with 2.91 and didn't work\n\nJump to keyframe command & shortcut doesn't work anymore when you duplicate a keyframe.\nDoesn't matter what amount of layers there are.\n\n1) Draw a keyframe\n2) Draw a nother keyframe at another time = works\n\nDuplicate 1) keyframe = doesn't work, please see the video\n\n\n[gp_duplicate_keyframe_jump.mp4](gp_duplicate_keyframe_jump.mp4)\n\nReason why this happens: clicking in the summary will deselect the corresponding greasepencil ID channel [on which the \"Jump to keyframe\" operator relies]\nClicking in the summary should not deselect every channel there is, but keep the channel representing the selected greasepencil object selected as well [same as for other object types such as meshes].\nThis goes hand in hand with selecting a grease pencil object in the 3DView or Outliner not selecting the corresponding ID channel [this should also be the case (same as for other object types such as meshes)]", "Rigid Body Constraints break when copying its empty object\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce RTX 2080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 446.14\n\nBroken: version: 2.90.0 Alpha\n\nCopying a rigid body constraint via CTRL+C / CTRL+V breaks it permanently (you have to remove and re-add the constraint to make it work again).\n\n- Create a simple rigid body constraint setup. You can use this file to start: [bug_rigidbodyconstraints_copy.blend](bug_rigidbodyconstraints_copy.blend)\n- Select all objects\n- CTRL+C to copy\n- CTRL+V to paste\n- Move the new objects a bit to avoid clipping, and play the animation - the active object just falls down\n- Fixed by removing and manually re-creating the constraint on the Empty.001 object\n\nThis does not happen when using SHIFT+D to duplicate.\n\n---\n\nSeems like the dependency graph loses the connections between the constraint and the rigid body objects after copying.\n\nWorking (before copy): \n\nBroken (after copy): ", "Keep Alt Key not work in the F-Curve properties\nwin10 64bit, gtx1060\n\nBroken: 2.79a 8928d99270f\nWorked: (optional)\n\nKeep alt key for multiple objects not work in F-curve property ( N panel )\n\nOpen my .blend file, when I select multiple curves and keep alt key to twice the Interpolation or Easing option, it's just worked in one cure,not for all selected, if this is bug, please fix, it very useful for lots curves, else I must one by one to change those, thanks.\n\n\n\n[bug.blend](bug.blend)", "Scene pre/post handlers are called for each main loop iteration\nTitle says it pretty much everything. just execute this in py console:\n\n```\ndef cb(sce): print(\"foo\")\n\nbpy.app.handlers.scene_update_post.append(cb)\n```\n\n…and you'll get endless 'foo' lines in your console, even when Blender is hidden and all.\n\nIssue is, main loop calls (through `wm_event_do_notifiers()`) `BKE_scene_update_tagged()`, which unconditionally fires those pre/post scene update handlers.\n\nNot quite sure what's to be done here really, not familiar at all with this area of the code.", "View -> Lock Time to Other Windows does not maintain scale when CTRL+TAB between two animation editor modes.\nOperating system: Manjaro Linux\nGraphics card: nVidia\n\nBroken: (example: 2.80, edbf15d3c044, master, 2018-11-28, as found on the splash screen)\nWorked: (optional)\n\nView -> Lock Time to Other Windows does not maintain scale when CTRL+TAB between two animation editor modes. \n\nExpand open the Timeline editor, now adjust the horizontal scale of the view (squeeze or expand), then Ctrl+TAB to swith to Graph Editor mode. Notice the scale does not stay in sych as it does when they are separate windows. ", "Wrong status bar keymap for release confirms and right click select\nOperating system: Manjaro x64\nGraphics card: HD5770\n\nBroken:\n2.80, 81ea815dcb6, blender2.8, 2018-12-12\nWorked: 2.79b\n\nWhen grabbing something(object, vertex, etc) by right clicking on it and then releasing/clicking on the same button the action is not performed, object doesn't return to its initial position. \nHowever if you use G or a Tool to grab instead of mice click Cancel by Right clicking option works.\n\nRight click on a Cube(in 2.79 thats enough to start moving an object but not here here you have to hold the button) AND HOLD the button\nDrag mice somewhere to move the Cube\nRight Click again to Cancel movement\nNothing happens.\n[blender_bug-2018-12-13_22.18.53.mkv](blender_bug-2018-12-13_22.18.53.mkv)\nI'm also attaching video file for it to be more clear." ]

[ "Animation. Keyframes. Copy - move to the current frame.\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1060 6GB/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 466.77\n\nBroken: version: 2.93.0\n\n[Animation. Keyframes. Copy - move to the current frame for replacing the previous one. If you copy keyframes and move them to the current frame it won't work properly. All will seem good but the last frame will play an animation relay on the previous parameters. ]\n\n\n[bug.blend](bug.blend)\n\n\n\n" ]

[ "Pixel Smearing in Compositor when using image node (image node lacks interpolation setting [in comparison to texture nodes])\nOperating system: Windows 10\nGraphics card: NVIDIA Quadro RTX 4000\n\nBroken: 2.79b , v2.83.8\nWorked: none known\n\nWhen an image is brought into the compositor and mixed with the scene, the result can be \"half pixel smeared\" depending on the even/odd pixel relationship between the image and the scene. It appears that some interpolation is happening on the image node's result in order to reconcile it with the center of the scene size, but this is undesirable as it breaks images that are intended to be pixel-perfect. \n\nAttached is a very simple, lightweight example .blend file (2.8) that contains 4 small example images (these images have dimensions 16x16 pixels, 15x16, 16x15, and 15x15).\nThese images exist as image nodes in the compositor.\nRender the scene, and then move the viewer node to the various outputs indicated to see how the resulting images have been smeared.[PixelError.blend](PixelError.blend)\nWhen these images are mixed with the scene, the resulting pixels from the small images will be smeared if there is a mismatch between the even-ness and odd-ness of the number of pixels in the image, and the number of pixels in the scene.", "Timeline scrubbing is laggy when Preview tab is open in Material Properties\nOperating system: Win10\nGraphics card: RTX3070\n\nBroken: 3.0\nWorked: N/A\n\n\n\nHaving the little preview tab open in Material Properties when scrubbing the timeline causes heavy lag, noticeable even on the default cube file. \n\nThe material can be a mesh material or a grease pencil material, but the result is the same.\n\nVideo example: watch?v=1Fw6Y-XzMyE\n\n", "Realtime compositor: Texture node output textures with artifacts\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2060 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 531.61\n\nBroken: version: 3.6.0 Alpha\nWorked: n/a\n\nTexture node support was introduced in 151a53110c\n\nIn realtime compositor, Texture node outputs textures with artifacts.\n\n\n\nIt happens with most procedural textures I tested. This does not happen when a custom image is used, or when Blend texture is connected from the Value output (instead of the Color output) of the node. (With other textures it happens with both outputs.)\n\nThis does not happen when running Blender with a single thread (`blender -t 1`)\n\nFile for quick test: [TextureNodeTest.blend](attachment)\n\n", "Allow OpenGL image drawing in UI colour space\nSince 7ad802cf3a images are stored with premultiplied alpha and in scene linear space. This produces the wrong colours when used for drawing UI elements.\n\nExample from the Blender Cloud texture browser, with the post-7ad802cf3a drawing on the left and the expected colours on the right:\n\n\n\n@brecht proposed adding a new shader that can be used from Python and that draws the textures in the correct colour space.\n", "Light Nodetree support\nWhen rendering shadowmaps, we could run a postprocess shader (using the depth texture to retreive position) to output a Mono or Poly-chromatic intensity texture.\n\nThis would enable the use of custom light shaders and IES light profiles.\n\nThis will have to use another sampler since the shadow texture use another format.\n\nEstimate : 2 weeks", "Add Image/Sequence to video editing Squencer: number of images are truncated\nOperating system: Win10\nGraphics card: GeForce GTX 1050Ti\n\n**Blender Version** \nbroken: 2.8 RC3\nworked: 2.8 build from ca 1 month ago\n\nWhen you add images by drag 'n drop in de video sequencer, the number of images in the resulting strip is truncated aaround 27. \n\n(1) Open Blender\n(2) from the splash-screen choose: video editing\n(3) Open a windows explorer and select ca 100 jpg files;\n(4) use drag n drop to drop the files in the sequencer;\n\nYou then have a strip of just 27 frames where the selected number of files is expected. This used to work in the beta's", "AgX has banding-like issues on hyper saturated gradients\nBroken: a9053f7efbe8b25429b11bccd679dbe81ea60e3f (introduction of AgX)\n\n\nWhen using the sRGB display device and the AgX view transform, fully-saturated colors can exhibit banding-like issues. It doesn't appear to be actual banding, but subjectively feels similar.\n\nThis issue was discovered by @eyecandy while testing AgX at Blender Studio, and the concern is that for more cartoony styles that tend to have a lot of single-color materials, this could manifest as visually objectionable artifacts.\n\nHere's a comparison between Filmic and AgX that highlights the issue:\n\n| Filmic | AgX |\n|----|----|\n|  |  |\n\n\n1. Load  into Blender (e.g. as a texture on a plane with an emissive material).\n2. Set the color management display device to sRGB.\n3. Set the color management view transform to AgX.\n\n", "GPUOffScreen example give strange \nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2080 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 512.96\n\nBroken: version: 3.1.2\n\n\n\n1. copy the code from here gpu.html#rendering-the-3d-view-into-a-texture\n2. run it \n\n", "Attribute node (Type: View Layer) doesn't update in real time in Cycles\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 511.09\n\nBroken: version: 3.4.0\nWorked:\n\nWhen using an Attribute node (type: View Layer) to get the position or angle of an object, the Attribute is not immediately updated in Cycles if Viewport Shading is Rendered.\nBut It works correctly in EEVEE and Material Preview.\n\n - Open attached file\n - Set Viewport Shading to Rendered\n # Move Empty and Rotate Sun\n\n[AttributeNodeBug.blend](AttributeNodeBug.blend)\n[221212_022211.mp4](221212_022211.mp4)\n\n", "VSE: Cycles displayed as Rendered in Scene strip makes Blender hang - and other Scene strip bugs\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce RTX 2060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 462.59\n\nBroken: version: 2.93.0\n\nThe Render modes of Scene Strips are either missing or broken:\n- When Cycles is set as render and Rendered set as display, it'll make Blender hang.\n- Wireframe and Rendered doesn't work with any Render. \n- The only thing working is Eevee with either solid or Material Preview. \n- Workbench and Cycles settings seems to just use the Eevee render, since these are producing the same results in Solid and Material Preview.\n\n{[F10177251](scenestrip.gif),size=full}\n\nTest file:\nCycles is set up as render in the 3D Footage scene. In the Master Edit scene, switch display mode to Rendered. Blender will hang.\nTry the various other display modes and renders(by changing the render in the 3D Footage scene), to see wireframe and rendered doesn't work with any render. \n[multicam_3d_c.blend](multicam_3d_c.blend)\n\n\n", "Move cache visualization to draw overlay\nCurrenlty the cache visualization isn't updated during playback, only when the screen is redrawn.\nThis needs an optimization of the drawing of the cache lines.", "EEVEE and Workbench missing animatable properties\nOperating system:\tWindows 7\nGPU:\t\t\tRadeon Vega 64, Divider version: 19.12.2\nCPU:\t\t\tIntel Core i7-5960X\n\n\nWorking-ish*:\t\t`blender-2.79b-f4dc9f9d68b-windows64` (official release) \nBroken:\t\t\t`blender-2.80-f6cb5f54494e-windows64` (official release) and later\n\n\nEEVEE and Workbench's render properties (`Properties Editor --> Render Properties`) are, for the most part, not animatable.\nThis is in contrast to Cycles which, *for the most part*, has all settings set to be animatable.\nBack in 2.79b the OpenGL \"Workbench\" Render options were supported:\n\n\n\nSome of the panels like Film, Simplify and Colour Management (which exist in all render engines) work as expected from 2.7X to 2.8X.", "Inconsistencies between 3D Viewport and Viewport Render Image using the GPU Shader Module\nOperating system: Ubuntu 18.04, Windows 10\nGraphics card: GeForce GTX Titan X, GeForce GTX 1080 Ti\n\nBroken: \n2.83.0\n2.90.0 Alpha\nWorked: Never\n\nRendering the `3D Viewport` with `3D Viewport/View/View Render Image` (i.e. OpenGL rendering) does not render elements drawn with the [GPU Shader module ](gpu.html).\n\nThe problem is illustrated in the following example, which shows the `3D Viewport` (left) and the rendered result of `View/View Render Image` (right).\nThe points drawn with the GPU shader module are neglected during OpenGL rendering.\n\n\nExcecute the [3D Lines with Single Color ]] example of the [[ https:*docs.blender.org/api/current/gpu.html | GPU shader module documentation ](https:*docs.blender.org/api/current/gpu.html#d-lines-with-single-color).\nThen, render the results with`View/View Render Image` as shown in the image below.\n\n\nThe lines visible in the 3D Viewport (left) are missing in the rendered result (right).\n ", "VDB sequences are rendered blocky here and there ( issues with Volume Displacement Modifier )\nOperating system: Windows 10 \nGraphics card: RTX 2070 CUDA\n\nBroken: blender-3.0.0-alpha+master.ef5a362a5b24-windows.amd64-release.zip\n\n\nBlender renders these blocky frames here and there during the actual render. The VDBs seem to be fine, the same botched render frames look fine in the viewport render. These anomolies are happening only during the actual render. See the attached image (middle right image) for what is happening please.\n\nThe same sequence renders without any issues in Houdini.\n\n(Cycles)\n\n\n\n\nHere is an actual movie render (with Eevee) that shows how horrible this issue is.\n[borkedframes_0001-0200.mp4](borkedframes_0001-0200.mp4)\n\n\n[#88966.blend](T88966.blend)\n[steam.4.0001.vdb](steam.4.0001.vdb)\n\n- Open .blend file\n- Render sequence with CUDA\n\nIssue should appear within 30 frames\n", "Ambient Occlusion node with image texture\nOperating system: macOS 12.6.8 – 16GB\nGraphics card: Apple M1 8/4-4\n\nBroken: 3.6.3 RC\n\nWhen I mix an image texture (here a PNG) with Ambient Occlusion node, its not working, or results a weird render at least.\nIn the render-preview the image is disappearing, but in the final render is there but partially and broken (i think).\n\nAttached a screenshot and a very simple blend file.\n\nI hope I am the idiot and doing something wrong, but after a full week I didn't find the answer to this strange behavior.\n\n**Edit: it seems an earlier resolved problem is back:** [#94050](94050)\nIf I switch to CPU-rendering, its working fine... :(\n\n" ]

[ "Unable to use image sequence shader in Eevee.\nOperating system: Windows 10 i7 4200k\nGraphics card: GTX1060 6GB x 2\n\n**Blender Version** 2.8 Beta\nBroken: Image Sequencer in Eevee\n2.80 ba1f178c1c8 2018-12-14 23:33\nHas never worked\n\nI have a UV Unwrapped plane and can display a single image on it easily. If I try to add multiple images the material shows as the pink 'No Material'. If I try to add\nthe Multiple Images or Image Sequence shader I get a brief display of an error and I cannot add the shader.\n\n\nAdd an image to your material (which will work) and then make that image multiple images.\n\nPlease see attached .blend file.\n\n[15th December For Report.blend](15th_December_For_Report.blend)" ]

[ "Maximum number of attributes that can be used with eevee shading needs to be documented\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1050 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 527.56\n\nBroken: version: 3.6.0 Beta\nWorked: This is the only version I have used\n\nThere are more than 20 parameters or attributes in my geometry nodes setup, and when I try to use those attributes in shader editor it doesn't update its values. Sometimes it updates when there is a viewer node connected.\n\nIts a big node tree and a lot of materials, so that's it.\n\n", "Cycles viewport unneccessary update calls\nOperating system: Ubuntu 20.04\nGraphics card: Nvidia MX150\n\nBroken: 2.83.17, 8a8c61190b06, Beta and 2.90.2, 8f21f867e715, Alpha\nWorked: 2.82a official\n\n[#76970.blend](T76970.blend)\n\n* set scene renderer to Cycles\n* configure two viewports and a shader editor in a single screen\n* set one viewport to cycles preview render\n* set second viewport to workbench texture color\n* add two image texture nodes and load them with different textures\n* by clicking on the texture node both viewports are redrawn.", "Script node with shader output crashes Eevee\nOperating system: OSX Monterey (12.6)\nGraphics card: Radeon Pro Vega 20\n\nBroken: git versions: blender-v3.2-release, blender-v3.3-release, master\nWorked: 3.2 release\n\nI'm testing the MaterialX standard surface shader. It seems to work when in Release versions, but in a debug build, it crashes when you connect the output of the shader to 'surface' in the terminal node.\n\nUsing the default scene, switch from Eevee to Cycles + Open Shading Language. Select the Cube, switch to Shading tab. Create a new material, and add a new Script node. Load the attached .oso file (to compile the attached .osl, you need to have MaterialX installed. I'm using 1.38.5). Attach the output of the script node to the 'Surface' input of the terminal. Boom! Where the crash occurs depends on whether your in Eevee preview, or Cycles preview.\n\n[out.osl](out.osl)\n\n[out1.oso](out1.oso)", "Realtime compositor: Texture node output textures with artifacts\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2060 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 531.61\n\nBroken: version: 3.6.0 Alpha\nWorked: n/a\n\nTexture node support was introduced in 151a53110c\n\nIn realtime compositor, Texture node outputs textures with artifacts.\n\n\n\nIt happens with most procedural textures I tested. This does not happen when a custom image is used, or when Blend texture is connected from the Value output (instead of the Color output) of the node. (With other textures it happens with both outputs.)\n\nThis does not happen when running Blender with a single thread (`blender -t 1`)\n\nFile for quick test: [TextureNodeTest.blend](attachment)\n\n", "Crash when rendering animation\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA RTX A4500/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 512.15\n\nBroken: version: 3.4.1\n\nBlender crash when rendering an animation\n\nIn an attempt to simplify my rendering of multiple scenes, I added an additional scene to a blend file with no objects itself. Its purpose is to be a place to put multiple *Render Layer* nodes in its compositor for all the other scenes in the file. In other words I hit \"Render\" when on that special scene and all other scenes are rendered in one go. Running this render as a single image works fine. Running render animation crashes after one frame.\n\nMy actual file has 7 scenes to be rendered. But to help demonstrate and reproduce this, I have simplified the file to just two scenes:-\n\"*Scene 1*\" is a simple scene containing a cube and a camera. Rendering this scene directly, works fine as still image or animation.\n\"*Render all*\" is a second scene that has no objects of its own (just a camera because my render farm seems to insist upon that). This scene is set to not be rendered (\"*Use for Rendering*\" unticked). However it's compositor has a *Render Layers* object which looks at *Scene 1* and which is connected to a *File output* node.\n\n\n[crash-test3.blend](crash-test3.blend)\n\nOpen file and render animation.", "Crash after copy pasting a material onto itself and switching to render\nOperating system: ArchLinux\nGraphics card: NVIDIA 3090\n\nBroken: Current main.\nWorked: Never.\n\n\nBlender crashes when copy pasting a material onto itself (via UI), then creating a new material slot, and then switching to the rendered view.\n\n1. Open a new scene in blender.\n2. Select the cube and go to its material tab.\n3. Next to the material list, on the right, click the down arrow and do Copy Material.\n4. From the same menu now do Paste Material.\n5. Add a new material slot using the + in the materials tab.\n6. Switch to rendered view.\n\nAttached a crash stacktrace taken with gdb, and main build with debug.\n\nIf it's useful, the crash was taken with a manually built blender, so it was a factory startup with Eevee as a renderer, but it also crashes when using Cycles.\n", "Mix/Add nodes breaks Eevee AO render pass if one socket is unfilled.\nOperating system: Windows-10-10.0.16299-SP0 64 Bits\nGraphics card: GeForce RTX 2070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 419.67\n\nBroken: version: 2.92.0, 2.93.0, 3.0 0efb627bbd\n\nIn Eevee, any material which contains an add shader node will be broken in the AO pass, if second socket is *invalid*, leaving the material either entirely black or white.\nThis is annoying if you're creating entirely black materials by leaving a socket empty.\n\n- Add a mesh, give it a material.\n- Put either an an add shader node or a mix node into the material somewhere, but only fill one of the input sockets\n- Enable AO with Eevee, and look at the AO render pass in render view. In 2.92, the object will be completely black, in 2.93 it will completely white.\n\n----------------\n- Render the scene with the file given.\n- For the second case, render by swapping the socket inputs of the \"Add Shader\" node.\n[T88779_mine.blend](T88779_mine.blend)\n\n\n\n\n\nAdditional example images.\n(2.93)\n[F10155389](image.png)\n[F10155391](image.png)\n\n(2.92)\n[F10155385](image.png)", "Texture properties not working with OSL enabled and unpacked textures\nOperating system: Linux-5.11.7-arch1-1-x86_64-with-glibc2.33 64 Bits\nGraphics card: Mesa DRI Intel(R) HD Graphics 520 (SKL GT2) Intel Open Source Technology Center 4.6 (Core Profile) Mesa 20.3.4\n\nBroken: version: 2.92.0\nBroken: version: 2.93.0 Alpha\n\nI have a texture with RGB channels set to 1 and alpha set to 0. When trying to use this texture with OSL, the alpha property (at least) of the image texture node does not work when the texture is unpacked. I set it to channel packed, expecting the color data to be always white. The problem is that when the texture is unpacked, the color data is black.\n\n1. Open the attached file.\n2. Set viewport shading to rendered.\n3. The plane shows as white (as it should).\n4. Unpack the texture.tga file.\n5. Toggle between preview and rendered viewport shading to refresh.\n6. The plane now shows as black.\n7. Pack the texture again.\n8. The plane shows as white again.\n\n[bug.blend](bug.blend)", "Cycles: Baking to an externally loaded image fails to update the image buffer results in Cycle's Rendered Viewport Shading\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2080 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 512.59\n\nBroken: version: 3.3.0, 2.93\nWorked: probably never\n\nBaking to an externally loaded image will not changed the baked image as displayed in the Viewport Shading's Rendered Mode.\nThe newly baked image buffer is shown correctly in the image editor and (EEVEE's) Material Preview, but wrong in Cycles's render view.\n\n- Open the attached .blend file.\n- Open the attached baked image in the image texture node.\n- Use it to bake any type of pass to the loaded image.\n- Observe that the image looks unchanged in the Rendered View by connecting the image texture node to the material output.\n[#98063.blend](T98063.blend)\n\n[2022-05-12 09-12-20.mp4](2022-05-12_09-12-20.mp4)", "Scene update bug\nSystem Information\nwindows7 64\nASUS ATI Radeon HD6870\n\nBlender Version\nBroken: (All)\nWorked: (None)\n\n\n\n\n\n\n\n\nopen this file, select plane and try to select for this plane any mesh - all this meshes don't have users, but was saved, try to select material - materials was saved, but don't have users, \nwe can revert, close and open again many times - materails and meshes will be saved.\ntry to add gruop instance - we will have all this groups is empty.", "EEVEE crash while rendering animation, cycles works.\nOperating system: Win10\nGraphics card: GTX1070MAX-Q\nCPU : i7-6700HQ\nRAM : 32GB DDR4\n\nBroken: 2.90\nWorked: -\n\nBlender crashes w/o any errors when I started to rendering animation in Eevee. \nIt happens on 2 laptops and it always happen in 154 or 155 frame while rendering.\nIn attachment you can download the project which gets this error. \nWhen I use cycles rendering it works perfect.\n\n[#80740.blend](T80740.blend)\n\n- Open file\n- Render frames 154 and 155\n", "Eevee - visual artifacts during irradiance volume baking\nWindows-10-10.0.16299 64 Bits\nGeForce GTX 1080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 419.35\nBlender 2.80 (sub 51)\n\n[irradiance_volume_bake_artifacts.blend](irradiance_volume_bake_artifacts.blend)\n - Open attached file\n - Switch to rendered mode (Eevee)\n - Press \"Bake Indirect Lighting\"\n\nYou should see squares flickering:\n\nVideo:\n[Desktop 2019.03.27 - 09.18.07.01.avi](Desktop_2019.03.27_-_09.18.07.01.avi)", "Momentary interface freezes while EEVEE rendering causes progress bar malfunction\nOperating system: Windows 10, 64 bit\nGraphics card: Intel HD Graphics\n\nBroken: 2.80, 2.81\n\n**Rendering in EEVEE has been causing momentary complete interface freezes for each frame**, which caused the progress bar on the frame render to malfunction and jump very quickly from one value to another, thus giving momentarily true but useless information (note that the progress bar never moved to an incorrect position, it just froze). One thing that assures me this is a bug is **minimizing all Blender windows causes the progress bar visible on the minimized icon to work properly** (but unfortunately one cannot see what frame it's on).\n\nOpen the default blend file.\nRender the animation. The interface on 2.81 has frozen for each frame.", "Eevee preview nodes with drivers performance drop Round 2\n[NodesPerformanceDropRound2.blend](NodesPerformanceDropRound2.blend) this is extend of previos task #71214\nSystem Information\nOperating system:Windows 10\nGraphics card: GeForce GTX 1060\n\nBlender Version\nBroken: Blender 2.81 Beta aff6446e064f and 2.82 Beta 092deb88b074\nWorked: 2.8\n\nShort description of error\nComplex shader with large amount of math nodes and with drivers slow down performance\nin 2.8 60 fps\nin 2.82 30 fps\n\nExact steps for others to reproduce the error\nCreate material with 30 node groups contains 30 math nodes, add driver to shader. Set viewport to material preview and watch performance droping", "Cycles: Bump node doesn't generate anything in certain conditions\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3090/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 522.25\n\nBroken: version: 3.3.0\nWorked: all versions from 2.79 to 3.4.0 alpha are affected, there is no working build\n\nI'm trying to generate a normal map using Bump node and color output of the env map as height map, it doesn't work on Cycles while it works on Eevee. Seems like the \"reflection\" output of Texture coordinate node is broken, but it works fine when I'm using another outputs though\n\nOpen the project, then click on viewport Rendered mode button - it'll start Cycles which for some reason can't generate bump map, then switch to Material preview mode - it'll start eevee which works as expected\n[bug_report.blend](bug_report.blend)\n\nEdit: typos and bad english grammar fixed, I believe" ]

[ "Image sequence as texture not previewed and not rendered.\n\nOperating system: Windows 7 Ultimate sp1\nGraphics card: Palit nvidia 1060\n\nBroken: 2.80.37\n\nImage sequence as texture not previewed and not rendered either in Cycles and Eevee. In particular Image texture preview (and render) became violet whenever switched from single image to image sequence in texture node.\nIn movie mode only first frame is previewed and rendered but not auto refreshed.\nThank you very much." ]

[ "Intel Mac Metal: UV Display Stretch Angle Fault Colors\nOperating system:Mac OS 13.3.1\nGraphics card:Radeon Pro Vega 48 8GB\n\nBroken: 3.5.0 - 3.5.1 - 3.6.0 alpha\nWorked: NA\n\nUsing Intel Apple, with GPU Backend Metal enabled\nUV Editing, when you turn on Display Stretch: Angle\nCorrect angles will show faulty colors.\nWhen I switched to OpenGL and restart.\nThe Stretching shows properly as it should.\nI tested using default blender with no addons. Clean install with Default Cube and the error persisted on the versions noted above.\n\n\n\n- Activate blender with GPU Backend: Metal\n- Use a default Cube, within the UV Editing tab\n- In Overlays menu, choose Display Stretch, and pick Angle.\n- Select all the Faces. The colors will appear faulty and extremely inaccurate.\n\n\n", "Multires spikes\nOperating system:\nGraphics card:\n\nBroken: Blender 3.00\nWorked:\n\n\nBased on the default startup or an attached .blend file (as simple as possible).\n\n1. Sculpt while having multires modifier on\n2. Move \"sculpt\" level up and down\n3. Go into edit mode? I am not sure if I did do this, but I read that some people did do it to cause the artifact to appear. \n4. Artifact appear\n5. Unable to fix, without restarting [Handle safeguard v2.blend](Handle_safeguard_v2.blend)\n\nI can read online that this seems to have been a bug for long, but I couldn't find a fix. \n\nCan I use multires on my objects? What exactly caused this, and how can I fix it? \n\n", "Remesh Modifier in Voxel mode produces surface artifacts when Adaptivity enabled\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce RTX 2080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.71\n\nBroken: \n - 2.83.9\n - 2.91.0 Beta\n - 2.92.0 \nWorked: never\n\nRemesh modifier produces artifacts on mesh which look like a result of bad triangulation.\n\n1. Open attached blender file\n2. Open modifier panel and select testMesh\n3. Change adaptivity value from 1 to 0 cm and back\n4. Observe black shading artifacts \n[remesh_bug.blend](remesh_bug.blend)\n", "Trembling when selected cube in edit mode\nOperating system: MacOS Sierra - Versione 10.12.6\nGraphics card: NVIDIA GeForce GTX 680MX 2048 MB\n\nBroken: 2.83.0 (2020-06-03), 2.90 (2020-06-12)\nWorked: 2.82a\n\nA strange flicker happens when I select the faces of the cube.\n\nI select the cube.\nI change the visualization in Edit Mode.\nI set select mode to Face.\nI select the faces of the cube but they do not remain selected instead, a strange flicker of the object occurs. Here a video example:\n\n[blender_283_bug.mp4](blender_283_bug.mp4)", "Grease Pencil shimmers in the viewport and renders\nOperating system: Windows 7 Ultimate\nGraphics card: Nvidia Quadro 2000\n\n2.92.0\n\nI have a video below, If you see carefully the lines are shimmering, Is there any way to avoid that? watch?v=aBIJbTTiLlU \nAttached a blender file as well.\nI tried \n- increasing samples of viewport \n- increasing anti-aliasing of grease pencil. \n- different render systems 'eevee' and 'cycles'. \n- Tried all possible file exports but the problem is in the viewport itself.\nShould I try denoising it? although that didn't work so well either.\nThe problem just isn't going way. Is it my system? Could you help?\n\n[room fix.blend](room_fix.blend)", "False color look gives incorrect values\nOperating system: Windows-10-10.0.19043-SP0 64 Bits\nGraphics card: GeForce GTX 1080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 431.07\n\nBroken: version: 3.3.0 Alpha\n\nThe False color in Blender is indeed different from the Filmic master on GitHub:\n\n\nThe testing EXR I used (background is 0.18 middle grey):\n\n\n- Load image in Image editor\n- Enable View as Render in side panel\n- Set False Color as View Transform in Color Management settings", "Sculpting Has Unexpected behavior when Saving work\n**Broken**: 2.83.4, 2.90 beta 97dc370c50ed\n**Worked**: 2.82\n\n\nGlitches in the non-active mesh after sculpting the active mesh, saving, and updating the viewport.\n\n\n[Ogre.blend](Ogre.blend)\n\n* Sculpt\n* Save\n* Refresh viewport (e.g., continue sculpting)\n\nThe artifacts should be visible immediately.\n\n**Original message**\n\nWhen I am sculpting and I go to save my work, after a successful save I get weird artifacts in the mesh that I havent made edits on in sometime. And it seems to be the same mesh every time. This has happened over 5 times so i figure its just a visual bug.\n\nGet 2 seperate meshes ready to sculpt, then sculpt on one and save your work. This is all i did to see the bug arise.\n\n\n", "VSE: Separate images does not handle resolution correctly\nOperating system: Windows 11 Home 64 bit\nGraphics card: NVidia GeForce 2060 RTX SUPER\n\nBroken: 3.0.1\n\n- Add image sequence to VSE where each image has different resolution\n- Separate images to discrete strips\n\nCheck resolution of separated strips - they are all same, but they should not be.\n\n\n\n\n\n\n\n", "Render error not shown in VSE\nOperating system: Windows 10\n\nBroken: 2.90 8956e9e5f6c0\nWorked: Unknown\n\n**Reproduction**\n1. In a scene's Compositing workspace, enable Use Nodes and delete all the nodes.\n2. In another scene's VSE, add the above scene as a strip.\n3. Render\n\n**Description**\n\nF12 on the outputless scene throws an error.\n\nThe VSE shows the checkerboard pattern even if there's another strip (e.g. solid black) below the outputless scene strip and does not throw any error. The outputless scene strip should either not interfere with the layers below it or show an error indicating it doesn't have an output.", "Sequencer: Driver on any strip property does not invalidate its cache [prefetch frames]\nOperating system: Darwin-18.6.0-x86_64-i386-64bit 64 Bits\nGraphics card: Intel Iris OpenGL Engine Intel Inc. 4.1 INTEL-12.9.22\n\nBroken: version: 2.83.9\n\nSequencer: Driver on any strip property does not invalidate its cache [prefetch frames]\nThis can be any driver on any driven strip property:\nDoes not matter which property is the driving property (could be x Location of default cube as well)\n\n\n[#82776.blend](T82776.blend)\n- Open file\n- `driven` strips opacity is driven by Cube 's X location\n- change Cube 's X location --> driven strips opacity value changes, but the cache is not invalidated\n\n---\n\nOriginal report:\n\n Driver on `wipe.effect_fader` doesn't invalidate cache for driven wipe sequences. It's only invalidates the cache of the currently displayed wipe. This leads to display errors.\n\nNB: The issue is the same for any other parameter (e.g angle, blur...)\n\n- Open this file: [Bug_Wipe_Driver.blend](Bug_Wipe_Driver.blend)\n- There are two wipes, the first is the driver, the second one has the `effect_fader` value driven. Prefetch is enabled.\n- Move the `effect_fader` value of the driving wipe, its cache is reloaded, but not the cache of the driven one.\n\n", "Blocky artifacts in various render passes when rendering Mantaflow volumes in Cycles\n## Simplified Report\n\n- Open the attached file and set the VDB path to the attached VDB volume.\n- The VDB volume file is the result of a Mantaflow simulation generated in Blender using Quick smoke.\n- Render the scene and look at the shadow pass.\n- The shadow pass contains blocky artifacts as shown in the attached image.\n\n[blockyVolumeShadowPass.blend](blockyVolumeShadowPass.blend)\n\n[fluid_noise_0060.vdb](fluid_noise_0060.vdb)\n\n\n\n## Original Report\n\n[system-info.txt](system-info.txt)\n\nBroken: 3.1, 3.4.0 Alpha, branch: master, commit date: 2022-09-27 23:42, hash: 5beaecb33e74\nWorked: most likely never\n\nRendering imported vdb generated with Mantaflow in Cycles gives blocky artifacts in various passes. \nMost visible artifacts are in shadow pass and denoising albedo. Less visible can be seen in volume direct and indirect.\n\nThis bug is particularly nasty as the artifacts are mostly invisible in beauty pass. \nHowever when volume have high density and is illuminated with intense light the artifacts starts to appear in combine pass.\nArtifacts in denoising albedo pass can ruin seemingly good volume render.\n\nI wasn't able to replicate the artifacts in EEVEE.\n\nHere is how the bug looks with Quick smoke on default cube (scene in attached blend file):\n\nshadow pass\n\n\ndenoise albedo pass\n\n\nAnd here is an example how this bug might affect production scene:\n\nshadow pass\n\n\n\ndenoise albedo pass\n\n\n\nvolume direct\n\n\n\nvolume indirect\n\n\n\nbeauty (cropped)\n\n\n\n\nReproduce render artifacts:\n- open first attached blend file\n- hit `F12`\n- observe shadow pass\n- change `Render Layer` output in `Compositor` if needed\n\nRecreate vdb (optional):\n- open second attached blend file\n- (optional) in `Properties` -> `Physics Properties` -> `Fluid` -> `Cache` set path for cache\n- in the same panel hit `Bake All` and wait until bake is completed\n- navigate to cache directory and open `noise` folder, pick one file (better to pick one from the end where the smoke have more complicated shape)\n- import vdb to the first file and place it so the camera can see it\n\n[volume_bug_render.blend](volume_bug_render.blend)\n[volume_bug_sim.blend](volume_bug_sim.blend)", "VSE: adding scene stripe resets animation of colour stripe with mask\nOperating system: Linux-5.13.0-41-generic-x86_64-with-glibc2.33 64 Bits\nGraphics card: AMD RENOIR (DRM 3.41.0, 5.13.0-41-generic, LLVM 12.0.1) AMD 4.6 (Core Profile) Mesa 21.3.8 (git-813ee839be)\n\nBroken: version: 3.1.2\nWorked: none tested\n\nI have added a colour channel and applied a mask to it. I have created and animated a clock as UI element.\nWhen I add the clock as scene stripe to the scene with the masked colour stripe, the mask of the colour strip is not applied correctly.\n\nI have no idea what really causes the issue or how I could reproduce it. See the attached blend file and video files.\n\n**Notes**\n- When the clock strip is muted in the VSE overview, the problem does no longer exist. So I guess it's some weird interaction between the mask and the scene strip\n- The bug can be solved when you create a new scene and add the colour strip as scene strip and on top of that the clock strip. It works, but performance is incredibly bad.\n\n**Files**\n - The actual output: [bug 01.mp4](bug_01.mp4)\n - How it should look like: [bug 02.mp4](bug_02.mp4)\n - blend file: [bug.blend](bug.blend)", "Visual bug when using strenght tool in 3D space left for Tool Panel\nOperating system: Windows 10 64Bit\nGraphics card: Nvidia GTX 960M\n\nBroken: 2.80 RC\n\nVisual bug in Sculpt mode, Vertex Paint, Weight paint and Texture paint\n\n1. Go to Texture Paint mode\n2. Set strength tool to 1.0 (shift F)\n3. Use strenght tool hotkey (Shift F) in the narrow 3D space left for the Tool Panel\n\n", "Blender Interface Does Not Respect Windows Colour Profiles\nOperating system: Windows 10 Home 1909\nGraphics card: Nvidia GTX 1080\n\nBroken: 2.83.0 on Windows\nWorked: 2.83.0 on OSX\n\nBlender's interface shows incorrect colours on high colour gamut monitors in Windows 10. Any monitor over 100% sRGB is showing colours to be incorrectly vibrant in every panel, including image viewer, 3D view and compositing.\nThis is believed to be due to Blender failing to reference the OS's installed ICC/ICM profile, or the interface entirely lacking any colour management that respects monitor profiles.\nSince OSX has system-wide colour management, this issue is not apparent. Windows does not have system-wide colour management, and apps must perform their own referencing of the installed ICC/ICM profile.\n\nOpen Blender on a high-gamut monitor in Windows 10. Compare to sRGB 100% monitor or OSX.\nError initially produced on a Dell UP2516D high-gamut monitor.", "Multiple Frame Rates/ Playback/ VSE scene strips\nOperating system: windows 10\nGraphics card: rtx 2060\n\nBroken: 2.90.0\nWorked: Never\n\nI'm using the VSE and I have 2 scene strips from two different scenes inside the VSE.\nThe first scene strip references a scene that is set to 12 FPS.\nThe second scene strip references a scene that is set to 24 FPS.\nThe 3rd scene is where I'm using the VSE, and this scene is set to 12 FPS.\n\nThe problem is, all the scene strips are playing back at 12 FPS by default in the scene where I'm using the VSE. In other words, the scene that is 24 FPS is playing back at 12 FPS.\nIf I'm not mistaken, this wouldn't happen with real footage. Two recordings with two different FPS dropped into a project would play back at different speeds.\nWhat I'm trying to do is have a video at 12 fps, and the put a 24 fps scene strip in the VSE so that i can put it in slow motion and have it play back with the same quality. \n\n- Open attached file\n- Run VSE\nNote that the cube shown on each strip has the same speed in the animation.\nIt was expected that one was twice as fast than the other.\n[_VSE scene strips.blend](_VSE_scene_strips.blend)" ]

[ "The alternate rows in the Sequencer are green (macOS)\nOperating system: Darwin-19.6.0-x86_64-i386-64bit 64 Bits\nGraphics card: AMD Radeon Pro 455 OpenGL Engine ATI Technologies Inc. 4.1 ATI-3.10.18\n\nBroken: version: 2.92.0 Alpha\nAfter f7223d5f72\n\nThe alternate rows in the Sequencer are green.\n\n\n\nSimilar to #81827 (MacOS lines that should be thick are green instead)" ]

[ "History filled with \"Double click to rename\" when user changed index of a UIList\n\n\nBroken: version: 3.2.0\nWorked: Never\n\n- open Blender\n- add particle systems\n- change particle systems index\n- see how your history got vanished by useless flags\n\nGifs: \n\n\nI believe in order to fix this bug,\na `SKIP_UNDO` flag need to be implemented at a property level \nexample: \n`my_uilist_idx : bpy.props.IntProperty(options={'SKIP_UNDO'},)`\n\nThis issue is not only linked with UILists indexes, but also with any properties used strictly GUI\n\nsee the blender dev thread where we talk about the general issue of using properties for interface\n4\n\n\n\n", "Keyframing Stretch property in IK won't work properly - iTaSC Solver\nDoesn't work in all versions 2.79b, 2.83, 3.00\n\nWhen using iTaSC solver for an armature, you can't keyframe or use a driver for the bone stretch property. If you do it will be buggy and won't work as expected.\n\nThe Stretch property I'm talking about\n\n\n1. Make a a small armature with iTaSC Solver\n2. Add IK Bone Constraint\n3. Animate the stretch property whether in the bone constraint or the bone property\n4. Keyframe your rig and see that it's not behaving properly\n\n**note: I know it may seem incomprehensible but if you click constantly \"Update Dependencies\" in the drivers panel you will see the expected result**\n\n[Bug report.blend](Bug_report.blend)", "Material Properties: Shortcut assignment for the \"New\" button does not relate to the interface.\nOperating system: Windows-7-6.1.7601-SP1 64 Bits\nGraphics card: GeForce GTX 980/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 425.31\n\nBroken: version: 2.81 (sub 16)\nWorked: (optional)\n\nThe shortcut for the big \"New\" button in Material Properties does not work for the button, which suggests giving a new material to the active slot. Instead it calls bpy.ops.material.new() and silently creates new materials in the data. This can also be tested with Quick Favorites.\n\n\n\n\n", "Geometry Nodes: Modifier input properties don't cause update for screen statistic\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 4090/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 528.24\n\nBroken: version: 3.4.1\n\nModifier input properties of geometry node don't cause update for screen statistic.\n\n1. Create object with geometry nodes.\n2. Add any Mesh Primitive node in modifier node tree.\nPut mesh primitive on group output. Updating the detail of a primitive affects the statistics of the scene.\n3. Put group input dummy socket in detail of a primitive node to create connection.\nChanging the details through the modifier settings is not reflected in the statistics.\n", "Modifiers after physics ones shouldn't invalidate cache\nOperating system: macOS-10.14.6-x86_64-i386-64bit 64 Bits\nGraphics card: AMD Radeon Pro 560 OpenGL Engine ATI Technologies Inc. 4.1 ATI-2.11.21\n\nBroken: version: 2.93.0\n\nIn current Blender, any change on modifiers invalidates.\nModifiers *after* physics modifiers won't affect the result of physical simulation, therefore they shouldn't invalidate the cache. \n\n\n1. Create a plane\n2. Give it Cloth and Subdivision modifiers, in this order\n3. Set the end frame to 10\n4. Play animation until the end\n5. Stop animation\n\nNow the simulation is cached.\n\n6. Change the viewport division level on Subdivision modifier\n\nNow the cache is invalidated.\n\n", "Improvement ideas for using Driver Context Properties to access scene custom properties\nThe #105132 pull request introduces access to properties of the current scene and view layer via an implicit rather explicit reference.\n\nThe existing implementation is perfectly fine for built-in properties, but I have some additional ideas aimed at making it easier and more convenient to allow scene-global custom settings via custom properties.\n\nBasically, since the properties have to be added by the user, it would be convenient to support the case when it doesn't exist. It also may be nice to allow setting overrides at several different levels of abstraction.\n\nThus:\n\n* Allow an optional default value that would be returned by the variable if the referenced property doesn't exist, instead of failing with an error.\n\n* Allow fallback from Layer to Scene, and then World: one World can be used by multiple Scenes, which can have multiple Layers.\n\nSince this all only makes sense for *custom* properties, maybe it can be a new mode specifically aimed at them, e.g. some kind of \"Layer/Scene/World Custom Property\", which has options for the custom property name (without the `[\"...\"]` quoting), channel index (for vectors), and default value.\n\nThese are things missing for the new driver things to fully cover the features of the View Layer mode of the Attribute material node.", "Animated particle emitter refuses transforms within simulation range\nOperating system: Win8.1x64\nGraphics card: 3xgtx580\n\nBroken: 2.79, 2.82\n\nAn animated object with a particle system it is not properly transformed in the frames within the simulation range.\nIn **Blender 2.79** the object jumps back to it's animated location on confirmation on hotkey or widget transformation.\nIn **Blender 2.82** the object doesn't even transform, but its particles does.\n\nTemporarily setting particle system Subframes parameter to 0 seems to be a workaround, eliminating the behavior.\n\nHere is a gif showing the bug in 2.82 and 2.79:\n\n\n- Open attached file\n- (G)grab move the emitter\n[particle-subframes.blend](particle-subframes.blend)\n\nIn blender 2.82 you must change the visibility of objects. If you want to skip this step, use this file:\n[particle-subframes-283b.blend](particle-subframes-283b.blend)", "Visual glitches With adaptive domain enabled when object is duplicated or transformed\nOperating system: windows 10\nGraphics card: 3080ti\n\n**Blender Version** 3.3 lts\n\nVisual glitches With adaptive domain enabled when object is duplicated or transformed\n\n- Open default scene\n- Add quick smoke on cube\n- enable `adaptive domain`\n- play animation\n- move domain around or duplicate it (snaps back to original place)\n- play animation again (will move to transformed position)", "Keyframed group inputs bound to index not property\nOperating system: Windows 10\nGraphics card: GTX 1070 (Laptop)\n\nBroken: 2.80.0 009dbc2bc9fb\n\nKeyframing a property on the group input stack, maintains the keyframe at the stack index rather than the mapped property. This means when the ordering of group inputs is changed the keyframe is maintained at the previous position/index rather than moving with the initial property.\n\n\nBased on the default startup\n\n1. Create node group with multiple inputs\n2. Keyframe 2nd property\n3. Push this group input up or down the input stack (while inside node group)\n4. 2nd property will stay keyframed rather than moving up/down stack with the correct property.\n", "Blender freezes with big custom data\nmacOS 10.14\nBlender 2.92, 2.93.0, \n\nI found if I store big data (like 5 millions of points) then Blender will freeze a bit. This is not a huge bug but it would be nice not to freeze blender.\n[tmp1.blend](tmp1.blend)\n\n\n\nFor example, a console looks like this if I do \"print(test2)\". And there is no freeze in the console.\n\n\n- Run the script\n- Hover over the custom properties section. (try to rotate the cube to observe the freeze, beachball may take time to show up)/\n\n\n\nMemory usage rises and falls on every hover.", "Library-Override Not able to edit \"Strip time Keyframe\" value in sidebar panel in graph-editor\nOperating system: Windows 10 Pro v.21H2\nGraphics card: NVIDIA GeForce GTX 1060 3GB\n\nBroken: Version:3.3.3, hash:8d94aeb604fa, branch: master, date:2023-01-17\n\nI think is a similar case to this \" blender/blender#81965 \", but more specifically for animated striptime. The striptime key-frame in Graph-Editor is not editable in sidebar and values are grayed out.\n\n\n1. link a collection/character/object\n2. object > relation > make library override\n3. animate\n4. stash action to create NLA track\n5. animate strip time\n6. try to change a keyframe's value in graph-editor's sidebar Fcruve > active keyframe.\nopen the StripTime_KeyLocked.blend file, try to change the selected key's value in graph editor's sidebar\n\n\n\n", "Geometry Nodes Inputs `is_animatable` Property Incorrectly Returns False\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3090/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.59\n\nBroken: version: 3.3.0 Alpha\n\nGeometry Nodes inputs' context menus ([WM_MT_button_context ](bpy.types.Menu.html#extending-the-button-context-menu)) incorrectly show the inputs' `is_animatable` property as `False` when the property does accept keyframes (see screenshot, attached).\n\n\n - Open the attached .blend file (attached, below)\n - Run the script in the Text Editor\n - Right-click the Input property of the Cube's Geometry Nodes Modifier\n - Open the System Console to see the `is_animatable` property of `context.button_prop` is `False`\n[is_animatable.blend](is_animatable.blend)", "Wrong editor type saved in workspace\nOperating system: Win11\nGraphics card: AMD Radeon (TM) Graphics (Included Graphics)\n\nBroken: 3.5.1\nWorked: N/A.\n\nMy properties editor is glitched and keeps resetting to nonlinear animation editor.\n(When saving file and reopening)\n\nIdk how to repreduce it, but here are steps for my blender file:\n1. Switch editor type to properties editor.\n2. Save file.\n3. Close blender .\n4. Open the saved file.\n5. Properties editor switched to nonlinear animation editor.\n6. (When repeating 1-4 with all objects deleted, then its working normally without the switch at restart).\n7. Viedeo: EDgFd6MIzjM.\n\n", "Workbench render engine does not update the Material Slots panel preview\nOperating system: Linux Fedora 35\nGraphics card: Intel\n\nBroken: 3.2.0 Beta, branch: master, commit date: 2022-05-10 06:30, hash: 11aa237858d4, type: release\n\n\nThe aspect of the preview sphere in the Material Slots panel is not updated.\n\n\n1 - In a factory default startup file, select \"Workbench\" as render engine\n2 - Select the \"Cube\" Object\n3 - Open the Material Property Panel, in Viewport Display set a different color (diffuse_color)\n4 - The aspect of the preview sphere in the Material Slots panel is not updated.\n\nThe users of my BlenderFDS (www.blenderfds.org) addon keep asking me: why?\nA correct UI feedback is important especially for new users, and I presume this should be quite easy to fix. \n\nThank you very much, I am grateful for your work.\n\n[system-info.txt](system-info.txt)", "Grease Pencil datablock property 'use_onion_skinning' doesn't work\nOperating system:\nGraphics card:\n\nBroken: 3.4\nWorked: \n\nThe datablock-level property \"use_onion_skinning\" on the Grease Pencil datablock doesn't seem to do anything. Toggling the property with Python doesn't affect the visibility of onion skinning.\nWorking as expected, this property will be useful when drawing among multiple Grease Pencil objects.\n\nOpen attached blend, toggle use_onion_skinning in the Python console: bpy.context.object.data.use_onion_skinning = False [onion_bug.blend](onion_bug.blend) " ]

[ "properties made with python (bpy.props) do not update with drivers and keyframes animation\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce RTX 2080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.71\n\nWhen users animate our bpy.props properties with keyframe or drivers, no update signal will be sent \n\n\n\n- create a python bpy.props.FloatProperty, with an update function (that print something for example)\n- Show this property in a panel \n- use #frame to animate this property then run animation \n- no update will be sent \n\nto test it, you can use the blend below, where i wrote a little script that register a new property and a panel\n\n[keyframe & driver don't update props.blend](keyframe___driver_don_t_update_props.blend)\n\n- related task? (Someone told me it's related, it doesn't looks like it, as this is not the same kind of properties)\nT63793\n\n- Potential problem? \n*\"RNA is not notified because drivers are probably working with pointers (modifying directly values)*\"\n\n- related blenderstack question\nproperties-bpy-props-dont-update-with-keyframes-and-drivers\n", "Custom Prop Does Not Fire Update When Driven By F-Curve\nWindows 7 x64 8GB\nnVidia 460gtx 1GB\n\n2.69 official release\n\nCustom Prop Does Not Fire Update When Driven By F-Curve\n**My guess is that only a draw() event is being issued and not and update() behind the scenes.**\n\n\n[custom_node_update_issue.zip](custom_node_update_issue.zip)\nI have attached a custom node AddOn and a BLEND file that uses that AddOn.\nDrop the AddOn into your AddOn folder and activate \"test_node\" under *User Preferences*. Then open the BLEND file.\nYou should see a series of three Float nodes.\nPut Blender and the Console side by side so you can view the AddOn output.\n\nThe first node is animated. **Scrub the Node Property Value** directly to observe that the node does correctly send an update to all nodes downstream of it's output. It also generated print statements in the console as events are fired.\n\n**Scrub the Timeline** and you will see that while the animated node and the next companion node do update, the third node does not. This is because the *update=updateProperty* def is not being fired when it is driven by an f-curve. This should happen. When I place a custom property under the object panel and animate it I do get an update event fired by the property. Also you will notice that not a single print statement is generated by the AddOn when you scrub the timeline even though the property value does change. This is the result of what I am calling a \"fake\" update event. \n\n```\nvalue = bpy.props.FloatProperty(name = \"Float\", default=DEFAULT_FLOAT, update=updateProperty)\n```\n\nThe*update()* event of the node itself is not firing when you scrub the animation or render the animation via F12.\n\nThe only reason we see the second node update when we scrub the timeline is because there is some kind of special code in Blender that forces this \"fake\" update to only the first node connected. This \"fake\" update event should be removed from the system.\n\nMy guess is that only a draw() event is being issued and not and update() behind the scenes.\n\nI know this is kind of a complex issue and please point out if my code is wrong. I hope this is enough info to describe the bug\n\n\n" ]

[ "VSE (Video Sequence Editor) scene strips have very slow playback in both Material preview and Rendered mode.\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1050 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 531.41\n\nBroken: version: 3.6.0 Alpha\n\nAdding the 3d scenes to the VSE and turning Material Preview or Rendered mode are making the playback very slow.\n\n1. With the default scene opened create a second scene.\n2. Switch to the Video sequencer.\n3. Add a first scene to the timeline.\n4. Switch \"Scene strip display\" to Material preview or Rendered.\n\nThank you.", "Automated performance testing\nSee here for documentation on the performance testing framework:\nPerformance\n\nThis system is separate from [Open Data ]() by design, for developers rather than users.\n\n**Tasks**\n\n- [ ] Automate running on buildbot\n - [ ] Nightly builds tracking performance over time\n - [ ] GPU support\n - [ ] Publish HTML results on builder.blender.org/download\n- [ ] Tests\n - [x] Cycles\n - [ ] Animation playback\n - [x] Geometry nodes\n - [ ] Object and mesh editing operators\n - [ ] I/O", "GPU subdivision improvements\nThis is a list of TODOs and ideas for improving GPU subdivision support ([D12406](D12406)).\n\n**TODOs**\n\n- Investigate if read and write operations can be made more coherent.\n- Investigate if compute dispatch size can be improved, to avoid too many calls.\n- Remove some useless OpenGL initialization from OpenSubDiv\n - We use our own patch evaluation shader, however we are still forced to compile the one from OpenSubDiv. This will require to implement our own GLComputeEvaluator.\n - `GLVertexBuffer` also initializes the OpenGL library. We can avoid this by using our own vertex buffer wrapper.\n- Compressed data structures for reducing memory usage\n - Using compute shaders forces us to only deal with ints and floats. However, this is using more memory than necessary (e.g. for normals).\n - This requires to do manual bit packing in the shaders, we need to pay good attention to alignements and data types (int vs. uint).\n- Use multithreading for CPU tasks when applicable\n - We still process some data on the CPU (e.g. polygons to materials map, edit mode flags). Although they are based on the coarse mesh, we could use multithreading to speed things up if needed.\n\n**Ideas**\n\n- Heuristic to disable GPU subdivision for small meshes, if they are hindering performance\n- The mesh wrapper used for on the CPU might not need all of the data to be subdivided (e.g. we do not need vertex colors for snapping), we could potentially speed up CPU subdivision by only subdividing what we need.\n- Potentially evaluate points on the GPU and copy back to the CPU side, this would require threads to have their own GL context.\n- See if OpenSubDiv data structures can be shared between the CPU and GPU code (e.g. the stencils buffers) so that we do not have to create duplicate versions when the CPU side mesh is requested.\n\n**Vulkan**\n\n- OpenSubDiv does not yet officially support Vulkan. The Pixar team seem to already have something working internally (1153). It is not clear when this will be released.", "Faster Animation Playback\n**Status:** Discussing Designs\n\n---\n**Team**\n**Commissioner:** @Mets @Hjalti\n**Project leader:** @Jeroen-Bakker\n**Project members:** @Sergey\n\n**Description**\n**Big picture:** Animators should be able to work in an interactive environment.\n\n**Non functional requirements:**\n\n* Animation playback in the viewport should play at constant framerate at 24fps.\n* `?` Motion curves should be update-able in realtime\n* Playblast shouldn't take much more than regular viewport playback.\n\n**Work plan**\n\n\n**Milestone - Animation Playback/Render Performance**\n\n**Status:** Execution\n**Time estimate:**: \n**Design:** #73429 (Approach Faster Animation Playback)\n**Engineer plan:** Will be created per improvement.\n**Related Tasks**\n* {icon circle color=red} #75134 (TaskScheduler Improvements)\n* {icon circle color=red} #75120 (DrawManager: Improved Task Scheduler)\n* {icon circle color=red} #75208 (Dependency Graph Evaluation Optimizations)\n* {icon circle color=yellow} F-Curve Evaluation Optimizations\n* {icon circle color=yellow} Profile and optimize modifiers (some suggestions are archived in #75207)\n* {icon circle color=yellow} Motion Path Performance\n* {icon circle color=yellow} Optimize VSE Playback for Animators\n* {icon circle color=green} #75124 (ActionEditor: Keyframes Drawing)\n* {icon circle color=green} #75127 (ActionEditor: Add callback function to ANIM_animdata_filter)\n\nThese tasks are ordered in priority. The changes to the task scheduler has most technical impact so we should execute that task with highest priority so we have more time to fix any related issue. The Draw Manager changes rely on features that will be realized by the task scheduler. (@Jeroen-Bakker will do the developments)\n\nChanges to the dependency graph can start from the beginning. Most likely they will be executed by @dr.sybren or @sergey.\n\nThe rest (modifiers, action editor) can be executed afterwards. This is also the time to look at improving motion curves (if this is still needed). (@Jeroen-Bakker will do the developments)\n\n**Milestone - GPU Subdivision**\n\n**Status:** Initial\n**Time estimate:**: \n**Design:** To be completed\n**Engineer plan:** To be completed\n**Related Tasks**\n\n* #68996 (OpenSubdiv GPU acceleration)\n\nThe final design for subdivision (part of the modifiers optimization) needs more collaboration between the GPU/Viewport team and @Sergey to come with an realistic approach.\n\n\n**Milestone - Animation Cache**\n\n**Status:** Needs Design and engineering plan to proceed\n**Time estimate:**: To be determined after Design/EngineeringPlan\n**Design:** #73481 (Design: Animation Caching)\n**Engineer plan:** Not started\n\nAnimation Caching Mechanism: RAM/Disk cache mechanism that can cache the output of an animation. During the animation the cache is invalidated.\nThis cache can also be used for viewport playback and viewport animation rendering.\n", "Vector transform node with constant input difference in Cycles and Eevee\nOperating system: win\nGraphics card: nvidia gtx 790\n\nBroken: 2.80, 009dbc2bc9fb\n\n\nEvee and Cycles have different background rendering. It looks like Cycles has 180 rotation for the backgraund.\n[2_8_setup_1.blend](2_8_setup_1.blend)\n\n[temp1.mp4](temp1.mp4)\n\n\n\n\n- Open a file\n- rotate in the viewport\n- switch to Evee, Cycles, Evee, Cycles.", "Render progress Infotip \"Time Remaining\" is calculated incorrectly in animations\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1050/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.71\n\nBroken: version: 2.92.0 Alpha\nWorked: none - bug has been present since this feature was introduced in 2.77 (confirmed present in e.g. 2.79, 2.90 and 2.91)\n\nWhen hovering the mouse pointer over the Render progress bar in the status bar, an \"Infotip\" appears showing the **Time Remaining** and **Time Elapsed** for that render (see attached screenshot). This works correctly for a single F12 render, and matches the values shown in the render window, but when rendering an animation (Ctrl-F12), the Time Elapsed shown is the *total* elapsed time for all frames so far, not just the elapsed time for that single frame. This isn't really a problem in itself, but this total time is evidently also being used in the Time Remaining calculation, which a result becomes more and more wildly inaccurate with each subsequent frame, presumably because the calculation wrongly assumes that that single frame has needed all the accumulated time to render however far it's got.\n\nFor example, if a single uniform frame took 100s, after 10% of the first frame of an animation it would correctly show **Remaining 01:30.00**, **Elapsed 00:10.00**, as in the render window. But on the 101st frame, after 10% progress it would show **Elapsed 2:46:50.00** (100x100+10 seconds), but **Remaining 25:01:30.00** (10,010x9 seconds), wrongly implying this single frame will take over a day to finish rather than about 90 seconds!\n\nPlease note that this report is *not* about the expected inaccuracy of the estimate early in the progress of a frame, which I realise is unavoidable given the variation in render speed for different tiles within a frame.\n\nI believe a fix for this bug might be to either:\n\n - change the Elapsed time in the Infotip to be the elapsed time for just the current frame (as it already is in the render window header) so the Remaining time will then be calculated correctly; or better still...\n - when rendering an animation, change the Infotip times and the progress bar to show the overall progress of the entire animation (i.e. frames rendered divided by total frames) instead of the individual frame progress (i.e. tiles rendered divided by total tiles). This would give a useful indication of the total (estimated) remaining time to render the full animation.\n\n\n\n\n\n - Open any version of Blender 2.77 or later, with the default startup file or any other scene.\n - Depending on the speed of the test hardware and the complexity of the scene, optionally select Cycles render engine, the CPU device, and/or increase the sample count, so that the F12 render time is at least e.g. 20-30s (this is because very short renders make it difficult to see the problem, as the progress bar jumps straight to e.g. 98%)\n - Now start rendering the scene as an animation using Ctrl-F12 (the default of 250 frames is fine).\n - While rendering, hover the mouse pointer over the Render progress bar in the status bar to show the popup Elapsed/Remaining Infotip (this applies to Windows 10 - I don't know if this is similar in Linux or MacOS)\n # Notice that the time values in the Infotip agree with those in the render window for the first frame, but with each subsequent frame the Elapsed time accumulates and as a result the estimated Time Remaining becomes more and more inaccurate.\n \n\n\nI realise this is not a serious bug, but it has been long standing and is will hopefully be quite easy to fix, so it would be good if this could be included in a future version.\n\nMany thanks, Ross\n[edited to hopefully correct my maths!]\n\n\n", "Viewport Slowdown on 11th gen Intel Xe graphics\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\n| Graphics card: | Intel(R) Iris(R) Xe Graphics Intel 4.5.0 - Build 27.20.100.8935 | @tommyoliversays, @CuriousHobbyist (#93114), @Jonathan-Monsonego (#93039), @Pyer (#93505), @Charlie3521 (#93257), @ImaRealBoi (#87147)\n| -- | -- | -- |\n| | Quadro RTX 3000 with Max-Q - 4.5.0 NVIDIA 472.39 | @Stay_Here\n\nBroken: version: 2.92.0, 2.93\nWorked: 2.83 LTS (better but not perfect, just takes longer for it to start occuring, maybe an hour as opposed to around 20 minutes in 2.92).\n\nWorking on a model slowly brings viewport down to single digit framerates.\nIt seems to be a twofold issue. Whenever you hit spacebar to playback animation, Blender starts using up more RAM. After working for about 15 minutes, 2.92 RAM usage increased from ~500MB at launch to over 3GB in the same scene while idle. Rotating the viewport also spikes GPU usage to over 50%, which immediately drops back to single percent digits when idling.\n|  |  | \n| -- | -- | -- |\n| Memory at launch | 15 mins after | Spikes rotating viewport\n\n- moving verts around over time or\n- hit spacebar to playback animation or\n> In #93114#1254781, @Stay_Here wrote:\n> The most likely scenario in my case is when two or more Blender(s) are launched. For example, open Blender.exe twice, then load two different characters in the two Blender windows. Let's say, character A and character B are loaded in the Blender window A and window B, respectively. After that, I spend some time on sculpting character A, and use the character B as a reference character. In this setup, Blender window B is seldom touched or manipulated, since it is just for reference. \n> \n> After about 10 mins to 2 hours on sculpting character A on the window A and open YouTube to listen some music, I jump to the Blender window B to navigate and change the view to the reference character B. At this moment, the lag in the solid viewport in the window B can be quite likely to happen. However, at the same moment, navigating in the window A is still smooth and unaffected.\n\n\n[imp2.blend](imp2.blend)\n\n", "Same pattern looks different for Cycles and Eevee\nOperating system: Windows-8.1-6.3.9600-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 660 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 472.12\n\nBroken: version: 3.1.1\n\nThis is the difference\n\n\n\n\nsimplified file\n[pixels tiled1.blend](pixels_tiled1.blend)\n\n\nold file \n[pixels tiled.blend](pixels_tiled.blend)\n\nI suppose the difference in Color Ramp or Noise node.\n", "Regression: GPU Subdivision: EEVEE doesn't refresh viewport correctly with Armature modifier\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.94\n\nBroken: version: 3.3LTS, master\nWorked: 3.2.2\n\nGPU Subdivision doestn't refresh viewport correctly an object when Armature and Subdivision Surface Modifiers are used.\n\n1) Start Blender.\n2) Add Armature object\n3) Switch to \"Shading\" workspace.\n4) Select the Cube and add `armature modifier` then `Subdivision Surface Modifier`.\n5) Switch to Properties Editor to \"Modifier Properties\"\n6) Add armature in armature modifier's object property. \n7) Switch between object-edit mode\nIt works fine when \"GPU Subdivion option\" is turned OFF (Prefrences > Viewport > Subdivision > GPU Subdivision)\n\nI added \"BeforeGPUSubDBug.blend\". You can reproduce this issue by loading the blend file and doing step 7)\n", "Slow initialization in the rendered viewport after the default normal map added\nOperating system: Windows-10-10.0.19043-SP0 64 Bits\nGraphics card: Quadro RTX 3000 with Max-Q Design/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 512.15\n\nBroken: version: 3.1.0\n\n\n[slow initialization time in viewport rendering.mp4](slow_initialization_time_in_viewport_rendering.mp4)\nThe slow initialization in the rendered viewport occurs after the default normal map node was added in the Skin shader. (The Skin shader is very simple - it contains just a Principled BSDF and Material Output.) I have tried both CUDA and Optix, and the slow initialization exists in both cases after the default normal map node added. \n\n\n[Sample to test slow initialization in viewport rendering.blend](Sample_to_test_slow_initialization_in_viewport_rendering.blend)\n\nSteps to reproduce the bug:\n\n1, Download the sample file and open Blender; \n\n2, Select the mesh, and go to the Sculpting mode, and then under the \"Remesh\" tab change the Voxel Size from 0.4 m to 0.004 m, or even smaller value if your computer can handle. Click \"Remesh\".\n\n3, Then click the rendered preview. You will find that the higher the poly-count, the slower initialization speed. \n\n4, Go to the Shading mode, and remove the Normal Map, and click the rendered preview. Your initialization speed returns to normal, even if the poly-count is high. \n\n", "Fix big LIghtCache restriction\nAs reported in #78529 saving big lightcache is impossible due to blocksize restriction.\n\nFor now we restrict the lightcache size that can be written to disk but it would be better to slice the huge textures to avoid such limitation.\n\nHowever this needs some refactors in GPUTextures namely the need of a properly exposed `glTexSubImage2D` and `glTexSubImage3D`.", "Clearcoat roughness is clamped to ~0.017 in Cycles\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 4090/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 527.56\n\nBroken: version: 3.5.0 Alpha\n\nAny value for clearcoat roughness lower than 0.017 is just 0\n\n- Open attached .blend file\n- Note that at 0.018 `Clearcoat Roughness` the roughness is visible\n- Set the roughness to `0.017` (just go back 1 frame in the timeline as the value is animated in the file)\n- Note that now the roughness is now invisible\nIt's perfectly sharp at 0.17, only after 0.18 it has roughness\n[Clearcoat bug.blend](Clearcoat_bug.blend)", "Cycles renders SSS is noticeably slower in 3.4 than in 3.3 with multiple GPUs\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: 2x NVIDIA GeForce RTX 2080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 526.98 (SLI)\n\nBroken: version: 3.4.0\nWorked: 3.3\n\nCycles render images with SSS noticeably slower in 3.4 than in 3.3\n\nI noticed that some of my scenes render in 3.4 noticeably slower than in 3.3.\nAfter some analysis, I figured out that only images with SSS shaders are affected.\n\nI created a simplified test scene, this scene takes on my computer in Blender 3.3 **4:58** min and in Blender 3.4 it's **6:10** min.\n\nIf SSS is disabled, the test file renders equal fast in 3.3 and 3.4.\n\nProbably this change is unintended and might be a bug.\n\n[speed test.blend](speed_test.blend)\n\n\nTest render 3.4 (OptiX) 6:13 min:\n\n\nTest render 3.3 (OptiX) 5:12min:\n \n", "Double click list renaming not working on heavy scenes\nVista64, 2.7a, gtx470\nWhen the scene is heavy, so that interface becomes slightly laggy, double click list rename stops working, especially for fast clicking.\n\n - Add subdivision modifiers to any mesh until interface becomes laggy.\n - Add a vertex group or UV map.\n - Double click to rename - renaming is not activated.\n", "Cycles does not generate the exact same images when a scene is rendered twice\nOperating system: Ubuntu 20.04\nGraphics card: Nvidia 2070 Super\n\nBroken: master\nWorked:\n\n\nThe results generated by Cycles are not 100% deterministic.\nAs a consequence path guiding ( #92571 ) can not be implemented deterministically.\n\n\n1. Start Blender and open a scene like `monster`\n2. Render the scene with 64spp and store the result as exr image (e.g., monster-run-0.exr).\n3. Repeat 1. and 2. and save the result as exr again (e.g., monster-run-1.exr).\n\n4. Use an image comparison tool such as tev (tev) and compute the difference. \n\nYou will see that, even if both renderings were performed on the same machine, the resulting images have minor differences.\nNote: It might be necessary to scale the diff-images to see the errors.\n\nRun 0:\n\nRun 1:\n\nDiff:\n\n\n\n\n\n\n \n\n\n \n\n \n\nBased on the default startup or an attached .blend file (as simple as possible)." ]

[ "Double render times with Cycles 2.69 vs 2.70RC \nWin7\nI7-2600K\n2 x GeForce GTX 580\nDriver: 334.89 \n\nBroken: 2.70 RC\nWorked: 2.69\n\n\nHello Devs,\n\nunfortunately I have almost double cycles rendertimes with 2.70 RC.\nMike Pans BMW renders in the same time but my files has problems.\nPerhaps it has something to do with linking. I link 28 objects in the szene.\n\nRendertimes:\n2.69 Resolution: 50%HD, 100spmls, 2x GPU 26 sec, CPU 3:13\n2.70 Resolution: 50%HD, 100smpls, 2x GPU 45 sec, CPU 6:29\n\nI can´t upload the production folder because it is to big. I tried to delete all necessary stuff but it´s still to big.\nBut here is a dropbox link: \nrender%20bug.zip\n\n(By the way I won a trailer competition for the local short filmfestival\nwith the video I made with the attached file: )\n\n\n\n\n" ]

[ "Consider using more explicit names instead of the generic 'Size' value for Empties.\nWhen creating an Empty, the pop-up window specifies the `Radius` for the Empty, presumably because the shape chosen could be a sphere, arrows, or any number of different types.\n\nHowever, when viewed in the Object Data Properties, the radius is displayed as a more generic `Size` value.\n\n\n\nThis leads to confusion, especially when comparing an Empty to a Cube. In the example below, the visual scale of these two items is identical, yet looking at their `Size` tells a different story.\n\n\n\nFor context, this came up as a point of confusion for a use case where I'm using non-scaled empties as a 'holder' for position/rotation/scale while applying a set of custom properties to be exported as .glb and used in threejs.\n\nMy suggestion would be to continue the use of `Radius` as the label in the Object Data Properties, instead of `Size` since that's what is actually meant by the value.\n\n", "How to handle volumes in spreadsheet\nVolume data is not suitable for spreadsheet, maybe we shouldn't show the individual voxel information there.\n", "Frame Selected does not respect GN output if it has different type from input\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: Radeon RX 570 Series ATI Technologies Inc. 4.5.0 Core Profile Context 23.3.1.230305\n\nBroken: version: 3.6.0 Alpha\nWorked: n/a\n\nI think this is an old known issue but I can't find this documented here.\n\nFrame Selected does not respect the output geometry by Geometry Nodes if its type is different from the input type.\n\nFor example, if the input is mesh and the output is curve, Frame Selected works on the original input mesh. (The same if the output has multiple types of geometry including the original type.) If the output is mesh it respects the output geometry even if it has noting to do with the input and has a wildly different bounding box.\n\nOn the other hand, a curve object converted to a mesh by Solidify modifier is still respected as the final output mesh.\n\n(Maybe related? #104738)\n\n[FrameSelectedWithGN.mp4](attachment)\n\n[FrameSelectedWithGN.blend](attachment)\n\n", "Apply Parent inverse dose not take scale in to account\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1650/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 531.61\n\nBroken: version: 3.5.1\n3.6 alpha\n\nApply Parent inverse dose not take scale in to account while using alt+g , alt+r , alt+s to restore the coordinate\n\nnot sure if this is design intended.\n\n**Exact steps to reproduce the error**\n\nBased on the default startup\n1. create 2 cubes in same size ( display in wire frame mode for easier visual)\n2. Scale 2 cube independently in any size\n3. Parent them in either way , using any following parenting method below.\n\nParent object \nParent object Keep transform\nParent object (keep transform without inverse)\n\n*** Parent object (without inverse) works fine as intended there is no need to try on this ***\n\n4. Select child cube Apply Parent inverse\n5. Select both cube alt+g , alt+r , alt+s restore the coordinate , you will see child cube is not at the same size as parent\nbut the transform panel is displaying 1,1,1 both on the child and parent cube\n\n--------------------------------------------------------------\nrelated issue been reported \n97648\n\n", "Blender 3.6's window has 1/4 content area, 3/4 transparent area under Linux\nOperating system: Arch-based Linux, kernel 6.3.7-arch1-1\nGraphics card: Intel Arc A779\n\nBroken: blender-3.6.0-beta+v36.2ef27684c56c-linux.x86_64-release, but has happened with all 3.6 daily builds so far.\nWorked: 3.5.1, the latest version.\n\nThe content area is 1/4 of the window size when the window is first open. 3/4 of the window is transparent.\nGnome 44.2, Wayland, fractional scaling enabled (100%, 175%), dual monitors (1920x1200, 4K).\n\n\n\n\nRun Blender 3.6\n\n", "Knife Project Resolution is View Dependent \nWindows 7, x64, gtx 580\n\nBroken: 85c38e7\n\nKnife project resolution seems to be based on the camera zoom level. Artists are expecting that it wouldnt operate like this, ending up with meshes which look inaccurate.\n\n1. Open attached file\n2. knife project\n3. zoom in a lot\n4. notice how horrible it is\n5. undo\n6. knife project\n7. notice the difference\n\n[knifebug.blend](knifebug.blend)", "UI change for the Armature Modifier\nThis task proposes a small update of the Armature modifier's UI.\n\nThe issue the UI currently suffers from is that it is unclear that checking \"Preserve Volume\" actually changes the deformation method, instead of just enabling an additional option in the modifier.\nThis is addressed by replacing the checkbox with a dropdown, containing the two available deformation methods: \"Linear Blending\" and \"Dual Quaternion\".\n\nOn the contrary, a similar method cannot be used for the \"binding\" checkboxes, since those two options are apparently not mutually exclusive.\n\nMockup:\n\n\nTooltips:\n- Linear Blending: Deforms the mesh without preserving volume. Prone to candy-wrapping artifacts, but does not suffer from bulging issues.\n- Dual Quaternions: Attempts to preserve the volume when deforming the mesh. Generates bulging issues around joints, but does not have any candiwrapping issues.", "Multiple mix nodes lead to incorrect results\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 512.15\n\nBroken: version: 3.4.0\n\nI'm not good at English and don't know how to explain it, it's a stripped down part of my project. When two orange nodes have the same input value, they get a value of 0.5. But when the value is 5, sometimes 8 or 9, the correct result cannot be obtained, it seems that it is not refreshed.\n\nI provided the video and the .blend file, hope that helps.\n\n[bug.blend](bug.blend)\n[20221213_16.48_793.mp4](20221213_16.48_793.mp4)\n\n\n", "Sculpt Mode - Face Set FK, mirror X is wrong when Scale/Translate using Pose brush with Lock Rotation When Scaling\nOperating system: macOS-10.15.7-x86_64-i386-64bit 64 Bits\n\nBroken: 2.92, 3.5.0 Alpha\nWorked: Never.\n\n\n- Change to pose brush.\n- Change brush deformation mode to Scale/Translate.\n- Enable brush Lock Rotation When Scaling.\n- Try to pose the hand.\n- The other hand is flipped.\n\n\n\n\n\nThis is the file I'm using:\n\n[20221205_model_human_sculting_pose_brush_using_facecets_0001.blend](20221205_model_human_sculting_pose_brush_using_facecets_0001.blend)\n", "Normal orientation for complex selection work unpredictable\nOperating system: Windows-7-6.1.7601-SP1 64 Bits\nGraphics card: GeForce GTX 660 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.71\n\nBroken: version: 2.92.0 Alpha\n\n\nHere is simple example with cylinder. Same surface with same normals changes it's normal orientation and scale-cage orientation after rotation.\nIt is unexpected and unpredictable.\n\n[2020-11-21_14-27-41.mp4](2020-11-21_14-27-41.mp4)\n\nIt work as expected only if I deselect one of opposite faces\n[2020-11-21_14-31-14.mp4](2020-11-21_14-31-14.mp4)\n", "Switching Curve Shapekeys in Edit Mode go haywire if relative to one another\nOperating system:\nLinux\n\nBroken in 2.79 onwards\nWorked in 2.78\n\nPossibly caused by\n6f1493f68f\n5e1d4714fe\n\nShapekeys on curves do not reset properly when switching between them in edit mode, and one shapekey is relative to another.\n\nSelect the circle and enter edit mode. Repeatedly switch between shapekeys 'Key 1' and 'Key 2' while staying in edit mode. 'Key 2' will change over time from just this.\n[bug_report_shapekeys_curve.blend](bug_report_shapekeys_curve.blend)\n\n**How to reproduce the scene**\nCreate a Curve object. Add Basis, Key 1, Key 2 shapekeys. Set 'Key 2' relative to 'Key 1'. Select 'Key 1' and translate/scale it. ", "Fluid domain renders as the original object when particle mesh is momentarily disappear\nOperating system: Windows-10-10.0.19043-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1050/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 511.65\n\nBroken: version: 3.2.0\n\nIf there is no mesh generated from active particles within the domain, the frame will render with the domain as its original surface.\n\nI have an emitter object that periodically (cycles) emits fluid into the domain. At times when there is no mesh generated from particles in the domain the rendering is correct -- the fluid mesh:\n\n\n\nAt some point when there is no mesh generated particles inside the domain, the render is incorrect -- the surface coloring on the domain boundary:\n\n\n\nIn the attached file, I would know when I might start to include the Fluid Domain in the rendering, but it would be rather difficult to tell when there are no more particles if velocities or emissions are complicated with noise.\n\nHere is the [StackExchange ](fluid-simulation-gives-a-obscured-render-when-active-fluid-particles-are-absent) discussion and an example blend file.[fluid render problem.blend](fluid_render_problem.blend)\n\n\n\n", "Wrong particle distribution\nOperating system: Ubuntu 18.04 LTS\nGraphics card: NVidia Quadro K2000D\n\nversion: 2.80 (sub 75), branch: master, commit date: 2019-07-29 14:47, hash: f6cb5f54494e, type: Release\n\nThe density of randomly generated particles inside of a deformed mesh is not uniform.\n\nThis behaviour is unwanted for a random distribution and leads to visual artifacts as well as inconsistencies in simulations. ('even distribution' does not solve this, the distribution probably has to be normalized on the actual mesh volume)\n\nSet up a particle system in a cube volume and set the distribution to 'random' and generate all particles in the first frame. Take one edge of the cube pointing in z-direction and pull it further away. Press 7 on the numpad to see the effect in the distribution. The effect becomes prominent with around 5000 particles.\n\nThis is the top view of a cube with one vertical edge dragged out:\n\n\n\n[#71731-particleVolume.blend](T71731-particleVolume.blend)\n", "Nodes to convert between cartesian/spherical coordinates\nThere should be nodes to convert between spherical and cartesian coordinates in a standardized way.\n\nThis point came up when discussing [D12746](D12746) .\n\nThey are multiple design aspects to be figured out:\n- Node layout\n- Naming and Scope\n- Units\n- Value Ranges", "\"Z\" changed to \"depth\" breaks consistency in compositing and in Render layers\n\nwin7 gtx 550ti\nOperating system and graphics card\n\nBroken by: 4cf7fc3b3a\n\n\"Z\" changed to \"depth\" breaks consistency in compositing and in Render layers\n\n\n- Make sure that the **z**-pass is checked in the Render Layers tab\n- Open node editor, notice that the**z**-pass is now **depth**-pass\n- Notice as well that everywhere else it is still consider as**z**-pass even in the manual z_combine.html\n\n\n\nAlthough **depth** is more meaningful name, it is breaking consistency. Suggestion: we bring back the old behavior or we change everywhere to z-depth or depth.\n" ]

[ "[Mantaflow] voxel shape changes to cuboid upon resizing domain \nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: AMD Radeon HD 7560D ATI Technologies Inc. 4.5.13399 Core Profile Context 15.201.1151.1008\n\nBroken: version: 2.83 (sub 15)\nWorked: 2.82a \n\nIf you scale the domain along one axis, the voxel in the corner of the domain will stretch and lose its cubic shape.\n\nAdd a cube.\nMake this cube a domain.\nDouble-scale this cube on the x-axis.\nYou will see this picture:\n\n" ]

[ "Zooming in and out of the node viewport, text keeps changing its size\nOperating system: Arch Linux\nGraphics card: NVIDIA GTX 980\n\nBroken: Blender 2.90.0 Stable Release\nWorked: I'm not sure if this has ever worked correctly but it looks ugly.\n\n\nWhen zooming in and out of the node viewport, the text keeps changing its size making it look like it's jumping all around. I think the view should only zoom in and out without ever changing the text's size.\n\n\nAdd a principled BSDF node and then zoom in and out.\n\n[nodetitle.webm](nodetitle.webm)\n\n", "GPU subdivision option with subdivision surface modifier collapses geometry (AMD GPU)\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: AMD Radeon HD 7800 Series ATI Technologies Inc. 4.5.14831 Core Profile Context 21.5.2 27.20.20903.8001\n\nBroken: version: 3.1.0 Alpha\n\nSubdivision surface modifier with GPU Subdivision enabled produce geometry that appear to be lost\n\n1. Open default scene.\n2. Add subdivision surface modifier to an object.\n3. Make sure GPU Subdivision option is enabled.\n4. ~~Uncheck Use Limit Surface~~\n5. ~~Disable GPU Subdivision to see correct result.~~\nLast two steps are not necessary for recreation of the bug since this change- 118a219e9d\n[Subdiv GPU Use Limit Surface 2022-01-15 22-13-33.mp4](Subdiv_GPU_Use_Limit_Surface_2022-01-15_22-13-33.mp4)\n\n------\n\nNOTE: This bug seems to only occur on AMD cards on Windows.\n\n**Reproducible on:**\n\n| Operating System | GPU | Driver |\n| ----------------|-------------|---------|\n| Windows 10 | AMD Radeon HD 7800 Series | 4.5.14831 Core Profile Context 21.5.2 27.20.20903.8001 |\n| Windows 10 | AMD Radeon (TM) RX 480 | 4.5.14760 Core Profile Context 20.45.37.01 27.20.14537.1001 |\n| Windows 10 | AMD Radeon(TM) 535 | 4.5.14831 Core Profile Context 21.5.2 27.20.21003.8013 |\n\n**Not reproducible on:**\n\n| Operating System | GPU | Driver Version |\n| -----------------|-------------|---------|\n| Linux Mint 20 | AMD Radeon RX550 | 4.5.14800 Core Profile Context 21.40 |\n| Linux Mint 20 | AMD Radeon RX550 | 4.6 (Core Profile) Mesa 21.3.0-devel |\n| Windows 10 | Intel(R) UHD Graphics 620 | |\n| Windows 10 | NVIDIA GeForce GTX 1050 | 472.39 |\n| Linux Mint 20 | NVIDIA RTX 2070 | 510.47.03 |\n| Linux | NVIDIA GeForce GTX 970M | 495.44\n|Windows-10-10.0.19043-SP0 64 Bits|Radeon (TM) RX 480 Graphics| 21.30.23.01, 22.2.3, 21.10.2 |\n", "When changing the width of frame, it also changes frame location.\nIf you create frame and set it into (0,0) location. Then grab right side of the frame node and make frames width bigger. Now using python read the frame \nlocation again and it is not (0,0) anymore. \n\n", "No fluid particles created when switching from gas to liquid\nOperating system: Linux-5.4.0-40-generic-x86_64-with-debian-bullseye-sid 64 Bits\nGraphics card: GeForce GTX 760/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 435.21\n\nBroken: version: 2.90.0 Alpha\nWorked:2.83 sometimes shows the same behaviour, and sometimes it works (???) however in 2.90.0 I can consistently reproduce this.[logs.zip](logs.zip)\n\nWhen 'manually' creating a fluid sim, no particles are created, while when using 'quick liquid' it does work as expected.\nChanging the cache to an empty directory correctly initializes the particles.\n\n- default startup\n- switch to wireframe view\n- duplicate startup cube\n- scale duplicate x 2\n- select original cube. go to physics properties. check liquid.\n- choose 'type = flow' in fluid panel\n- choose 'flow type = liquid' in the fluid settings panel, leave the rest of the settings as is.\n- select larger cube\n- go to physics properties, check fluid\n- check 'type = domain'\n- check 'domain type = fluid'\n\nNo particles are created inside the small cube. \n\nGoing to the cache settings and selecting a different (empty!) cache folder solves it makes it create particles.\n\nWhen I set MANTA::with_debug to 1 and I compare a run between 'quick fluid' and 'manual fluid' the first obvious difference is that has_data is 1 in the 'manual' fluid run.\n\nCould it be that erasing the cache for the unused gas cachedata fails somehow?\n\nAttached are two console outputs , one from quick fluid and one using the manual steps above (except that I used a different scale factor for the domain in these runs) \n\n\n(edit: removed embedded text files and added them as a zip)\n(edit2: cleared up some mixup with the bugreport I originally wrote for 2.83)\n(edit fix small typo)", "Setting the brush radius unit to \"scene\" does not automatically use correct stroke spacing distance\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce RTX 2070 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.38\n\nBroken: version: 2.91.0 Alpha\n\nThe brush settings have an option to fix the radius to either the view or scene, called \"Radius Unit\".\nIf the Scene setting is used, it will not automatically use the correct spacing distances for the stroke.\nThis needs to be toggled manually with separate stroke setting called \"Spacing Distance\" from view to scene.\n\nIf the viewport orientation is very close to the world origin, rotated towards an object that is further away from the world origin, it will lead to a much higher spacing distance.\n\n[2022-05-13 10-24-15.mp4](2022-05-13_10-24-15.mp4)\n\n**Exact steps to reproduce**\n- Move the default cube 4 units away from the world origin\n- Add a multires modifier and subdivide 5 times\n- Enter sculpt mode\n- Use the draw brush\n- Switch the Radius Unit setting from View to Scene\n- Orient your viewport so that you are positioned very close to the world origin, looking at the cube\n- Sculpt on the cube (Notice the very high spacing distance)\n- Rotate your view around to the other side of the cube (Now the viewport is positioned further away from the world origin)\n- Sculpt on the cube from that angle (The stroke distance should be much smaller now)\n\n- To fix the issue switch the stroke setting \"Spacing Distance\" from View to Scene\n- Retry sculpting from the same angles", "Subdivide with smoothness not symmetrical (not deterministic, depends of order of vertices in face)\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GT 430/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 388.13\n\nBroken: version: 2.83.4\nWorked: Never (2.8+)\n\nMesh is no longer centered along axis after using Subdivide with Smoothness greater zero. Projection of mesh is not exactly rectangular when looking form top.\nThe position of vertices, created with Subdivide with applied Smoothness, is wrong/arbitrary only if the vertex is placed inside the original face/polygon not along the edges of it. (Quads have one vertex inside face after one subdivision, Triangles need atleast two subdivisions for the error to appear because first subdivision creates new vertices only on edges)\n\nFirst example:\n1. Open subdivide_smoothness_plane.blend\n2. Apply Subdivide with Smoothness to the mesh\n3. Mesh is no longer centered along y-Axis. Compare position of vertices created inside the two quads.\n[subdivide_smoothness_plane.blend](subdivide_smoothness_plane.blend)\n\nSecond example:\n1. Open subdivide_smoothness_multiple.blend\n2. Apply Subdivide with Smoothness to the mesh with MORE then one Number of Cuts\n3. Mesh is no longer centered along y-Axis\n[subdivide_smoothness_multiple.blend](subdivide_smoothness_multiple.blend)\n", "Vertex Transformation, Number Rounding Problem, versions 2.79 and above!\nOperating system: Windows 7 x64 Ultimate SP1\nGraphics card: onboard Intel HD 4600\n\nBroken:\n2.79 (latest) AND 2.80 latest beta\n\nWorked: (optional)\n2.77a works correctly!\n\n\nThis problem is happening on blender versions 2.79 and 2.80! It is NOT present in 2.77a.\nHow to replicate: Selecting a vertex and clicking on its axis transformation on the right pane (N), for example on the Y-axis, enter the number 47.15426 for example.\nSo far, so good. But after clicking on that same field again (to edit or just copy the value to the clipboard), it gets automatically rounded off (just by clicking on it) to 47.1543 for no reason.\nBasically, when I want to select the coordinate of that vertex (47.15426 is its correct value) and just copy it to my clipboard so I can paste it into another program, that number gets changed automatically to 47.1543 inside Blender and it stays that way after de-selecting the Y-axis number field in the Transformation tab!\n\nIn 2.77a version, no such thing can happen, I have been using it since 2016 and there weren't any problems with that.\nI have tested both the 32-bit and 64-bit versions of the affected 2.79 and 2.80 releases, all have this issue.\n\nOpen the attached .blend file: [vertex_coordinate_rounding_problem.blend](vertex_coordinate_rounding_problem.blend)\nThe vertex is already selected. Just navigate with your mouse to the Y: axis transformation and click on the field with the number. It will change from 47.15426 to 47.1543 (which literally lowers the precision of the float by a decimal point). If you try that in Blender 2.77a, this will not happen.", "Set parent (keep transform without inverse) has no effect on scale\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: GeForce GTX 750 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.71\n\nBroken: version: 3.2.0 Alpha\n\n\nSet parent (keep transform without inverse) has no effect on scale. End result is different if use Set parent (without inverse)\n\n[parent_scale.mp4](parent_scale.mp4)\n", "Sculpt Mode: Edit Voxel Size hotkey conflict\nOperating system: Win 10\n\nBroken: blender-2.90-51ac411ce80c-windows64\n\nThe \"Edit Voxel Size\" and the \"Repeat Last\" functions are using the same hotkey (Shift+R), and as a result, the \"Repeat Last\" command is not working in sculpt mode.\n", "Multires: Artefacts from lower subdiv level changes\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 980/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 442.50\n\nBroken: version: 2.90 (sub 0)\n\nWhile I was testing MultiRes in 2.90, I found a bug when sculpting on lower subdiv levels which recently got activated. If you for instance smooth the sculpt at a lower subdiv level and save your sculpt while at any lower subdiv level that isn't the highest one, your sculpt will deform with artefacts. Here's a demonstration:\n\n[Desktop 2020.05.03 - 17.27.17.02.mp4](Desktop_2020.05.03_-_17.27.17.02.mp4)\n\n\nHere's the file I was working on.\n[Urbosa Sculpt.blend](Urbosa_Sculpt.blend)\n\nHere's what you need to do to reproduce:\n\n1) Smooth an area with a lot of sharpness to it at the current subdiv level (it can be any level except 5 which is the highest one).\n\n2) Save the project.\n\n3) Artefacts should appear now. Change to other subdiv levels, including level 5 and repeat the save just to compared the results. The highest subdiv level should always work. An alternative way to not reproduce the bug is to smooth at a lower subdiv level and then go to the highest level available and save. Then the artefacts don't appear either.", "mathnode socket doesn't restore when voronoi signal is turned to \nOperating system: Linux-4.15.0-66201911071317-generic-x86_64-with-debian-buster-sid 64 Bits\nGraphics card: GeForce GT 630/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 390.116\n\nBroken: version: 2.82 (sub 6)\nWorked: (optional)\n\nthe mathnode after a voronoi tex node ejects the noodle when you switch voronoi to \"n-sphere radius\" and keeps hiding any values.\n\nopen blender\ncreate a material for an object of your choice\nadd voronoi and mathnode as shown in the picture\nswitch voronoi from f1 to n-sphere radius\nnotice that the mathnode disconnenćts, which is fine, butit should re-open the slot then, right?\n\n\n\n", "View -> Lock Time to Other Windows does not maintain scale when CTRL+TAB between two animation editor modes.\nOperating system: Manjaro Linux\nGraphics card: nVidia\n\nBroken: (example: 2.80, edbf15d3c044, master, 2018-11-28, as found on the splash screen)\nWorked: (optional)\n\nView -> Lock Time to Other Windows does not maintain scale when CTRL+TAB between two animation editor modes. \n\nExpand open the Timeline editor, now adjust the horizontal scale of the view (squeeze or expand), then Ctrl+TAB to swith to Graph Editor mode. Notice the scale does not stay in sych as it does when they are separate windows. ", "Fix operators that affect hidden geometry\n(NOTE) This task is part of the community & sub tasks of the workboard. Anyone is free to pick up this task and contribute. For any questions or needed reviewers, please tag @JosephEagar and @JulienKaspar .\n\nThere are a number of operators in sculpt mode that affect hidden geometry. As a general rule this should be avoided.\n\n- [ ] `bpy.ops.sculpt.face_set_edit` to Grow/Shrink, Fair or Delete face sets needs to have the \"Modifiy Hidden\" property disabled by default and disabled in all menu and keymap entries.\n\nList of further operators:\n- [ ] `bpy.ops.sculpt.face_sets_init` \n- [ ] `bpy.ops.sculpt.face_sets_create(mode='SELECTION')`\n- [ ] `bpy.ops.mesh.paint_mask_extract()`\n- [ ] `bpy.ops.mesh.paint_mask_slice()`\n\n", "Edit Dyntopo Detail Size: Add resolution value for visual feedback\n(NOTE) This task is part of the community & sub tasks of the workboard. Anyone is free to pick up this task and contribute. For any questions or needed reviewers, please tag @JosephEagar and @JulienKaspar.\n\n## Issue\n\nWhen using the `sculpt.dyntopo_detail_size_edit` on `R` while Dyntopo is set to 'Constant' or 'Manual' Detailing in the dyntopo settings, \nit can easily happen that the overlay doesn't show the wires that indicate the resolution.\nThis happens when the value is much higher than the current brush size, like when working on a very small object.\n\n\n\n## Suggestion\n\nIt would be give better feedback to the user that this is happening by displaying the 'Resolution' value\n- in the header\n- in the viewport itself, like with `object.voxel_size_edit` on the same shortcut.\n\nThe first solution would be the easiest and most consistent with other operators.\nThe second solution is also very welcome but would need the text to be placed outside of the brush cursor (Perhaps even dynamically placed depending on the availible viewport space).\n\n\n\n", "Particle System Orientation Axis set incorrectly after file load\nOperating system: Windows 10\nGraphics card: Nvidia GTX 1070\n\nBroken:\n2.82 Beta, 507a331f0182, blender2.8, 2020-01-27\n\nAfter undoing any change, every particle system in the scene resets its orientation axis. To fix this, you have to manually change the orientation axis to something else and back again. There may be other situations where the particle orientation gets reset, but this one is easily reproducible.\n\n1. Open the attached file\n2. Set the ParticleSettings>Rotation>Orientation Axis to Global Z\n3. Shift-A and add a cube (or any change at all to the scene)\n4. Ctrl-Z\nThe particles are not longer aligned on the global z-axis. \n\nAlternatively, it also breaks if you:\n1. Open the attached file\n2. Set the ParticleSettings>Rotation>Orientation Axis to Global Z\n3. Save the file\n4. Close and reopen the file\n\n[ParticleBug.blend](ParticleBug.blend)\n" ]

[ "Fluid domain and voxel proportions inconsistency with scale.\nOperating system: Windows-7-6.1.7601-SP1 64 Bits\nGraphics card: GeForce GTX 660 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 442.59\n\nBroken: version: 2.83 (sub 17)\n\nVoxel proportion of domain depends on when i applied scale.\nIt leads me to problem: If I want to have proper cubic voxel, I need: assign domain, change resolution to lowest, scale, apply, change resolution back.\n[vp.blend](vp.blend)\n[2020-05-22_12-40-04.mp4](2020-05-22_12-40-04.mp4)\nI can assume, that sometimes it is fun to have stretched voxels. But most time — it is not.\n\n\n1. Make cube, assign domain, scale X 4\n2. Copy it, apply scale\n3. Make cube, scale X 4, apply scale, assign domain\nCompare results." ]

[ "After duplicating a collection with rigidbody objects, the simulation doesn't apply to the new rigidbody objects\nOperating system: Windows 11\nGraphics card: NVIDIA GeForce RTX 3080 Laptop GPU\n\n\nBroken: `master` branch\nWorked: Don't know if it ever worked\n\nAfter duplicating a collection with rigidbody objects, the simulation doesn't apply to the new rigidbody objects\n\n**Steps to Reproduce**\n\nIn a new blender project, add a new collection. Add a cube to the collection. Add a rigidbody to the cube. Verify that playing the simulation causes the object to fall under gravity.\n\nCopy and paste the cube to create a new cube inside the collection. Move the new cube so both are visible. Verify that playing the simulation causes both cubes to fall under gravity.\n\nIn the outliner right-click on the collection and select \"Duplicate Collection\". Move the two new cubes so all four cubes are visible. Run the simulation.\n- The two cubes in the original collection will fall under gravity.\n- The two cubes in the duplicate collection do not move and are not affected by the simulation.\n\n \n\nIt appears the two cubes in the duplicated collection are not added to the RigidBodyWorld collection.\n\n", "Clearing library overrides from the object menu differs from the outliner\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: Quadro RTX 3000 with Max-Q Design/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 536.67\n\nBroken: version: 3.6.1 Release Candidate\n\nPlease see the attached movie. Using the \"wrong\" menu will cause the material properties panel to become unresponsive (beginning of movie) with overrides still in place until you use the outliner (end of the movie).\n\n1. Link some objects from a file\n2. Override them down to the materials\n3. Swap out some materials\n4. Use the clear library override from the object menu\n\nResult:\n\nThe material overrides aren't cleared, but you also can't access the properties panel anymore.\n\nExpected:\n\nThe material overrides should be cleared.\n\nWorkaround:\n\nUse the outliner instead.\n\n", "Shadow catcher problem with holdout collection\nOperating system: *Windows 10 Pro 64 bits*\nGraphics card: *2x NVIDIA GTX Titan X*\n\nBroken: *master 2.81a*\nWorked: *Worked on blender 2.79b*\n\nWhen shadow catcher is activated in an object located in a collection with the holdout option, alpha is truncated.\n(more visible when there is motion blur)\n\nBased on the default startup :\n- set render engine to cycles\n- set Film parameters to Transparent\n- activate Motion Blur\n- set world backgound color to white (better visibility)\n- delete light\n- create a plane, scale x2\n- move plane to a new collection\n- set plane to shadow catcher\n- set collection 2 (called by default, with the plane) to holdout\n- animate camera for have motion blur (more visible when there is motion blur)\n- in viewport render, you will have that : \n- render.\n\nwithout motion blur :\n antialiasing problem\n\nwith motion blur :\n\n\nsame render result with and without alpha on composite node :\n\n\n\n.Blender here : \n[#73376.blend](T73376.blend)", "Design discussion: Add object instancing to support transforming linked objects?\nThis is an initial task to start discussion on this topic. More details would have to be looked into for an actual design.\n\n### Motivation\n\nLinking objects in Blender has the inherent design limitation that it also links the object's transform, which means you can't transform an object that was linked. This is a very basic and important operation. With library overrides, that issue is mitigated; however IMO we should look at this as a feature for production use-cases, not as something for more informal, low-overhead usages. This would be a complex solution for the user to manage for a simple problem. They need rather deep understanding of how Blender data-management works to manage library overrides.\n\nEspecially for the Asset Browser, it would be good to have a simple solution that behaves as users expect. Apparently what they do expect is that they can simply link objects, while being able to transform them still. Not being able to edit other object properties seems fine.\n\n### Possible Solutions\n\nFor collections we have actually addressed this issue by supporting linked collection instances. That means, while the collection and its contents are linked, the collection is placed relative to an emtpy that is local to this file. So by transforming the empty the entire collection is transformed. The idea of instancing can be brought to objects so transforming linked objects becomes possible in effect.\n\nSome ideas on how object instancing could be supported on linking:\n* **Object is made local, the mesh, materials, etc stay linked** Simple to implement - downside is that object properties, modifiers, constraints, physics and the like would not be linked anymore.\n* **Instancing via empty.** This would be very similar to collections. An empty can instance an object which means it acts as a parent of the object.\n* **Wrap into collection instance.** Effectively the same as instancing via an empty, but the user would see an intermediate collection. It's simple to add this, but kind of forces an implementation detail onto users.\n\n### Other Approaches\n\nInstead of trying to support object instancing, other things could be done on linking:\n* **Automatically create library overrides for transforms**. Another simple solution technically and at least saves the user from having to know about or setup the library overrides when linking. But obviously pushes users towards using library overrides, which they sooner or later will still have to deal with.", "Custom Properties Improved UI\nCurrent UI:\n\n\n\n- [ ] Allow reordering of custom properties in ways that make the most sense to an artist\n- [ ] Allow editing from the sidebar\n- [ ] Decrease the space between properties\n- [ ] Using an 'X' icon to delete\n- [ ] User and edit icon instead of \"Edit\"\n- [ ] Better location of icons\n- [ ] Group properties by the selected object/bone (see image)\n- [ ] Add a drop-down to select the data type\n- [ ] Allow the use of checkboxes to represent booleans\n\n", "2.8 - Particles in multi-level nested collections not visible\nOperating system: win10x64\nGraphics card: AMD Radeon Pro WX 7100\n\nBroken: 2.80, 54ffc4e19dc4, 2019-02-25\n\n\n\nIn nested collections, objects with particles work only in two levels of nesting. Particles in further levels of nesting seem to ignore collections instancing.\n\n\nRelated file: [particle-instances.blend](particle-instances.blend)\n\n- create cones with hair particles on a collection called **branches** \n- create a cube on a collection called **tree** \n- add a collection instance of branches within the collection called **tree** (1st level of nesting of branches) \n- create a new collection called **trees** and add a few collection instances of tree. \n- for all objects in trees, the particles are visible ( 2nd level of nesting) \n- create a new collection called **instances** and add a collection instance of trees. Here, the particle settings are not replicated. (3rd level of nesting) .\n- For the sake of testing this further, create a new collection called **instances2** and add a collection instance of instances . Also, here the particle settings are not replicated. (4th level of nesting) \n\n\nIn collection Instances, only the initial particles are visible and their location seems to be reset based on the initial location\n\n\n", "Asset Browser: Instance option shifts objects from the nested collection to the world origin\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3050 Laptop GPU/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 512.78\n\nBroken: 3.5\nWorked: unsure\n\nIf we have a collection as a asset which contains nester collection, after spawning such collection with instance unchecked will keep \nobjects from nested collection at world origin\n\n- Download zip\n- Open default scene add set this as asset library path; `Example/Asset Browser`\n- Open asset browser -> select `User Library`\n- Drag & drop `Example_Object`\n- Uncheck `Instance` from redo panel. \nNotice collider objects shifts to World origin\nFound this issue by @JelSadones while discussing #103216 (Linked objects only count once towards the triangle count)\n- - - \n**Additional Info**\n- Open Prop_Example.blend\n- Remove `BoundingBox` and `CustomColliders` collection and move all objects under single collection (i.e. `Example_Object`)\n- Save file then open a default file.\n- Now drag & drop this asset and uncheck Instance option (all objects will be at the instanced position)\n\n[Example.zip](Example.zip)", "Animation: Replacing bone layers + groups with bone collections\nThis design describes the replacement of armature _bone layers_ & _bone groups_ with a new _bone collections_ system.\n\nImplementation is in #109976 and #112271.\n\n## Goals\n\n- Named collections, instead of numbered layers, should give a better indication as to what the collection contains / is for.\n- Increased flexibility in number of collections.\n- More similarity to the scene object collections, including features like 'not selectable' which can increase clarity of how to use a rig by animators.\n- Compatibility with library overrides, so that new shot-specific bone collections can be added.\n- Unification of numbered bone layers (for organisation) and named bone groups (for coloring).\n\n## Properties of the System\n\n- Armatures have one or more bone collections.\n - If there is only one collection it cannot be removed.\n- Every bone is in zero or more bone collections.\n- Bone collections **cannot be nested**. At least not for 4.0, can be added later.\n- Bone collections must have a **unique name** within the armature.\n- Library Overrides:\n - New bone collections can be added.\n - Existing ones cannot be reordered / removed.\n- Bone collections have the following properties:\n - Name\n - Selectable: 'set' wins over 'cleared'\n - Visible: 'set' wins over 'cleared'\n - `IDProperties` (aka Custom Properties) for use by add-ons.\n- Bones will get individual color properties.\n - The color system of bone groups will be used, which allows selection of 'default', 20 theme-defined colors, and a custom RGB value.\n - Bones have a color defined as part of the armature data. Here 'default' is the default grey color.\n - Pose Bones have a color defined as part of the object/pose data. Here 'default' means 'uncolored', and will thus show the edit bone color.\n\n## Operations\n\nThese operations can be performed on/with bone collections:\n\n- (De)select one or more collections themselves\n- (De)select all bones in active collection\n- (De)select all bones in collections containing the currently selected bones\n- Create a new collection\n- Delete a collection. Bones in the collection that are now 'orphaned' are assigned to the 'default' collection.\n- Move collection up/down in the list.\n- Move selected bone(s) to a collection.\n- Select Bones of Same Color.\n\n## User Interface\n\nFor Blender 4.0 the UI is kept intentionally simple. I'm thinking of reusing the Bone Groups interface, with the addition of a few per-collection icons (visible, selectable, color palette) as well as an indicator of which collection(s) contain the selected bone(s).\n\n## Conversion from Old to New\n\n- Layers converted to `Layer {NUM}` collections.\n- Bone groups converted to `{name}` collections.\n- Selection Sets converted to `SelSet {name}` collections.\n- Maybe even automatically merge these when they contain the same bones?\n\n## Uses of the Current System\n\n### Bone Layers\n\n- Visibility\n- Logical grouping\n\n### Bone Groups\n\n- Coloring of logical groups\n- Set by rigger\n\n### Edit Mode vs. Pose Mode\n\nThis describes Blender's **current** behavior:\n\n| Property / Action | Edit | Pose |\n|---------------------------------------|------|------|\n| Name Bone | ✅ | ✅ |\n| Set Bone Layers | ✅ | ✅ |\n| See Bone in Group Color | ❌ ¹⁾ | ✅ |\n| Create / Assign Bone Group | ❌ ¹⁾ | ✅ |\n| Select Bones in Group | ❌ ¹⁾ | ✅ |\n| Set Bone Group Color | ✅ | ✅ |\n| Set Armature Viewport Display options | ✅ | ✅ |\n| Set Bone Viewport Display options | ❌ ²⁾ | ✅ |\n\n¹⁾ The new system should allow these in Edit mode too.\n²⁾ Bones can only be hidden, which is separated between edit mode & pose mode.\n\n## Future Work\n\nThese are ideas for future improvements / extensions:\n\n- Nestable bone collections.\n- Integration with the Outliner.\n- Per-collection option whether bones are deforming or not (rather than setting that per bone)\n- Per-collection viewport display mode (octo, line, bbone, etc)\n- Possibly unification of Selection Sets as well.\n\n## Code Location\n\nThis work will kickstart a new module in `source/blender/animrig`. This will become the place for new animation/rigging related code. The intention is that also a subset of the already-existing animation-related code from both `blenkernel` and `editors` is going to move there, to help clean up the API and make it more consistent.\n\n-------------------------\n\n## Ongoing Work\n\n**Targets for 4.0-bcon1: tracked in #109976**\n\n**Targets that could be pushed back to 4.0-bcon2:**\n\n- [x] Add ID Properties support to bone collections, including a UI.\n- [x] Removal of bone group RNA access & UI\n- [x] Ensure that library overrides can add new bone collections & assign bones to them (#112271)\n- [x] Removal of bone groups from the UI, including the Ctrl+G menu\n- [x] Add bone collections to the outliner (replacing the current list of bone groups)\n- [x] Operators to move/add selected bones to a collection:\n - [x] move to collection (similar to 'm' for objects & collections, from viewport)\n - [x] add to collection (similar to 'shift+m' for objects & collections, from viewport)\n- [x] Select Similar operator (by collection, by color) / Shift+G menu\n- [ ] Fixing the Collada I/O, as that also handles bone layers.\n- [x] Make bone collections on linked-but-not-overridden armatures usable. The object itself may be overridden, so changing the active bone collection should work (even when it cannot be saved), and then 'Select Bones' and 'Deselect Bones' operators should work.\n\n**Potential push back to later:**\n\n- [ ] Add selectability toggles to bone collections (only for pose mode)\n- [ ] Add armature-global override to selectability, so that a rigger can still opt to select all the bones, then lock them down for animators.\n- [ ] Managing bone assignments to collections from the outliner.\n", "Linked data-block has \"pointer\" property still active\nBroken: version: 3.4.0 Alpha, branch: Unknown, commit date: Unknown Unknown, hash: `rBUnknown`\nWorked: Probably never\n\n\n\n[torus.zip](torus.zip)\n\nOpen torus.blend. It has a linked collection containing the selected object `Torus`.\n\nAs you can see in the image, the \"Object\" rna properties (parent, and the array modifier one) are both active:\n\n{[F13627034](image.png), size=full}\n\nIn fact you can even click on the property to change it (the popup show up, but clicking on it at least doesn't change it):\n\n{[F13627039](image.png), size=full}", "Outliner: Unexpected behavior when duplicating object hierarchies\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA RTX A4500/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 512.15\n\nBroken: version: 3.4.1\n\n**may not be a bug** - duplicate a parent object (i.e. an object that has child objects). The duplication automatically selects the new object in the 3D viewport but this selection differs from that shown in the outliner. This means dragging to, say, a new collection does not work as expected (only the parent object is moved leaving the child behind).\n\nThe set up (this is the attached blend file):-\n1. Create two cubes (or any meshes) and parent one to the other to form a hierarchy.\n2. Add the object(s) to a collection.\n3. Create another top level collection which is empty.\n\nThe aim:-\nTo duplicate the object(s) (not the collection) and copy the duplicate to the second (currently empty) collection.\n\nThe steps:-\n1. Right click the parent object in the outliner and choose \"Select Hierarchy\". Note that the outliner shows the selected objects with a blue background bar (as expected).\n\n2. In the 3D viewport, use Shift or Alt D (doesn't matter which) to create new duplicate objects. Note that the 3D viewport shows the duplicate parent and child selected. The outliner shows a new object selected (with blue bar) but the tree is collapsed so we can't see the child.\n\n3. In the outliner, click and drag the selected parent object to the empty collection. Note that only the parent moves, the child is left behind. This child has then to be dragged separately.\n\nAfter step 2, every visual cue leads the user to believe that the duplicate hierarchy is selected. And this is the intuitive behaviour expected. But expanding the outliner tree shows the child object is not, in fact, selected (no blue bar). So an additional step of having to reselect the hierarchy is required before step 3.\n\nThe main issue is that with lots of objects and detailed hierarchies, just forgetting that additional \"select hierarchy\" step before dragging means that child objects can be scattered all over the place away from their parent and the user is left picking through the large list trying to sort it out.\n\nThis may not be a bug, it may be working as designed. But I cannot think of a use case where the current functionality would be useful so I'm assuming it's a small selection bug.\n\n", "Custom property in workspace \"Layout\" can not be added to driver\nOperating system: Linux-4.10.0-28-generic-x86_64-with-debian-stretch-sid 64 Bits\nGraphics card: Mesa DRI Intel(R) Ivybridge Mobile Intel Open Source Technology Center 3.3 (Core Profile) Mesa 17.0.7\n\n\nBroken: version: 2.80 (sub 75)\nWorked: (optional)\n\n\n\n\n\nWhen adding custom property in any workspace can't use it in any drivers. It works with object's custom property world's custom property, but not in workspaces.\nI think it's a bug.\n\nThank's in advance.\n\n\n\n\n\n\n[workspace_custom_property_drivers.blend](workspace_custom_property_drivers.blend)", "Hold Out does not hold out Collection Hierarchy\nOperating system: Linux-5.19.0-50-generic-x86_64-with-glibc2.35 64 Bits, X11 UI\nGraphics card: NVIDIA GeForce RTX 3070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 535.54.03\n\nBroken: version: 3.6.1\n\nNested collections do not get held out if only the top level collection is held out.\n\n\nNote attached blend file.\n\n\n\nNote the foreground collection with 2 subjects in it. When holding out the foreground, the nested collections (collections held within the \"Foreground\" collection) are not held out.\n\n", "Multi-object-Mode: EditMode transform properties\nEdit-mode buttons currently only show median values for the active edit-object.\n\nThe Mean Crease value of the N panel doesn't work for multiple objects:\nHere i have 2 objects selected\n\nIn Edit Mode changing the Mean Crease value does change its effect only on the selected object:\n", "Reference/background images hide additional information in the Properties Editor\nOperating system: Linux-6.4.13-200.fc38.x86_64-x86_64-with-glibc2.37 64 Bits, WAYLAND UI\nGraphics card: AMD Radeon Graphics (renoir, LLVM 16.0.6, DRM 3.52, 6.4.13-200.fc38.x86_64) AMD 4.6 (Core Profile) Mesa 23.1.6\n\nBroken: version: 4.0.0 Alpha\n\nReference/background images hide additional information in the Properties Editor\n\n- Add > Image > Reference / Background\n- observe the Image panel in the Properties Editor\n\n- compare with the Image panel in the Image Editor\n\n\n", "Properties Editor Popover\nWith properties search and outliner to properties tab switching we need a place in the properties editor for various settings. These options may include:\n* Outliner Tab Switch. There isn't an automatic solution that works well enough, so an explicit toggle is needed. [Auto, Enabled, Disabled]. Auto will keep the current behavior of only switching tabs on shared borders\n* Property search: Full tooltip text search would be useful, but could make performance suffer so a toggle would be useful\n* Tab hiding. Some workspaces don't need to display certain tabs and they can be hidden by default" ]

[ "Ctrl-G creates ghost collections\nSelecting couple of objects and hitting Ctrl-G (or going from menus Object->Collection->CreateNewCollection) doesn't prompt for a collection name.. also, nothing new shows up in outliner's collection view. but, if you switch over to \"Orphan Data\" in outliner, you see it actually created some collections. \n\nIf you select objects, and hit \"m\" (Move to Collection->New Collection) it correctly prompts for collection name and it works as expected.\n\nso.. hmm, wrong operator in Ctrl-G ? \n\nalso, I'm not sure if Shift-Ctrl-G (Remove from collection) really does anything. You can keep removing and removing but they are always there... \n\n\n.b ", "Collections Bug... Collectios created via 3DView Objects/Collections/Create New Collection does nor register in the Outliner... There are descrepancies in how the 3DView identifies objects according to Outliner\nOperating system: Windows 8.1 service pack 2\nGraphics card: Intel graphics Intel(R) Core(TM) i7-4700MQ CPU @ 2.40GHz\nand Nvidia GeForce GT 745M/PCIe/SSE2\n\nBlender version: 2.80 (sub 35), branch: blender2.8, commit date: 2018-12-05 02:05, hash: 1fa527bfa3a, type: Release\nbuild date: 04/12/2018, 18:16\nplatform: Windows\n\nCollection naming and identification discrepancies between Outliner and 3DViewport\n\nWhen adding Collections via Outliner alone... seems fine... each newly added collection gets a sequential number and is added to list.\nHowever, when Using the 3DViewport Menu options... Object/Collections/Create New Collection... the input box appears to allow a new Collection name to be specified... default simply says Collection... typing in Collection4 and hitting return does not add a Collection called Collection 4 to Outliner List... however, if you then select and object in 3DView and choose Add to collection Instance... Collection 4 is listed as available (even though not in Outliner.... Also Object/Collection/Add to Active Collection can recognize it as well.... but still the Outliner doesn't see Collection 4....\nThus items dragged in Collections are not updated as belonging to correct Collection in 3Dview, and 3DView has different idea about what objects are in what Collections that doesn't match Outliner organization... its as if they have their own database with no re-concilation or agreement... So how do the objects maintain proper grouping and layering and visibility, etc. between the 2 Editors? ... not very well.\nBased on the default startup or an attached .blend file (as simple as possible)." ]

[ "Faster Animation Playback\n**Status:** Discussing Designs\n\n---\n**Team**\n**Commissioner:** @Mets @Hjalti\n**Project leader:** @Jeroen-Bakker\n**Project members:** @Sergey\n\n**Description**\n**Big picture:** Animators should be able to work in an interactive environment.\n\n**Non functional requirements:**\n\n* Animation playback in the viewport should play at constant framerate at 24fps.\n* `?` Motion curves should be update-able in realtime\n* Playblast shouldn't take much more than regular viewport playback.\n\n**Work plan**\n\n\n**Milestone - Animation Playback/Render Performance**\n\n**Status:** Execution\n**Time estimate:**: \n**Design:** #73429 (Approach Faster Animation Playback)\n**Engineer plan:** Will be created per improvement.\n**Related Tasks**\n* {icon circle color=red} #75134 (TaskScheduler Improvements)\n* {icon circle color=red} #75120 (DrawManager: Improved Task Scheduler)\n* {icon circle color=red} #75208 (Dependency Graph Evaluation Optimizations)\n* {icon circle color=yellow} F-Curve Evaluation Optimizations\n* {icon circle color=yellow} Profile and optimize modifiers (some suggestions are archived in #75207)\n* {icon circle color=yellow} Motion Path Performance\n* {icon circle color=yellow} Optimize VSE Playback for Animators\n* {icon circle color=green} #75124 (ActionEditor: Keyframes Drawing)\n* {icon circle color=green} #75127 (ActionEditor: Add callback function to ANIM_animdata_filter)\n\nThese tasks are ordered in priority. The changes to the task scheduler has most technical impact so we should execute that task with highest priority so we have more time to fix any related issue. The Draw Manager changes rely on features that will be realized by the task scheduler. (@Jeroen-Bakker will do the developments)\n\nChanges to the dependency graph can start from the beginning. Most likely they will be executed by @dr.sybren or @sergey.\n\nThe rest (modifiers, action editor) can be executed afterwards. This is also the time to look at improving motion curves (if this is still needed). (@Jeroen-Bakker will do the developments)\n\n**Milestone - GPU Subdivision**\n\n**Status:** Initial\n**Time estimate:**: \n**Design:** To be completed\n**Engineer plan:** To be completed\n**Related Tasks**\n\n* #68996 (OpenSubdiv GPU acceleration)\n\nThe final design for subdivision (part of the modifiers optimization) needs more collaboration between the GPU/Viewport team and @Sergey to come with an realistic approach.\n\n\n**Milestone - Animation Cache**\n\n**Status:** Needs Design and engineering plan to proceed\n**Time estimate:**: To be determined after Design/EngineeringPlan\n**Design:** #73481 (Design: Animation Caching)\n**Engineer plan:** Not started\n\nAnimation Caching Mechanism: RAM/Disk cache mechanism that can cache the output of an animation. During the animation the cache is invalidated.\nThis cache can also be used for viewport playback and viewport animation rendering.\n", "Undo Makes Completed Renders Disappear From View Render Window Slots\nOperating system: Windows 10 Build 19043\nGraphics card: Nvidia RTX 3090\n\nBroken: 2.92\nWorked: Can't remember, but I don't recall this happening before\n\nUndo makes completed renders disappear from other slots in the View Render window. Some of them are preserved, but the more you undo, the more disappear from other slots.\n\n1. Load default scene (with the cube)\n2. Render scene\n3. Go to the cube's shader and change its base color\n4. Switch to next unoccupied slot in the View Render window\n5. Render scene\n6. (optional) Repeat steps 3-5 as many times as you like\n7. Go to View Render window and scroll through slots to verify all renders are there (with J key)\n8. Activate undo (ctrl+z)\n9. Go to View Render window and scroll through slots. Some will be missing.\n10. Activate undo again and more will be missing.\n\nThis doesn't seem like intended behavior, and I've replicated this issue on a different machine. Hope this helps!\n\nEdit: I've replicated this issue in 2.91, 2.90 and the latest LTS release. I assume intended behavior is for undo to change the shaders back, but for renders you've completed to remain in their slots.", "Menus open on upward mouse movement\nOperating system: Linux (Debian bullseye)\nGraphics card: Nvidia GTX 1070\n\nBroken: current master 2.90.0 (fc672ce8e24e)\nWorked: Never\n\nWhen there are a lot of modifiers in the stack which make the evaluation slow there is a weird effect with the dropdown menu. It somehow glides when dragging to the next modifier.\n\n\n\nTo the default cube, add array modifiers until your memory usage is at roughly 300MiB. Then drag over the drop down arrows like shown in the gif.", "When \"Pie Menu on Drag\" is enabled in user prefs, Ctrl + Z has a tendency to open the shading menu by accident\nversion: 2.80 (sub 74), branch: master, commit date: 2019-07-05 21:52\nbuild date: 2019-07-06, 01:03:01\nplatform: Linux\n\nThis may be more of a design issue than a bug, but it's happened to me about 100 times now and it's quite irritating.\n\n**To reproduce:**\n\n - enable User Preferences > Keymap > Preferences > Pie Menu on Drag.\n - open a complex scene (one where the undo operator will have a significant lag) \n - fill the undo history with events, as though you've been working for a while\n - Press Ctrl + Z a couple times while moving the mouse a bit, rolling your hand so that Ctrl is released just before Z (in the manner of a busy artist)\n\nIf the user happens to release Ctrl earlier than Z while moving the mouse, this can bring up the shading menu, interrupting the user's workflow. This may sound like an edge case, but when working quickly, it's actually really frequent and annoying! It seems to especially effect large scenes where there is a lag in the effect of the undo operator.\n\nIdeally, the \"Pie On Drag\" feature should not listen for key presses if those key presses were part of a different combination that has yet to be completely released. What do you think?", "Inset faces and Bevel Edge command has bugged interaction with Undo command and shape keys\nOperating system: Windows-7,10\n\nBroken: version: 2.83 (sub 2)\nBroken: version: 2.82 (sub 6)\n\n\nIn edit mode, while selecting a shape key, the Inset operation (I) followed by any other operation (Translating vertices or extruding or whatever), followed by an Undo (Ctrl-Z) command causes the shape key selector to hop to the \"basis\" shape key. This also occurs when using the `Bevel` command.\n\nFrom Default Startup:\n1. Create two shape keys: \"Basis\" and \"Key 1\"on Default Cube. Selection will be on \"Key 1\"\n2. Go into edit mode\n3. Inset a face on any of the existing faces with \"I\"\n5. Do ANY other operation or operations, such as moving a vertex or extruding. The bug occurs with any number of operations done after the Inset command.\n6. Undo (Ctrl-Z)\n7. Observe Shape key information as the undo command causes the selection to switch from \"Key 1\" to \"Basis\".\n8. Keep `Undoing` until you `Undo` the `Inset` operation.\n9. Notice that the shape key selection jumps back to \"Key 1\" as you `Undo` the `Inset`. \n", "Performance: EditMesh to Mesh on every modifier run\nThis is a part of performance regression investigation task.\n\nEvery movement of vertex will request modifier stack to be re-run. This involves conversion of EditMesh to Mesh, no matter whether there are modifiers or not.\n\nSince 84cf670d1d it seems that we can simply delay creation of `mesh_cage` in `editbmesh_calc_modifiers` and only do it if there is actual modifier which is to be applied. We can then have `runtime.is_original` set to truth for all Mesh IDs which are covered by Copy-on-Write, and set it to false when modifier runs.\n\nThis will solve about 50% of slowdown in the attached file (where EditMesh to Mesh takes almost same amount of time as batch cache creation).\n\n[megaheavy_mesh_edit.blend](megaheavy_mesh_edit.blend)", "Undo does not revert everything from Library Override/Make Local\nOperating system: Linux-5.14.7-2-MANJARO-x86_64-with-glibc2.33 64 Bits\nGraphics card: AMD VERDE (DRM 2.50.0, 5.14.7-2-MANJARO, LLVM 12.0.1) AMD 4.5 (Core Profile) Mesa 21.2.2\n\nBroken: version: 3.0.0 Alpha\n\nUndo does not revert adding a material slot to library override that was made local.\n\nin the attached LibraryOverridesUndoBug destination.blend that links a cube with \"Material 1\" from the attached LibraryOverridesUndoBug source.blend :\n- With the linked cube selected, use ID Data - Make Library Override (from the right-click menu of the Outliner)\n- use ID Data - Make Local (from the right-click menu of the Outliner)\n- in the material properties, switch Link to Object\n- add a second empty material slot\n- press Ctrl-Z three times and select the cube again - the icon of the empty second slot is still there (vanishes on resync)\n\n[LibraryOverridesUndoBug source.blend](LibraryOverridesUndoBug_source.blend)\n[LibraryOverridesUndoBug destination.blend](LibraryOverridesUndoBug_destination.blend)", "Undo/Redo shortcut still active in the Blender File View window\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 452.06\n\nBroken: version: 2.91.0 Beta\n\nUndo/Redo shortcut still active in the Blender File View window\n\n* Open Blender with the default cube.\n* Move the cube a few times.\n* Open the “Preferences” window in the menu “Edit | Preferences...”\n* Then do Ctrl + Z to undo. The undo shortcut still works. It is the same for the redo shortcut.\n\nThe undo/redo is also active from any other windows of the “File” menu like “Open”, “Link”, “Save”, “Append”, etc. All links showing the “Blender File View”.\n\nIn the “Blender File View” window, I noticed that the buttons at the bottom disappear when doing the undo shortcut.\n\nI tried a few other shortcuts like Tab to go to edit mode and G to move the cube. They do not seem to be active when the focus is on the “Preferences” or “Blender File View” window.\n\nThe undo/redo is also active when the “Blender Render” has the focus.\n\nAs a consequence, someone accidentally hitting the undo shortcut will get his work being modified behind the window and probably will not notice.\n\n\n{[F9111517](BlenderFileView_Open.jpg) size=full}\n\n{[F9111519](BlenderFileView_AfterShortcut.jpg) size=full}", "Cycles: Changing certain settings with `Persistent Data` enabled can lead to extremely long initialization times on re-render\nOperating system: Linux-5.15.0-1-amd64-x86_64-with-glibc2.32 64 Bits\nGraphics card: NVIDIA GeForce RTX 3090/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 495.44\n\nBroken version: 3.1.0 Alpha\nBroken version: 2.93.6\nWorked: Unsure\n\nWhen changing certain settings on some objects while having `Persistent Data` on in Cycles, can lead to extremely long initialization times on re-renders. For example I may have a particle count set to `10,000,000` and I render the scene for the first time. It takes roughly 2 minutes. Re-rendering the scene only takes 10 seconds because of the `Persistent Data` option. However, if I then **decrease** the particle count to `1,000,000`, the render takes a really long time to start (I waited 15 minutes before giving up)\n\n1. Change the render engine to Cycles.\n2. Change render settings such that the render is quick (lower sample count, turn off denoising, reduce the resolution).\n3. Enable `Persistent Data` in the `Render Properties -> Performance` section of the `Properties editor`\n4. Create a hair particle system with lots of objects. I did this by creating a plane, giving it a hair particle system set to count `10,000,000` and set it to \"render as\" a Suzanne object\n5. Render the scene. Take note of roughly how long it takes for the render to initialize. Roughly 2 minutes for me.\n6. Render the scene again. (I'm not sure why but this steps seems to be necessary)\n7. Reduce the particle count on the plane. I reduced it to `1,000,000`.\n8. Render the scene again. Notice how the render takes a really long time to start. (I gave up after waiting 15 minutes.)\n\nHere is a file that does step 1-4 for you:\n[Persistent Data Long Intialization Time.blend](Persistent_Data_Long_Intialization_Time.blend)\n\n**NOTE:** This is not an out of memory issue, at least from what I can tell. Blender gets nowhere close to using the entirety of my RAM (64GB) or VRAM (24GB) during tests using CPU or GPU for testing.", "2.91 - Undo in Sculpt Mode switches tool to inflate brush (specific to attached file)\nOperating system: Win 10 64-bit\nGraphics card: Nvidia GFX 1060 6GB\n\nBroken: \n - 2.90.1\n - 2.91.0, from steam, seems *file specific*\n - 2.92.0\nWorked: 2.91.0, same version, fresh file\n\nWith attached file: In sculpt mode calling undo changes the tool to \"Inflate\" besides undoing.\n\nAdditionally, when undo is called via menu, the undo operation happens only when undoing so far back sculpt mode was entered. On reentering Sculpt Mode, tool is set to \"Inflate\". \n\nThis issues seems to be tied to this specific file, as it's not reproducible in a fresh file. The file was created with blender 2.91.0. I sculpted multiple objects using both Dynamesh and Remesh in Sculpt Mode. I deleted the sculpts to keep the file size small and since the same behavior persisted with newly created geometry (in this case a cube). I don't know how to debug this any further, any hints appreciated. As a workaround I will try to reimport the geometry into a new file.\n\nBased on the attached .blend file [SculptModeBug.blend](SculptModeBug.blend)\n\nUndoing with shortcut:\n - Select object \n - enter sculpt mode\n - select any other tool than \"Inflate\"\n - modify mesh with tool\n - hit ctrl + z\n - tool switched to \"Inflate\"\n\nUndoing via menu:\n - Select object \n - enter sculpt mode\n - select any other tool than \"Inflate\"\n - modify mesh with tool\n - modify mesh with tool\n - call \"undo\" from menu \"edit/undo\" - this has no effect\n - call \"undo\" from menu \"edit/undo\" until the point object goes back to Object Mode - changes get undone\n - On entering Sculpt Mode again, the active tool has switched to \"Inflate\"\n", "DrawManager: Threading for ibo.fdots_nor/hq\nCurrently done in finish with comment that it is faster.\nWe should benchmark when it is faster and document this with the code.\nOr change it to use threading.\n\nIt could be faster, but can stall other batches.\n\nThe comment was written before the previous optimization that supports better batching. I don't see why the current structure would be faster than implementing it with poly looping.\nFile: `draw_cache_extract_mesh.c` section `Extract Facedots Normal and edit flag`.\n", "EEVEE Material Preview works significantly slower than before. \nOperating system: MacOS Mojave 10.14.6 (18G9323)\nGraphics card: Intel Iris Graphics 550 1536 MB\nMemory: 16 GB 2133 MHz LPDDR3\nProcessor: 3,3 GHz Intel Core i7\n\nBroken: (example: 3.0.1, 2.93.8 edbf15d3c044, master, 2018-11-28, as found on the splash screen)\nWorked: 2.83.19 \n\nSwitching to \"Material Preview\" view mode (or Rendered) and any manipulation of a material (like adding new node) takes significantly longer than in previous version. \n\n1. Open attached file\n2. Click \"Material Preview\" view mode\n3. Wait until material is fully visible in the viewport\n\nI know my machine is not a great performance beast and my OS is pretty outdated (but I can't upgrade because of other apps that I use), but you can clearly tell that 3.0.1 works much slower. I tried also 2.93.8 version (LTS tab on blender.org) and the result is the same. Only 2.83.19 worked faster. I attached a video, it's realtime screen recording (the only editing within the video is blurring the mess on my desktop). Also you can find a blend file that was use to perform the test. [performance-test.blend](performance-test.blend)\n\n[blender_performance.mp4](blender_performance.mp4)", "Instancing Geometry on Imported Alembic Particles produces extremely long render times when motion blur is enabled.\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3090/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 535.98\n\nBroken: version: 3.5.1 Release Candidate, commit date: 2023-04-03 16:00, hash: `ef7e83c7ff15`\n\nInstancing geometry on Imported animated Alembic Particles using geometry nodes produces extremely long render times when motion blur is enabled. Turning off motion blur or realizing instances reduces the render time back to its normal average length. \n\n\n", "Modifier are still evaluated even when not required.\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3090/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 496.13\n\nBroken: version: 3.0.0\nWorked: never\n\nModifiers are evaluated, even when blank, for example a lattice modifier without any lattice, would still make depth graph calculation or something like that, that make each action (select, move, rotate, etc.) verry slower.\n\nCreate/ import a large object (60M poly mesh for example).\nAdd lattice modifier, let all option blank. \nDisable viewport visibility of this modifier.\nTry to move/rotate/scale the object in the viewport, the performance is really low.\n\nIn order to transform your mesh, you need to remove all modifiers, and add then back once transform is convenient.\n", "Extremely Long time delay using geometry nodes with simple mesh boolean node 37706ms\nOperating system: Linux-5.13.0-41-generic-x86_64-with-glibc2.31 64 Bits\nGraphics card: Mesa DRI Intel(R) HD Graphics 4600 (HSW GT2) Intel Open Source Technology Center 4.5 (Core Profile) Mesa 21.2.6\n\nBroken: version: 3.2.0 Beta\n\nExtremely Long time delay using geometry nodes with simple mesh boolean node and difference (it takes **37706ms** for the mesh boolean node to do a cylinder difference)\n\n\nI've included the blend file which shows the times when you change the item to do a boolean difference 1 cylinder/count to 2 cylinder/count. \n\n**When I use the Boolean Modifier this time lag doesn't occur**. I've included this in the blend file for testing also.\n\n**PS: this time lag also occurs in Blender 3.1.2**\n\n\n\n[prego lid gn.blend.zip](prego_lid_gn.blend.zip)\n" ]

[ "Optimized per-datablock global undo\nGlobal undo is slow on scenes with many objects. A big reason is that undo writes and reads all datablocks in the file (except from linked libraries). If we can only handle the ones that actually changed, performance would be much improved in common scenarios.\n\nThis proposal outlines how to achieve this in 3 incremental steps. Only 1 or 2 of these steps would already provide a significant speedup.\n\nAt the time of writing there is no concrete plan to implement this, but I've had this idea for a while and it seemed useful to write down in case someone likes to pick it up.\n\n## ~~1. Read only changed datablocks~~ DONE\n\nLinked library datablocks are already preserved on undo pop. We can do the same thing for all non-changed datablocks, and only read the changed ones.\n\nWe can detect which datablocks were changed by checking that their diff is empty. Addition and removal of datablocks also needs to be detected. We need to be careful to exclude runtime data like `LIB_TAG_DOIT` from writing, to avoid detecting too many datablocks as changed. Most changes to runtime data go along with an actual change to the datablock though.\n\nSince unchanged datablocks may point do changed datablocks, we need to reload datablocks in the exact same memory location. We can safely assume the required memory size stays the same, because if it didn't those other datablocks' pointers would be changed as well. In principle other datablocks should not point to anything but the datablock itself, but there may be exceptions that need to be taken care of *(e.g. pointers to [armature] bones hold by [object] pose data...)*.\n\n`Main` fields other than datablock lists may need some attention too.\n\n## ~~2. Optimize post undo/redo updates~~ DONE\n\nThe most expensive update is likely the dependency graph rebuild. We can detect if any depsgraph rebuild was performed between undo pushes, and if so mark that undo step as needing a depsgraph rebuild. This means the undo step is no more expensive than the associated operation.\n\nThis requires the dependency graph to be preserved through undo in scene datablocks, similar to how images preserve image buffers. This would then also preserve GPU data stored in the dependency graph. If the dependency graph supports partial rebuilds in the future, that could be taken advantage of.\n\nThere may be other expensive updates, to be revealed by profiling.\n\n## 3. Rethink how we handle cache preservation across undo steps.\n\n#76989 seems to point to some pointer collision issue, similar to the reason that lead us to add session uuid to IDs when we started re-using existing pointers in undo.\n\nNote that according to the report, there were already issues before undo refactor - hard to tell though if they had the same root cause?\n\nThis needs to be investigated further.\n\n## 4. Write only changed datablocks\n\nWhile undo/redo is usually the most expensive, the undo push that happens after each operation also slows down as scene complexity increases. If we know which datablocks changed, we can only save & diff those.\n\nThis could be deduced from dependency graph update tags. However there are likely still many operations where tags are missing. If it is not done correctly, then instead of a missing refresh there is a more serious bug of not undoing all changes. So this needs verification throughout the Blender code for correct tagging, before it can be enabled.\n\nDebug builds could verify correctness by testing the diff is empty whenever there was no tag.\n\n\n-----------------------------\n\n\nCurrent Status and Plan\n****\n\nThere is an experimental option in 2.83 to address step 1 and (part of) step 2. This was developed in branches and is now in master. Implementation of step 3 has not started.\n\n- [x] [id-ensure-unique-memory-address]() implements the idea of never re-using a same memory address twice for IDs. This is required to allow usage of ID pointers as uids of datablocks across undo history.\n - [x] Add storage of addresses history for each ID (has to be stored in blend memfile, can be ignored/reset when doing an actual blendfile reading).\n - [x] Discuss how to best implement the 'never get same allocated address again' process, current code here is basic, we could probably be much smarter and efficient with some changes to our MEM allocator itself.\n - [x] We are going to try and use session-wise uuids for data-blocks instead.\n - [x] Make new undo behavior controllable from Experimental preferences area.\n- [x] [undo-experiments]() Implements detection of unchanged datablocks at undo, and re-using them instead of re-reading them.\n - [x] Switch to using id addresses instead of names to find back a given data-block across undo history (depends on completion of work in `id-ensure-unique-memory-address` branch).\n- [x] [undo-experiments-swap-reread-datablocks]() implements re-using old ID address even for changed IDs that need actual reading from memfile. Also adds support for more refined depsgraph update flags handling from undo. Both changes combined allow to reduce dramatically depsgraph work needed during most undo steps.\n - [x] Switch to using id addresses instead of names to find back a given data-block across undo history (depends on completion of work in `id-ensure-unique-memory-address` branch).\n- [x] Investigate crashes due to invalid or used-after-free memory in depsgraph after undo step.\n- [x] Investigate why some undo steps are still using full reload of the memfile. Some may be necessary, but think we can improve here.\n - [x] In particular, the absolute last undo step (the first to be undone) is always seen as 'all data are changed', for some reasons...\n- [x] Investigate how to discard runtime data from 'data is changed' detection (probably by nullifying those pointers at write time?).\n- [x] Fix slow undo of adding/duplicating/deleting datalocks. The undo diffing does not appear to be smart to match up the right datablocks in such cases, causing many unrelated datablocks to be marked as changed, which also means excessive memory usage. The `session_uuid` could be used to find matching datablocks." ]

[ "Edge/Vert Slide fail to properly \"redo\"\nCurrently vertex & edge-slide use the mouse input to calculate which edge to slide, the direction to slide, and the **Even** toggle is set at runtime.\n\nUsing operator redo is not re-running with the same options, so redoing the action fails.\n\nWe could simply disable redo with this operator, however being able to toggle **Correct UV** is important.\n\nAdding this report to merge in duplicates, since there are various reports on this, which all have the same root-cause.", "Rigidbody crash undoing while animation plays (which was started at frame 1 or below)\n[blender_system_info.txt](blender_system_info.txt)\n\n[blender_debug_output.txt](blender_debug_output.txt)\n\n[animation_john2.blend](animation_john2.blend)\n\nOperating system: Windows 10\nGraphics card: Nvidia 2070 Super\n\nWorked: -\n\nBlender crashes when object is rotated to a steep vertical angle and animation player is run.\n\nFor animation_john2.blend\n- Play Animation\n- Select a bone (could be any other operator as well I think)\n- Undo (this will also jump back to frame 0 -- where we started -- which is the important part, starting animation at frame 2 or higher does not result in the crash)\n- Redo\n\nRepeat until crash. Occurs between 2-5 repeats.\n\nAlso occurs at other similar high angles.", "Outline of plane in Object mode does not Render after applying transforms\nBroken: Current main.\nWorked: N/A.\n\nOutline of plane in Object mode does not Render after applying transforms and looking Down the plane in top down view or any view.\n\n1. Delete the default cube.\n2. Add a Plane.\n3. Numpad 7 for top down view.\n4. selct the plane and rotate on x axis 90 degrees\n5. Apply all transforms.\n\nFrom Blend File Just Select the plane and press numpad 7 to go to Topdown view and Out line should stop rendering.", "Fluid simulation result not visible after reloading file\nOperating system: MacOs\nGraphics card: Intel Iris Plus Graphics 640\n\nBroken: 2.91\nWorked: -\n\nHello. After I opened project, that I closed before, the fluid simulation didn't exist. It is very easy, to know the issue. Thanks for attention.\n- New file\n- Quick liquid effect on default cube\n- Save file\n- Close Blender\n- Open the same file again\n- The domain is just a smooth shaded cube until you play the file long enough for it to loop back to the first frame.", "Viewport display: in front should be listed under visibility as it also affects final render\nOperating system: macOS-11.4-x86_64-i386-64bit 64 Bits\nGraphics card: AMD Radeon Pro 5500M OpenGL Engine ATI Technologies Inc. 4.1 ATI-4.5.14\n\nBroken: version: 3.0.0 Alpha\nWorked: unknown\n\nUnder object properties, viewport display: in front affects both the visibility in the viewport and in the final render, and as such should probably be listed under visibility rather than viewport display.\n\n-", "GPencil: New operator to set Stroke start point\nThis operator allows to set the start point for any cyclic stroke.\n\n[SetStartPoint.mp4](SetStartPoint.mp4)", " Custom normals polishing\nThere are a few issues with custom normals workflow at the moment. For example it needs auto-smooth at the moment depending on the artist pipeline.", "Poor stroke quality when a model is far from the world origin\nOperating system: Win 10\nGraphics card: Nvidia Rtx 2070\n\nBroken: 2.83, master\nWorked: perhaps never?\n\njittery strokes when mesh is far from the world origin in perspective mode\n\nHow to reproduce this bug : \n- Create a grid (set the location to x : 2000 m and y : 1000 m)\n- Create a multires modifier, with 5 subdiv.\n- Sculpt with the clay brush\n- In Perspective mode (left stroke) the stroke is ugly\n- In ortho mode (right stroke) the stroke is clean\nTo fix the stroke poor quality, i had to fine tune the Clip start and the clip end values of the 3d view. \n[#102747.blend](T102747.blend)\n", "Freestyle wireframe-only rendering is broken with Ztrans faces (lines disappear when Ztrans alpha and Ztrans spec are set to zero / lines behind transparent faces are not rendered / marked edges are not sometimes rendered when faces overlap)\nOSX 10.11.2 / Macbook Pro Retina 15\" Mid 2014 / GeForce GT 750M\n\nBroken: 2.76b (also broken in 2.76)\n\nRendering wireframes with controllable wire thickness requires using Freestyle but Freestyle requires filled faces even when rendering only edges. To only render edges and make the faces invisible one must set Ztrans alpha to zero.\n- When other line objects are behind line-only objects their lines are culled (however ordinary objects are rendered properly)\n- When both Ztrans alpha and spec are set to zero no lines are rendered\n- When faces overlap only some marked edges are rendered (even when faces are set to be transparent)\n\nObserve and render the attached .blend file.\n[freestyle-lineclipping.blend](freestyle-lineclipping.blend)", "2.83 3D viewport performance\nOperating system: Ubuntu 20.04 LTS\nGraphics card: Mesa DRI Intel(R) HD Graphics 4600 (HSW GT2), version: 4.5 (Core Profile) Mesa 20.0.4\n[system-info.txt](system-info.txt)\n\nBroken: 2.83\nWorking: 2.82\n\nProblem seen in Workbench, Wireframe and Eevee.\nViewport performance it's really bad, I'm moving around just the initial cube scene at 15-20 Fps, older Blender versions movement is perfectly smooth.\nNo matter the scene/blend file, 2.83 offers worse viewport movement performance compared to older versions,\nIt's important to mention that only the viewport movement is being affected by this performance drop, everything else including rendering works smooth as butter just like older versions or even better. Splitting up the scene in windows kinda fixes this problem.", "Visual glitches With adaptive domain enabled when object is duplicated or transformed\nOperating system: windows 10\nGraphics card: 3080ti\n\n**Blender Version** 3.3 lts\n\nVisual glitches With adaptive domain enabled when object is duplicated or transformed\n\n- Open default scene\n- Add quick smoke on cube\n- enable `adaptive domain`\n- play animation\n- move domain around or duplicate it (snaps back to original place)\n- play animation again (will move to transformed position)", "Blender 3.2.1 / 3.4 - Cycles - Persistant Data and Motion Blur artifacts on GPU\nOperating system: Windows 10, Ubuntu 21.04\nGraphics card: GTX 1080, RTX 3080\n\nBroken: 3.2.1, b5e92c3dfe\nWorked: 2.79\n\nPersistant data is bugs out during renders sometimes.\n\n1. Open attached blend file in Blender 3.2.1\n2. Press f12\n3. Press right on the keyboard (frame 771)\n4. Press f12\n5. Press right on the keyboard (frame 772)\n6. Press f12\n7. Notice how its not as expected.\n8. Press f12 without framing forward\n9. Notice how it is fine\n\ngood\n\nbad\n\n\nThis means when rendering animations, it can produce frames which bad, which then require more re-renders.\n\nWindows and Linux have both been tested and results are the same.\n\nTested CUDA, Optix and both have issues. CPU renders fine. \n\nDisabling Persistant Data renders fine.\n[bug.blend](bug.blend)\n\nNOTE: We had #95475, #96822 already for Persistent Data and Motion Blur issues (these should be fixed), I can repro this case though.\n\nEDIT:\nAs another task #100729 has been merged into this file, there is another test file [BUG.blend](BUG.blend)\n", "EEVEE - Screen space reflections artifacts\nOperating system: Linux-5.3.0-7625-generic-x86_64-with-debian-buster-sid 64 Bits\nGraphics card: GeForce GTX 1080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 440.44\n\nBroken: version: 2.83 (sub 2)\nWorked: ?\n\nSSR shows some strips when I have bad data for normals.\n\nOpen the file:\n[ssr_sripts.blend](ssr_sripts.blend)\n\nIf you click anywhere multiple times, viewport gets darker and darker.\n\nBefore:\n\n\nAfter clicking the glossy node header a few times:\n\n", "Render problems\nOperating system: Windows 10 Pro Ameliorated\nGraphics card: RTX 3070 driver version 472.12\n\nBroken: v3.6 and v3.61\nWorked: v3.51 <- Using the same file created in v3.61\n\nWhile rendering a frames of an animated character, the current rendered animation frame usually shows in the 3D viewport in v3.51. However, in the broken version, it fails to show the current frame being rendered, nothing gets updated visually. Then eventually in the 3D viewport the character goes into its T pose for a few frames then the character starts losing visibility of its layers, at that point Blender crashes. \nI will create a .blend file if I am the only one getting this bug.\n\n", "Render: Option to use fullscreen depthbuffer for tracing\nAdd an option to allow the use of a more precise depth buffer when doing screen space effects.\n\nThat may solve some issue regarding noise or self intersection." ]

[ "\"Implement resolution divider in the Blender Internal\" plus Freestyle calls freestyle render over and over and over for each subdivided resolution\nHello and thanks for the \"resolution divider in blender internal\" (Differentail #[D609](D609))... works great!\n\nHowever, if you also turn on \"Freestyle' render option and then turn on \"rendered\" preview in View3d, the freestyle renderer is called once for each lower resolution, which make preview unreasonably slow.\n\nIt is easy to reproduce with factory settings like this:\n\n1) start new factory startup file\n2) enable \"Freestyle\" button\n3) enable \"Viewport Shading | Rendered\"\n4) remove the default cube\n5) add a Suzanne.\n6) add a 3x subdivision surface modifier\n7) (try) to navigate in Viewport\n\nIt does work as \"expected\" but what it does is call freestyle renderer over and over. \n\nFor just the default cube freestyle is quick enough but complex scenes causes a \nnoticeable delay in navigating as the freestyle renderer is called multiple times.\n\nI suggest that Freestyle is only called once after the highest resolution is complete." ]

[ "When double clicking to selecting text the first characters aren't selected\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 536.23\n\nBroken: version: 3.6.2\n\nWhen you double click on any text field with the intention to select all of the text when you start dragging to select the other words the first characters dont stay selected\n\ndouble click on a word and then drag your mouse to select the next or previous words. Some of the characters will always become unselected when you drag the mouse as seen in the attached video\n\n\n\nusually when double clicking a word the entire word stays selected regardless of where the initial mouse location was when double clicked but in blender the characters before or after the mouse location when double clicked don't remain selected when you drag to select more words", " Alpha channel of Tint Modifier don't Work with Fill material but work with stroke material.\nOperating system: Windows-10-10.0.19043-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1050/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.94\n\nBroken: version: 3.3.1\n\nhi, it's been a while i didn't make a report for a bug, :) , well i'm back.\nthe Tint Modifier work well with `Stroke `material for every color channel R,G,B,A , but for Fill material we have problem with Alpha channel that don't work at all.\nthe first video show stroke works well with ALpha channel :\n[2022-10-16 13-12-36.mp4](2022-10-16_13-12-36.mp4)\nthe file\n[aplha color of tint modifier work well with stroke material.blend](aplha_color_of_tint_modifier_work_well_with_stroke_material.blend)\nsecont video show the problem with FILL material and alpha channel:\n[2022-10-16 13-14-27.mp4](2022-10-16_13-14-27.mp4)\nthe file\n[aplha color of tint modifier don't work with fill material.blend](aplha_color_of_tint_modifier_don_t_work_with_fill_material.blend)\nhope to get fixed.thank you for reading it\n\n", "No \"affect alpha\" in brush settings in the Image Editor\nOperating system: Linux-5.3.0-40-generic-x86_64-with-debian-buster-sid 64 Bits\nGraphics card: GeForce GTX 980 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 435.21\n\nBroken: version: 2.82 (sub 7)\n(as well as the latest alpha: version: 2.83 (sub 5))\nWorked: 2.80 (Checkbox existed in Image Editor brush settings, but had NO effect on brush behavior)\n\nIn the 3D Viewport as well as the Properties Editor while in Texture Paint mode, all brushes offer the \"affect alpha\" option in the \"Advanced\" section of the Brush Settings. In the Image Editor that option is missing from the \"Advanced\" section for no apparent reason.\n\n\nOpen Blender with factory settings. Switch to \"Texture Paint\" tab. Compare the \"Advanced\" section of the active brush's tool settings in the Image Editor to that in the 3D Viewport or Properties Editor.\n\nSurely this has to be a well-known issue and exists only due to technical limitations of some kind.\nWhat should I tell my students why this inconvenience exists and when it will be addressed?", "Grease Pencil Sculpt/Vertex Paint Tool is exponentially slower when using tablet instead of mouse\nOperating system: Win10\nGraphics card: RTX 3070 FE\nTablet: Xp-Pen Deco 03 (Wireless)\n\nBroken: 3.1\nWorked: N/A\n\nWith high-geometry Grease Pencil strokes, the Sculpt Tool (and the Vertex Paint Tool) is exponentially slower when using tablet. \n\nVideo explanation:\nDavVZXsxBmI\n\nBlend File:\n[SculptToolExponentiallySlowerWithTabletThanMouse.blend](SculptToolExponentiallySlowerWithTabletThanMouse.blend)\n\n", "\"Dots\" stroke method is ongoing even without moving the cursor when using a pen\nOperating system: Linux-5.15.0-50-generic-x86_64-with-glibc2.35 64 Bits\nGraphics card: Quadro GP100/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 510.85.02\n\nBroken: version: 3.4.0 Alpha\n\nThe Dots stroke method is meant to apply the strength of the brush on each stroke dab.\nThis is working like intended when using the mouse. If you don't move the cursor, the stroke stops.\n\nBut when using a pen, even when holding the pen perfectly still, the stroke is ongoing on the stop.\nThis way the stroke method is working very much like the Airbrush stroke method,\n\n- Add a cube and subdivide it many times\n- Use the draw brush in sculpt mode\n- Set the stroke method to \"Dots\"\n- Draw a stroke and hold it on a single spot without moving the cursor or releasing LMB\n- Try this with a mouse and a pen and see how they compare.", "GPencil: Screen-space stroke size incorrect\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTS 450/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 391.35\n\nBroken: version: 2.92.0 Alpha\nWorked: none found\n\n**Description**\n\nWhen drawing a stroke with a defined pixel size and stroke-thickness-space is set to screen space, the radius of the drawn stroke should measure the set pixel size. This is not the case. The strokes radius measures half the size.\n\n**Steps to reproduce**\n\n - Create a grease pencil object\n - Set the stroke-thickness-space to screen space in objects grease pencil settings\n - Set the brush size to 100px\n - Draw a single dot.\n - Print the screen\n - Measure the dot size in any image editing app.\n # You will see that the radius of measures 50px and the diameter measures 100px. Instead of 100px and 200px respectively.\n\nThis issue can be further seen by trying to change the brush size using the `f` key. The brush size cursor is drawn at the correct size but is double the size of the stroke.\n\n**Investigations**\n\nThis relates to my original report #82707 wherein I thought the cursor size was incorrect. I have now had time to review the code enough to understand how the cursor is drawn; and how the brush size is defined and passed to the rendering engine.\n\nI have confirmed that the brush size takes the following path.\n\nBrush.size user setting: RADIUS → bGPDstroke.thickness: RADIUS → gpencil_vert.glsl thickness: RADIUS\n\n*Simplified for explanation purposes. Path steps missing.*\n\nI originally thought thickness was defined as a diameter but this is not the case.\n\nI also thought this would be a noticeable fix but I haven’t found where the issue lie yet so I thought I would report again. \n\nNote: When drawing a stroke with stroke-thickness-space set to world space the stroke is mapped to the users screen correctly. When world space is used a stroke with a brush size of 2000px is drawn with a radius of 1m in world space as defined internally. \n\n[Bug-GPencil-Screen-Space-Stroke-Size-Incorrect-Report-VID.mp4](Bug-GPencil-Screen-Space-Stroke-Size-Incorrect-Report-VID.mp4)\n", "Grease Pencil Fill Tool Breaks After Scroll-Wheel Zooming Window\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 980/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 457.30\n\nBroken: version: 2.91.0 Beta\nAlso tested in 2.83.9 & 2.92 Alpha on Windows, and 2.91 Beta on Mac with same results.\n\nIf an 'unclosed' area made with grease pencil lines on its own layer overlaps and is 'closed' with grease pencil lines on a second layer, zooming the 2d workspace in and out will cause the fill tool to 'break' and fill the whole screen; even when the fill tool is set to: Advanced-Layers-Visible. \n\nCreate closed shape on one layer. Create 'open' shape on second layer that overlaps and is closed by grease pencil lines from the first layer. Zoom in and out in workspace while using the fill tool in a separate 'fill' layer. \nPlease see attached .blend file and screen capture: \n\n[FillTest.blend](FillTest.blend)\n\n[FillTest.mp4](FillTest.mp4)\n\n\nI really hope this isn't just user error. Thank you for taking a look!\n", "Nodes: Different mouse drag effect for float properties\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 980 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 522.25\n\nBroken: version: 3.5.0 Alpha\n\nDrag step inconsistent of float data type that between of geometry nodes in modifier and editor.\nAnd float of vector also.\nThis caused the mouse to drag far to reach the maximum value when going from 0-100.\n\n\n[drag_step.blend](drag_step.blend)\n\n", "Sculpt brush Reverts to Default When Undoing\nOperating system: Windows-10-10.0.25336-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2080 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 531.41\n\nBroken: version: 3.6.0 Alpha\nWorked: 2.79\n\n**Description of error**\n\nWhen using a non-default brush such as clay strips, then undoing, the brush switches to the default one (draw brush). Now this is frustrating to the user because you need constantly switch back to your brush after every ctrl + z. Stupid bug that interrupts the workflow.\n\nNow this bug is not limited to sculpt mode, it happens in vertex painting, weight painting, texture painting and hair sculpt mode.\n\nHowever this bug doesn't happen to most users because for it to happen you need to have two main windows.\n\n* Add a new main window (`Window > New Main Window`) and switch the workspace (say \"shading\")\n* In primary main window: switch to sculpt mode\n* Choose clay strips\n* Press ctrl + z\n* Observe the brush is draw brush and not clay strips 😵‍💫\n\nNote: I think this has to do something with workspace since switching them can fix this bug but also crash blender\n\n[video](https://projects.blender.orgattachment)", "Liquid Emission unlimited\n\"Surface Emission\" value ignored I suppose, see the screenshot and the project file below:\n\n[test_emission.blend](test_emission.blend)\n\nThat's looks the same or almost same for \"Surface Emission\"=1,10,100,...\n\nBlender Version 2.90.0 (2.90.0 2020-08-07) on MacOS Mojave.", "Setting the brush radius unit to \"scene\" does not automatically use correct stroke spacing distance\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce RTX 2070 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.38\n\nBroken: version: 2.91.0 Alpha\n\nThe brush settings have an option to fix the radius to either the view or scene, called \"Radius Unit\".\nIf the Scene setting is used, it will not automatically use the correct spacing distances for the stroke.\nThis needs to be toggled manually with separate stroke setting called \"Spacing Distance\" from view to scene.\n\nIf the viewport orientation is very close to the world origin, rotated towards an object that is further away from the world origin, it will lead to a much higher spacing distance.\n\n[2022-05-13 10-24-15.mp4](2022-05-13_10-24-15.mp4)\n\n**Exact steps to reproduce**\n- Move the default cube 4 units away from the world origin\n- Add a multires modifier and subdivide 5 times\n- Enter sculpt mode\n- Use the draw brush\n- Switch the Radius Unit setting from View to Scene\n- Orient your viewport so that you are positioned very close to the world origin, looking at the cube\n- Sculpt on the cube (Notice the very high spacing distance)\n- Rotate your view around to the other side of the cube (Now the viewport is positioned further away from the world origin)\n- Sculpt on the cube from that angle (The stroke distance should be much smaller now)\n\n- To fix the issue switch the stroke setting \"Spacing Distance\" from View to Scene\n- Retry sculpting from the same angles", "Grease Pencil vertex paint select is using sculpt mode selection mode\nOperating system: Linux-6.4.6-gentoo-x86_64-AMD_Ryzen_Threadripper_1950X_16-Core_Processor-with-glibc2.37 64 Bits, X11 UI\nGraphics card: AMD Radeon RX 7900 XT (gfx1100, LLVM 15.0.7, DRM 3.52, 6.4.6-gentoo) AMD 4.6 (Core Profile) Mesa 23.3.0-devel (git-50c29e1ffa)\n\nBroken: version: 4.0.0 Alpha\nWorked: \n\nWhen selecting with mouse clicks the selection mode in the header determines which component is selected.\nIn Grease Pencil sculpt mode this behaves as expected. If it's set to \"points\", selecting will select points. If it's set to \"stroke\" it will select entire strokes.\n\nWhen switching to vertex paint mode this is not the case. Instead it will select components based on the selection mode that is set in sculpt mode.\n\nThe ideal behavior imo would be to have all grease pencil modes share the active selection mode.\nThis makes it easier to manage and simpler to use.\n\n- Change the keymap preferences to right click select (Selection shortcuts in left click select are partially broken atm)\n- Add a Gpencil Monkey\n- Switch to Sculpt Mode\n- Enable point selection mode\n- Select some points with `RMB` and `Shift RMB`\n- Switch to Vertex Paint mode\n- Make sure to use stroke selection mode\n- Select some strokes with `RMB` and `Shift RMB` (Notice how it will instead select points)\n\n", "If Grease Pencil is set to \"Stroke Depth Order: 3D Location\", blending always with the background and not with previously painted strokes\nOperating system: Ubuntu 20\nGraphics card: GT 1030\n\nBroken: 3.0\n\nIf Grease Pencil is set to \"Stroke Depth Order: 3D Location\", brush color blending seems to use only 1-bit transparency in many cases.\n\nIt is a known issue (see T75684). The linked bug is closed as a presumed limitation of Grease Pencil. However, what the dev calls a limitation there seems to be an unrelated feature (from user perspective, possibly not from a technical perspective). Namely, \"the strokes are always drawn as a billboard face to point of view\" prevents them from looking flat, despite being drawn with a brush defined in 2d space.\n\nI do not know Blender internals, but visually it looks like the problem is elsewhere: blending of colors in \"Stroke Depth Order: 3D Location\" binarizes (at least in some cases) the brush' alpha (opacity) channel. In effect, a hardly visible pixel may completely replace what is behind it. Possibly there is a way to make the scene look correctly at least when rendered.\n\nSee the model attached to the linked bug.", "Texture Paint mode tools swap when exiting maximized area\nOperating system: Linux-5.17.5-76051705-generic-x86_64-with-glibc2.35 64 Bits\nGraphics card: Quadro RTX 6000/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 510.68.02\n\nBroken: version: 3.3.0 Alpha\nWorked: unknown\n\n[texture_paint_tool_swapping-2022-05-10_12.52.16.mp4](texture_paint_tool_swapping-2022-05-10_12.52.16.mp4)\n\n- Go to texture paint mode\n- open second image editor\n- maximize the area of the 3D viewport\n- change the active tool\n- un-maximize the area\n", "GP Fill Tool not working on Mac with Metal GPU Backend\nOperating system: macOS-13.4-x86_64-i386-64bit 64 Bits\nGraphics card: AMD Radeon Pro 5500M OpenGL Engine ATI Technologies Inc. 4.1 ATI-4.12.7\n\nBroken: version: 3.5.1\n\nWhen my GPU Backend is set to Metal, the Fill Tool in Grease Pencil is not working. It either doesn't fill, or fills randomly (and yes, my shapes are closed). If I switch my GPU Backend to OpenGL, the Fill Tool works as expected (although I have to click twice)\n\n<video src=\"attachment\" title=\"IMG_8032.MOV\" controls height=\"360\"></video>\n\nWith the Metal Backend turned on in Preferences...\n1. I open a new 2D Animation file.\n2. I select Ink Pen as my brush type.\n3. I draw some simple shapes.\n4. I switch to the Fills layer.\n5. I switch to the Fill Tool.\n6. I switch to a fill material.\n7. I click inside the drawn shapes.\n8. Some shapes fill and some shapes don't.\n\nIn this file there is a case where the Fill tool works with OpenGL but not with Metal: [fill_metal_bug.blend](attachment)" ]

[ "Sculpt pressure sometimes start too big\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 770/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 461.40\n\nBroken: version: 2.93.0 Alpha\nWorked: version: 2.93.0 Alpha\n\nWhile quickly sculpting strokes with a tablet, Blender will sometimes start with a bigger size *(from pressure)* than the following areas of the stroke. \nThe inconsistency is like either a delay in finding the correct starting pressure but leaving the initial one, or that Blender would previously find the wrong starting pressure and now behaves differently. \n\nThe way I would describe how I'm drawing strokes is like flicking paint from a brush but trying to make it light. \nFast with the intent of being soft. \n[2021-02-18 02-36-09.mp4](2021-02-18_02-36-09.mp4)\n" ]

[ "Grease pencil brush sizes incorrect\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce RTX 2060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 457.30\n\nBroken: version: 2.92.0 Beta\n\nGrease pencil brush sizes are not accurate, not even close. A 500px radius(!) size brush (which should be 1000 pixels wide since we're talking about a \"radius\") has actually about a 50px radius.\n\nOpen a new 2D animation file, increase brush size to the max of 500px, draw a line. Does that look like 1000pixels wide to you? Or ist it me? Maybe I need to work all the way zoomed into the canvas to see the accurate size? How big is the factory canvas?\n\n", "Resize a incoming fluid cause fluid to stop flowing in\nOperating system: windows 11\nGraphics card: RTX\n\nBlender 3.6.1\n\n\nGood morning\n\nFirst thank you for the work done on Blender.\n\nI think I found a bug. In fluid simulations when we resize an incoming flow, it often happens that the flow is no longer able to generate a fluid.\nI have seen this with eco-spheres\n\nThanks in advance .\n\nBased on the default startup or an attached .blend file (as simple as possible).\n\n", "Curves sculpting collision with surface.\n", "Grease Pencil: various issues\n*Note: this tasks can be split up once we get feedback on issues, for now this is to discuss changes to make.*\n\n----\n\n\n- Brush pixels are not pixels,\n```\nshouldn't we use a meaningful value?\n(pixels or blender units).\n```\n\n- Grease pencil brushes have material pointers which are globally overwritten.\n\n```\nWhen you set a material, all brushes are looped over and have their material values clobbered.\n```\n\n```\nWhy have a material attached to the brush at all? This could be a scene level setting (or at least not stored per-brush).\n```\n\n```\nSee: `BKE_brush_update_material` called by rna `rna_Brush_material_update` when setting any material.\n```\n\n- Grease pencil sculpt uses it's own brush system,\n```\nis there a good reason for this?\n```\n\n```\nSince these brushes are fairly advanced (pressure & curve mapping options)\nwouldn't it be better to use the existing brush system?\n```\n\n *Note: while particle system for eg: has 'simple' brush types like this, grease-pencil sculpt brushes are more advanced and duplicate more of what we have for regular brushes used elsewhere.* \n\n- Grease pencil weight brush seems to have no way to paint a weight value (only add weight).\n\n```\nAs with mesh weight paint we should be able to paint a weight.\n```\n\n- In edit-mode, vertex selection exists but is hidden by default (0.0 opacity).\n\n```\nShouldn't they be visible?\n```\n\n----\n\n### Lower Priority\n\n- Internally grease pencil sculp is called:\n\n - grease pencil edit (`GP_BrushEdit_Settings`, `GP_EditBrush_Data`, `GP_BRUSHEDIT_FLAG_APPLY_UV`)\n\n```\nThis is highly confusing since we have grease pencil edit-mode, can't we rename these `GP_Sculpt*`. ?\n```\n\n - grease pencil paint (`GPENCIL_OT_brush_paint`)\n\n```\nPaint is easily confused with draw, seems this is for sculpt/weight painting.", "Weight Paint gradient operators wait for mouse input\nOperating system: Linux-6.4.6-gentoo-x86_64-AMD_Ryzen_Threadripper_1950X_16-Core_Processor-with-glibc2.37 64 Bits, X11 UI\nGraphics card: AMD Radeon RX 7900 XT (gfx1100, LLVM 15.0.7, DRM 3.52, 6.4.6-gentoo) AMD 4.6 (Core Profile) Mesa 23.3.0-devel (git-50c29e1ffa)\n\nBroken: version: 4.0.0 Alpha\nWorked: --\n\nThis is based on feedback from #108791\nWhen using the gradient operators in weight paint mode, they don't start on shortcut press but require a mouse press to start and release to confirm.\n\nThis worked well with the previous default shortcut with mouse clicks but when reassigned to keyboard shortcuts this is very inconsistent with other operators in Blender and seems unintentional or outdated.\n\nA good counter example for a better behavior is \"Expand by Topology\" in sculpt mode, which only waits for mouse input if started from a menu. If executed with the keyboad shortcut `Shift A` it will start and then confirm with `LMB`.\nIf the shortcut is remapped to a mouse press, it will start on press and confirm with release. \n\n- Enter Weight Paint Mode\n- Press `Shift A` and use the operation\n- Try the same in Sculpt Mode\n\n", "Most sculpt tools do not work while model is animated\nLinux 64-bit\n2.93.2\nMost sculpt tools do not work as expected when the object is spinning like it's on a pottery wheel. Some tools like the clay tool work okay, but other tools like smooth and scrape do not. I think it has to do with the stroke method; the stroke accounts for screen space but not the local transform of the object if it's moving during a stroke. Video of my attempt here:\n[pottery.mp4](pottery.mp4)\nI made this as a joke and it looks like garbage, but unfortunately it went viral and everybody thinks it's a great idea so now more people are gonna try it and complain that it doesn't work. It's a valid workflow that you have to account for now apparently. I'm really sorry.\n - Open the attached file\n - Play the animation\n # Try to sculpt the thing while it spins\n[bowl.blend](bowl.blend)\n", "Prompt when entering sculpt mode on non-uniform scaling\n### Motivation\n\nIn the sculpting workflow it is a common step to always apply the scaling of an object so that the tools, brushes and operations work as expected.\nThis is adding extra steps for the user for something that could be automated or fixed. It also confuses most beginners who need to learn this the hard way.\n\n### Proposal\n\nAdd a prompt when entering sculpt mode on an object that has non-unform scaling. This popup will behave similarly to enabling dyntopo while having object data that could be destroyed by using that option:\n\n\nThis popup will ask the user \" Warning!\"\n\"You are entering sculpt mode on an object with non-uniform scaling values. We recommend to apply the scaling of the object before proceeding.\"\n\nIt will then offer 2 options: \"Apply Scale\" which is highlighted and will be used when pressing enter, and \"Don't Apply\" which enters sculpt mode regardless without applying the scale.\n\nIf the scaling is applied the normals should also be recalculated if the scale was negative. This will save the user some trouble of going into edit mode and doing this manually.\n\n### Issues\nCertain bugs would lead to progress being destroyed such as #74361.\nThese need to be resolved first.\n\nAlso ff the user wants to keep the scale intact it will also prompt them every time they enter sculpt mode. Potentially this is not ideal but still better than not having the prompt.", "Canvas picker doesn't get immediate update with change of active canvas\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2060 SUPER/PCIe/SSE2 NVIDIA Corporation 4.6.0 NVIDIA 536.40\n\nBroken: version: 4.0.0 Alpha\nWorked: n/a\n\nCanvas picker was introduced with 3b5df8a7ea\n\nCanvas picker doesn't get immediate update with change of active canvas. It only gets updated after other action after that.\n\nex: weight paint\n\n\n\nIn case of sculpt paint mode it's dimmed until the first stroke of paint and other update action (hover on canvas picker in this case) after that.\n\n\n\n", "GPencil: Fill tool Ignore Transparent strokes setting broken\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.94\n\nBroken: version: 3.5.0 Beta\n\nFill tool not ignoring transparent strokes when the settings is enabled. When you enable Show boundary lines, it shows that transparent strokes are not a boundary but the fill still uses them as boundaries.\n\n- Open attached file and try the fill tool. Under the Advanced menu 'Ignore Transparent' and 'Show Lines' are enabled.\n\n", "Sculpt Mode Multires Design\nThe purpose of this task is to describe the main uses cases and functionality that should be provided by a Multires sculpting system, as well as to discuss a plan on how to implement that in Blender.\n\n\n## The main use case of Multires\n\n### The problem Multires solves\n\nWhen you are working on any art medium (this also applies for painting), you should always try to work from the general shapes to the details. In case of digital sculpting, this means that you should always try to work with the minimum number of vertices that represent the level of detail you are trying to add to the model.\nThis is also true for the current workflow we have implemented (The Voxel remesher). By default, it provides a low poly mesh (about 10K vertices), that you should deform trying to get main shape of the object you want. After that, you increase the resolution of the remesher until you have enough vertices to represent the next level of detail you want to work on, and you continue from there.\nThe reasons you want to work this way are:\n\n- Performance: It is faster (more responsive, without lag) to deform a mesh with 10K vertices than a mesh with 25M vertices. If you use a big brush size needed to modify the object proportions on 25M vertices at the same time, it is going to lag. (Unrelated note, but we should not try to optimize Sculpt Mode to support these kind of operations as they don’t make any sense with a Multires system in place).\n\n- (Most important) Noise: If you sculpt with more vertices than needed for the size of details and shapes you are trying to get, you are going to end with a lot of uncontrolled shape noise produced by the tools. This is not a specific problem with Blender brushes and tools, this happens in any software and in every sculpting system. If you use the Draw brush to add some volume on a big area directly onto a 25M vertices mesh, you are going to need need to clean manually all small uncontrolled imperfections that big stroke stroke produced if you want that to be the final surface. If you start sculpting from a sphere with 1M vertices, you are going to spend more time cleaning shape noise than sculpting or the sculpt is just going to look bad. This is why nobody works this way. You don’t generate unintended shape noise if there aren’t enough vertices to represent it.\n\n\nThat is the main problem Multires sculpting solves. With Multires, you are able to choose at any moment the level of detail that is appropriate for the change you want to make in you model, solving both the performance and the noise issues. This gives an artist a huge amount of freedom as you know you can always go back to tweak low frequency details (proportions, general volumes) after sculpting mid/high frequency details (wrinkles, skin alphas...) if you consider that they don’t look right, without loosing any work.\n\nWithout this, the rest of the sculpting workflows (the Voxel remesher, Dyntopo) are the same as sculpting with real clay or marble, you can only sculpt bigger first shapes, then mid details and them the small details. If it does not look right, there is not way to fix it, you need to delete what looks wrong and start again. This is why those workflows are often used for concepting general shapes, and you want two switch to Multires as soon as you can. Sculpting a complex model without using Multires, on any software, is extremely (and unnecessarily) hard and it requires an insane amount of skill from the artist to workaround the limitation. It is like sculpting without using undo. It is too risky to create a high poly sculpt with projected alphas without this feature.\n\nIn order to solve this problem, the main thing Multires needs to guarantee is that when you sculpt shapes on your model in the appropriate resolution level, all the other levels are updated accordingly to reflect the changes you did. This is what people often refer as ¨detail propagation¨.\n\n### Multires Design\n\nWhen designing a sculpting Multires system that provides the previously mentioned functionality, We need to take the following points into consideration:\n\n1. All Multires subdivision levels contain valid information (sculpted shapes, mask, colors...) that should never be lost or damaged. \n\n2. The user wants to modify any subdivision level, at any time, without restrictions (there should not be limits of how much the mesh should be deformed or “not allowed modifications”).\n\n3. The user wants to delete and rebuild higher/lower subdivision levels in he/she considers that the shape information that level contains is not longer relevant. \n\nFor example, if the user wants to do a high poly sculpt, he/she switches to the higher level subdivision possible that the computer allows to create the high frequency detail projecting alphas. When that is done, the user may want to switch to a lower level to tweak proportions, or to a mid level to tweak small volumes with a soft brush. All the work the user put into applying the alphas should be propagated with the lower subdivision level changes with the Multires functionality and the model should look fine.\n\nIf, for some reason (a bug or an design issue in the implementation that did not consider point 1), switching and modifying lower levels may introduce mesh artifacts in the higher ones (the ones that contains the alpha projections and high frequency details), it will destroy all the work done previously by the user in that levels. \n\nIn that case, the user is again limited to not going back to select the mesh resolution needed for the modifications he/she wants to make in the mesh. A Multires system like this is not solving the intended problem and there is no reason to use it. It provides the same functionality as any of the other sculpting workflows (Voxel, Dyntopo), but even more limited as the user can’t modify the base topology to adapt it to a new shape. It does not have any reason to exist.\n\nIf you project requires you to use a fixed predefined topology in your model (for animation, for example), subdividing the mesh manually and deforming it with a cage or a datatransfer provides the same functionality. You can also get a lower subdivision level by duplicating the model and using unsubdivide if you need it for texture baking or animation. This is not the main problem Multires should be trying to solve.\n\n \n**Any unwanted mesh artifact in the detail propagation, no matter how small it is, invalidates all Multires sculpting features and makes it not functional for its intended use case, as it leads to irreparable data corruption.**\n \n\n### Current Multires Modifier\n\nConsidering the previously design and use case description as valid Multires sculpting system:\n\n1. Multires in 2.81 does not support switching levels. This invalidates the feature completely and makes it not functional, but prevents the users to corrupt the mesh data if they use it. This is like not having Multires sculpting at all.\n2. Multires in 2.80 creates artifacts and vertex explosions when modifying any level, creating irreparable mesh data corruption. Releasing Blender with Multires in this state and without any visible UI warning could have destroyed the projects of many users in an irreparable way. Luckily, most users were not using it to create their models in the previous versions because it was also not working as intended. \n3. Multires in 2.79 and before creates a visible grid pattern on higher levels when modifying lower ones. This introduces unwanted shape noise and effectively destroys valid sculpted information, making it not functional for its intended use case. Users are limited to work considering of how much they can deform lower levels by the amount of noise they are predicting Multires will produce in the higher ones. This is not something that should be happening in a production ready implementation of the feature.\n\nThe most advances tools Blender has right now that were added in the last releases (Quadriflow, Elastic deform, IK Pose brush...) are designed to work with a Multires system:\n\n- The main purpose of Quadriflow is to produce subdividable geometry for a Multires. When you are done sculpting with Dyntopo/Voxel remesher, where you can’t have the freedom to choose the level of detail you want to work on, you can quickly generate a base mesh, subdivide it and reproject the details back. That way you can tweak again general proportions and low frequency detail.\n\n- These brush tools are designed to tweak proportions/big shapes of models, which is something you most likely want to do after sculpting more detail to have an idea of how your model looks like. Multires solves that problem.\n\n- These brushes can’t be used directly on high poly meshes. These tools needs to have disabled some sculpt mode optimizations in order to work, so they are much slower than regular brushes. In fact, if you try to use the IK pose brush directly on a high poly mesh, most likely you are going to run out of memory. If you don’t, it is going to create terrible artifacts between each segment. In order for these brushes to work as intended on high poly meshes, you need to use Multires to switch to a lower level, use the brush, and then switch to a higher level again to get the intended result.\n\n## Multires implementation reference\n\nSculptGL contains a valid and functional Multires sculpting implementation. You can try it by sculpting different details at different levels and switching between them, as well as creating and deleting subdivision levels:\n\nsculptgl/\n\nThis implementation of Multires provides:\n\n- Detail propagation that works across all levels. You can choose the desired level you want to sculpt on and that deformation is propagated correctly to all subdivision levels. There are not visible artifacts of any kind.\n\n- You can both create and delete levels in both directions (higher levels and lower levels) with subdivide/delete higher and reverse/delete lower. \n\nThese features are the minimum and required a Multires implementation in Blender should provide, as they are pretty much the definition of a Multires sculpting system. Any other Multires related feature or optimization (being part of the modifier stack, baking, shape keys, layers, exporting, reprojection, tool compatibility, Quadriflow...) should always come later. There should not be a design decisions that prioritize any other Multires feature or optimization without having those minimums perfectly working, bug free and functional.\n\n\n## Proposal\n\nGiven the fact that, as described here, Multires was never a functional feature for its intended purpose and:\n\n- The code is extremely hard to read and maintain. I still can’t understand what it is doing in some places or what was the reasoning to make it work that way in the first place. \n- New features are planned that require Multires support (masking tools, vertex colors, visibility operators). Adding Multires support for those features should not be a project bigger than the feature itself. This was the case with most of the 2.81 tools, and probably with the future planned features (Sculpt vertex colors or visibility management, for example)\n- We already have and sculpt tools abstraction layer in place. Making changes on the underlying data structures of sculpt mode should be easier than before\n- The original Multires comes from before 2.5. The basics of how it should works were never correct and a lot of features and modifications were added during the years. We are now trying to build state of the art sculpting features on top of a broken foundation that has more than 10 years. \n\nSome options are:\n\n\t- Replace Multires modifier with a simple implementation based on the SculptGL code (something we know it works) and start again from there. We can keep the current Multires modifier for compatibility, but we need to evaluate if supporting the features and fixes described in this task on the current code (and make it work exactly as the SculptGL Multires) is going to take more time than implementing something based on the SculptGL code from scratch. We also need to look that the new features that are planned and evaluate how hard they are going to be to add to the current Multires code (colors, per face data...) knowing that the SculptGL implementation already supports them.\n\t\n\t- Try to add this functionality to the current Multires modifier. In that case, the priority should always be supporting the functionality described in this task first, and the bring all the rest of the features back. In that case, it would be needed better documentation and code organization to be able to integrate new features into it easily. ", "Grease Pencil problem (especially soft pencil) \nOperating system: Windows-10-10.0.19042-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 970/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 497.09\n\nBroken: version: 3.0.0\n\nSince couple version of Blender I cannot use a soft pencil (mainly) brush in Grease Pencil 2D animation mode. The issue depends on how much You zoom. When You zoom in and use same size it's not visible. When You zoom out and draw with same size dots are visible. \nIt's behavior is very strange now and produces bad results (visible dotted lines or other which are visible on attached file. Some of these issues occur when the brush is small (line type set as dots). This is much more more visible when stroke stabilization is on. The bug was not there about 2 years ago and I was able to use the soft pencil without such problems. When line type is set to line the pencil stroke has strange contour around and when self overlap is off it produces artifacts on edges. \n\n\n\nJust choose a soft pencil and make it small. Check it with stroke stabilization as well. Then change the line type to line with self overlap on and off. \n\n", "Sculpt mode -> Curve Stroke causes Undo to behave differently\nOperating system: Windows 10 x64\nGraphics card: Nvidia 980 GTX\n\nBroken: v 2.82 Alpha, 10-12-2019\n\nIf Brush stroke is set to Curve and some sculpt detail is added (Ctrl + Left Click), (Ctrl + Z) or Undo does not work. Instead of Undo for the sculpt detail, the curve control points are removed.\n\n1. Start with default cube and add Multi-Res modifier, subdivide a couple of times.\n2. Go to sculpt mode and with the SculptDraw brush set the stroke to Curve.\n3. Draw a curve using (Ctrl + Right Click) to add some control points.\n4. Add sculpt details using Ctrl + Left Click. \n5. Press Ctrl + Z. Instead of removing the details the control points of the curve will be removed first. \n6. Press Ctrl + Z repeatedly. When the last control point remains then the sculpt detail will be removed.", "Sculpt Paint Tool: Strength doesn't use unified paint settings\nOperating system: Linux-5.18.6-arch1-1-x86_64-with-glibc2.35 64 Bits\nGraphics card: NVIDIA GeForce GT 1030/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 515.48.07\n\nBroken: version: 3.2.0\nBroken: version: 3.2.1 Release Candidate\nBroken: version: 3.3.0 Alpha\n\nThe strength of the Paint Tool in sculpt mode only accounts for the current brush strength when use_unified_strength is toggled True.\n\n1. Select cube\n2. Go into sculpt mode\n3. Select paint tool\n4. Swap paint tool's color (just so it's faster to get something other than white)\n5. Adjust Tools strength to 1.\n6. Toggle unified strength to true (Unified paint strength is usually .5 automatically)\n7. Paint on cube's verts.\n8. The strength of the brush will still be at 1.0 instead of 0.5\n\n----\nHere is a video showing the bug.\n[Bug render0001-4632.mp4](Bug_render0001-4632.mp4)\n", "Grease Pencil: unable to change erase size while using the Draw tool\nOperating system: Linux-5.19.0-76051900-generic-x86_64-with-glibc2.35 64 Bits\nGraphics card: AMD Radeon RX 5700 XT (navi10, LLVM 13.0.1, DRM 3.47, 5.19.0-76051900-generic) AMD 4.6 (Core Profile) Mesa 22.0.5\n\nBroken: version: 3.5.0 Alpha\nWorked: ???\n\nCan't change the size of the eraser brush while erasing while using the draw tool (`Ctrl + LMB`), I can do it with the scroll wheel on the mouse but not with the shortcuts assign for it `[]`\nAlso applies to annotation tool.\n\n- go to greasepencil draw mode\n- draw something\n- hold control to erase\n- try to change the size of the eraser brush while erasing (Either scroll wheel or brackets)", "PBVH task scheduler performance issues\nThis is a task to investigate a possible issue in the PBVH task scheduler. \n\n**Issues**\n - Some users report bad performance in sculpt mode, while others can work without problems on really high poly meshes with similar hardware.\n\n - On my main computer, brush strokes with TBB disabled feel more responsive regardless the mesh vertex count. When TBB is enabled, strokes usually stop generating new samples in the middle of the stroke or take a fraction of a second to start. I know this is happening to some other users. \n\n - Simple operations that need to loop over all vertices on high poly meshes are extremely slow for no specific reason. (doing x = 1.0f - x on a 25M elements array often takes up to 5 seconds). When the operation is more complex (like updating the normals or updating the nodes on the GPU), this can take minutes to finish. \n\nWe are not getting the reports of people that may be affected by this issue because most of them assume that Blender sculpt mode performance is just bad by default, so I can't tell if this is a problem with a specific configuration of hardware/software or a general problem in how the scheduler works. It would be nice to have more data from this test (and sculpt mode performance in general) on different hardware/software configurations to have a better understanding of how we can improve it. \n\n**How to test:**\n\n - Apply this patch: [P1202](P1202.txt)\n - Add a cube, subdivide N number of times and apply\n - Enter Sculpt Mode\n - Run \"Invert Mask\" (press A + drag left)\n\n\n**My results**\n\nOperating system: Linux-5.4.2-1-MANJARO-x86_64-with-glibc2.29 64 Bits\nCPU: Intel i7 6700K @ 4.00GHz x 8\nRAM: 32GB\nGraphics card: GeForce GTX 1080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 430.64\n\n\nSubdivide default cube N number of times:\n\n| N Subdivisions | Vertices | PBVH Nodes | TBB ON (seconds) | TBB OFF (seconds) | DIFF |\n|--------------|----------|------------|------------------|-------------------|---------|\n| 11 | 25165826 | 7596 | 2.474073 | 0.571511 | 432.90% |\n| 10 | 6291458 | 1920 | 0.307948 | 0.159395 | 193.20% |\n| 9 | 1572866 | 464 | 0.111459 | 0.096845 | 115.09% |\n| 8 | 393216 | 112 | 0.002289 | 0.007297 | 31.37% |\n\n\nSubdivide default cube 6 times, use multires to N level.\n\n| Level | Vertices | PBVH Nodes | TBB ON (seconds) | TBB OFF (seconds) | DIFF |\n|-------|----------|------------|------------------|-------------------|---------|\n| 6 | 26763264 | 3696 | 1.076635 | 0.322183 | 334.17% |\n| 5 | 7102464 | 912 | 0.074586 | 0.113303 | 65.83% |\n| 4 | 1990656 | 232 | 0.012938 | 0.048273 | 26.80% |\n| 3 | 614400 | 56 | 0.004059 | 0.012778 | 31.77% |\n\n\nAs you can see, a simple loop over all vertices in the mesh is way faster on my setup with TBB disabled.\n\nIf I change the ##LEAF_LIMIT## in ##pbvh.c## to 100000 (x10) in order to increase the amount of work each node has and reduce the number of total nodes: \n\n| N Subdivisions | Vertices | PBVH Nodes | TBB ON (seconds) | TBB OFF (seconds) | DIFF |\n|--------------|----------|------------|------------------|-------------------|--------|\n| 11 | 25165826 | 760 | 0.106868 | 0.210985 | 50.65% |\n| 10 | 6291458 | 184 | 0.029088 | 0.105376 | 27.60% |\n| 9 | 1572866 | 56 | 0.007168 | 0.029653 | 24.17% |\n| 8 | 393216 | 8 | 0.002261 | 0.006321 | 35.77% |\n\nIf we compare the two results with TBB on:\n\n| Vertices | 1000 LEAF_LIMIT (default) | 10000 LEAF_LIMIT (new) | Perf. gain |\n|----------|-----------------|------------------|------------|\n| 25165826 | 2.474073 | 0.106868 | 2315.07% |\n| 6291458 | 0.307948 | 0.029088 | 1058.68% |\n| 1572866 | 0.111459 | 0.007168 | 1554.95% |\n| 393216 | 0.002289 | 0.002261 | 101.24% |\n\nIt seems like if we avoid starting TBB with a large number of nodes the performance improvement is huge. Increasing the leaf limit makes these kinds of operations up to 23 times faster, but that would be like disabling multithreading for small brush strokes.\n\n- Is there a way to fix this in the scheduler?\n- Can we do something like sending small chunks of nodes to TBB instead of starting the loop with 7000 nodes at the same time?\n- Can we find a better optimal leaf limit to get a better workload per core suitable for modern processors by comparing the results on different hardware? Dyntopo leaf limit is in 100 elements, which I think it does not make much sense. " ]

[ "Start of a brush stroke always results with maximum pen pressure using Wintab and Windows Ink, latest 2.93 alpha and Huion WH1409 v2.\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: Radeon RX 580 Series ATI Technologies Inc. 4.5.14758 Core Profile Context 21.2.2 27.20.14535.1000\n\nBroken: version: 2.93.0 Alpha\nWorked: 2.91\n\nStarting a brush stroke always results in maximum pen pressure, even tho i use low pressure. Bug in Wintab and WIndows Ink but Windows Ink also gives no brushstroke except the initial maximum pen pressure dot. See video.[2021-02-17 19-00-25.mp4](2021-02-17_19-00-25.mp4)\n\nAll drivers are up to date.\n\n**EDIT: EDIT: I restarted my PC and it seems Windows Ink is working fine now. Proper pen pressure and no pressure spikes but instead Wintab gives me full pen pressure all the time. Weird...**\n\nUse a Huion tablet and sculpt something\n\n" ]

[ "Multi-Button Drag: not working properly for Object Dimensions\nWindows 7\n\nBroken: 2014-02-11\n\nMulti-Button Drag - D270 - doesn't work properly in N panel for Dimensions\n\nIf you drag over all 3, only X and Z change, Y remains unchanged. As mouse button is released, the first selected property will reset, only the last hovered is kept (and scale changed)", "Threshold of the 'Select Similar' operator (Shift G) is very limited at large scales.\nHi,\nsince I'm working on a project in Unreal Engine it needs to be actually done in centimeters (meaning in mesh-terms it's 100 bigger than meter mesh).\n\nHere is the blend file: [ShiftG_SelectSimilarArea_CantPutInMoreThan1.000.blend](ShiftG_SelectSimilarArea_CantPutInMoreThan1.000.blend)\nTry to select a face and Shift+G -> Select Similar (not sure if the shortcut stayed that way)\n\n\nI'm running into an issue here and there where some functions expect the mesh to be in meters and work worse in centimeters (or not at all)\n\nFor example here I was using Shift+G and Area to get the faces with the same area. Because the areas in the centimeters model are so big number-wise it doesn't find any match. So either I think the amount needs to be unlimited (I would put in 100 and that would probably have the same results as when the mesh is 100x smaller) or this needs to be addressed from the unit scale side of things.\n\nThis is in Meters (scaled the mesh 0.01), we get a good window selection because they are mostly the same:\n\nAnd in Centimeters nothing more than the one face is selected.\n\n\nMaybe the scale in Units should be used in these operators to make the mesh smaller?\n\n\nThanks for looking into this.", "Unable to select multiple Grease Pencil layers with Python\nOperating system:\nGraphics card:\n\nBroken: 3.4\nWorked: \n\nSetting the Grease Pencil Layer property \"select\" with Python doesn't seem to work. Tried various update tags with no luck. Similar layer properties, such as the booleans \"lock\" and \"hide\", work as expected with the same code as below.\n\n[select_bug.blend](select_bug.blend)\nOpen Blender from a console/terminal and keep that window in view, open the attached .blend, click the play button in the Text Editor header to run the example script (which sets all layers' select property to True), then hover the mouse cursor over the Python console and press Enter to run the command: print([gpl for gpl in bpy.context.object.data.layers if gpl.select])", "Applying Data Transfer Modifier - only the active Shape Key is taken into account\nOperating system: Win 10 x64\nGraphics card: 3080 TI\n\nBroken: 3.3.2, 86abbf717691, master, 2022-11-28\nWorked: ??\n\nWhen applying data transfer modifier to an object (to transfer vertex groups), it only takes into account the selected shape key, not all others too.\n\nExample:\n\nDownload test Blend:\n[datatransfer_shapekeys.blend](datatransfer_shapekeys.blend)\n\nThis is the cube's vertex group:\n\n\nPlane is moved down by shape key \"Key 1\". All looks good in the viewport, this is how the vertex group should look when you apply it:\n\n\nBut if the 'Basis' shape key is selected instead of 'Key 1' when you apply the modifier, it ignores Key 1.\n\n\nSee how the position of the vertex weight is off as compared to the first picture, as if Key 1 were disabled.\n\n\nBut if you were to have Key 1 selected when you apply the modifier, it works as expected:\n\n\nThis is troublesome with objects that have many shape keys, all with different influences on the affected area. When you apply the modifier, it should take into account every active shape key.\n\nThis gets even less predictable if you have multiple shapekeys besides the `Basis` shapekey.\nHere I have two (both with 0.5 weight).\nYou get different results selecting `Key 1` or `Key 2` and none look like what you had in the viewport:\n[datatransfer_shapekeys_multiple.blend](datatransfer_shapekeys_multiple.blend)", "UV: Mesh won't Unwrap at all\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3060 Laptop GPU/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 536.23\n\nBroken: version: 3.6.0\n\nWhat is happening is that when i unwrap my mesh nothing happen, even if i isolated one face and tried to unwrap it wont work, also even faces you extruded from the faulty face won't unwrap, but adding a new plane from the add menu inside the same mesh unwraps normally, even if i connected the plane to the faulty face.\n\n1. Create a plane.\n2. Give one edge a Mean Bevel Weight from the side bar.\n3. Give the plane a Bevel Modifier and the Limit Method set to Weight (visually nothing will change since its just a plane but the error will still reproduce).\n4. Apply the Modifier and Unwrap the plane. (normally the plane comes unwrapped by default, try changing the UVs out of bounds or scale them down just so you can notice the change ).\n\nThanks you blender developers.", "Subdivision modifier does not interpolate zero weights.\nOperating system: Linux\nGraphics card: Intel UHD\n\nBroken: 2.90.1\nWorked: ???\n\n\nThe subdivision modifier usually interpolates both value and membership across subdivided vertices.\n\nThis allows it to be used in non-destructive, modifier-based workflows that operate on or using vertex groups.\n\nIt does not do this, however, when vertex weights are equal to zero. The original vertices retain their (zero-weighted) vertex group membership, while new vertices in between them have no vertex group membership.\n\nAn example of why this is problematic is screenshotted below, with the file attached:\n\n\n\nEach Subdivision'd sphere has a bumpy texture applied using a Displace modifier after a textured VertexWeightMix modifier.\n\nThe rightmost sphere has its vertices weighted to 0.01, and interpolates and behaves correctly.\n\nThe leftmost sphere has its vertices weighted to 0.00. It intuitively (and, numerically/extrapolatively) should be indistinguishable to the 0.01 sphere. Because its weights aren't correctly interpolated, it instead ends up with a bunch of spikes at each of its original vertices.\n\nThe middle sphere has vertices weighted in a gradient from 0.00 to 0.50. It serves as an example of a feasible real-world application which would be broken by this kind of behaviour. Perhaps the original vertex weights act as a sort of base heightmap for a terrain model, and the changes made by the Subdivision and textured VertexWeightMix modifiers add an extra layer of detail. In this case, the clear and useful behaviour would be for behaviour to be continuous in the transition from 0.01 to 0.00. Instead, there is a sharp cutoff at 0.00, and inconsistent and discontinuous behaviour after that.\n[ZeroWeightSubdiv.blend](ZeroWeightSubdiv.blend)\n\nApply a subdivision modifier to a mesh with zero vertex weights, then check the vertex group membership of its vertices afterwards.\n\nOr download and open the attached `.blend` file.", "X-ray toggle button in weight-paint mode is ignored (honors edit-mode setting instead)\nOperating system: Linux\nGraphics card: Intel\n\nBroken: 2.91.0-a509e79a4c77-linux64\nWorked: 2.80 - Xray was not changed when switching modes there\n\nWhen in weight-paint mode, with select vertices enabled, the \"Toggle X-ray\" button in the upper right corner is ignored, does not allow to select vertices through the object (on both sides). But instead, if you switch to the edit mode, enable the x-ray mode, and then switch back to the weight-paint mode, then the vertex selection through object is turned on correctly. So oddly enough, the weight-paint mode x-ray is controlled from edit mode, ignoring its own x-ray toggle button.\n\n**Steps to reproduce**\n[xray-weight-paint-bug.blend](xray-weight-paint-bug.blend)\n\n- Open file\n- Toggle X-ray mode", "Sculpt: Evaluate multiple paint steps at the same time.\nCurrently a paint step is evaluated one by one. Redoing multiple checks and calculations. In case those checks and calculations are similar we could improve performance by evaluating multiple paint steps in the painting kernel.\n\nWhen sculpting this isn't possible as geometry is changing, but when only effecting colors or attributes this could improve performance.\n\nChecks/calculations that would return the same result:\n* Clipping tests: In order to do this we should split clipping tests from brush tests.\n* Most texture evaluation\n\nWould especially improve texture painting performance as most effort at this moment was done to reduce the calls to these tests.\n\nA possible approach would be to collect the steps in a list, still evaluate them one by one. But brushes that can benefit would store the needed data in a list containing dynamic data per step.\nWhen the paint frame finishes the brush will be triggered to do the actual evaluation.\n", "Select Loop Inner-Region issue\nWin8 64bit | GTX680\n\nBroken: 2.77 509270e\n\nWhen working with face loop, this tool doesn't always work as expected.\n\n\n\n\nNot sure if this can be a bug. Please check.", "Vertex weight not carried through bevel modifier\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 4080 4.5.0 NVIDIA 528.24\n\nBroken: version: 3.4.1\n\nVertex weight not carried through bevel modifier in some cases\n\n\n[105012.blend](attachment)\n- open blend\n- apply visual geometry on the `good` object\n\n- apply visual geometry on the `bad` object\n\n\n\n\nOK, some more testing (which solves the issue at least on the example I posted [also if there is a second bevel modifier with custom profile on there] -- as well as to some degree for the original file as well [which is posted below]):\n\n- open the file\n- go into editmode on the `bad` object\n- select all **edges** (edge select mode)\n- `Mesh` > `Sort Elements` > e.g. `Cursor Distance`\n- voila!\n\n-----\n\n\n\n**Original report**\n\nWhen wishing to create a bevel with the modifier on a specific vertex group aftern another Bevel modifier, the vertex group does not carry over.\n\n\n* Create a geometry\n* Mark some edges to bevel with bevel weight set to 1\n* Add an edge loop in to a vertex group\n* Add a bevel modifier and use Weight\n* Add 2nd bevel modifier and use vertex group. The area where the first Bevel modifier has been executed does not become incorporated into the vertex group\n\n\n\n\n\n\n\n\n", "Solidify (Complex, Constraints) - very rare glitch\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 528.02\n\nBroken: version: 3.5.1\nWorked: 3.1.2\n\nbehavior changed in c2e8e68b6517b91eb14270707d7182c1727861c4\n\nSolidify modifier creates a weird glitch at a specific vertex.\n\nI have a model with about 40000 verticies that I wanted to solidify. Everything works fine, except the surroundings of a single vertex. The topology does not seem any particular there. The model is made out of triangles, but the glitch also appears if I convert it to quads. I copied the vertex and the surrounding faces to a sperarate blender file and recorded how the glitch appears depending on moving one of the verticies:\n\n(the glitch looks exactly the same when the vertex is part of the larger object)\n\nSolidify Modifier Settings:\nMode: Complex\nThickness Mode: Constraints\n(other settings does not affect the glitch)\n\n", "Dynamic vertex groups (Dynamic Paint, Weight Proximity modifier, Geometry Nodes, ...) do not update density of particles in viewport display\nOperating system: windows 10\nGraphics card: Quadro M620\n\nBroken: 2.90\nWorked: Never\n\nI want to use my dynamic vertex group to generate particles on a plane.\nIt just works when I change parameters during the playback but it's not realtime when I hit play.\n\n- Open attached file\n- Play animation\n- Change any property of the particle during the playback\n[dynamic_paint_test.blend](dynamic_paint_test.blend)", "Intersect Boolean does not work properly when mesh has a degenerate face\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: Radeon (TM) RX 480 Graphics ATI Technologies Inc. 4.5.14760 Core Profile Context 20.45.37.01 27.20.14537.1001\n\nBroken: version: 2.91 - 3.0.0\nWorked: n/a\n\nThe `Intersect Boolean` Exact mode does not work correctly when the mesh has at least one degenerate face.\nIn this case, it was tested on a mesh with a triangle whose vertices are collinear\n\n- Open attached file - (The edited mesh has a degenerate triangle)\n- Go to `Face` -> `Intersect (Boolean)` - (The operation will not work because of that triangle)\n\n- Undo the operation\n- Move the triangle vertex a little so that the face is no longer degenerate.\n- Reselect that loose geometry ({key L})\n- Redo Boolean operation (now it works)\n[intersect_w_degenerated_tri.blend](intersect_w_degenerated_tri.blend)\n\n", "UV space aware triangulation\nPlease see #102683 for details, since the issue is described and discussed there. The basic suggestion is:\n\n> Add an option to the \"Triangulate Faces\" operator. It currently has Quad/N-gon methods for Beauty, Fixed, Longest edge etc. It seems reasonable to add options for \"Prevent UV overlaps\" and specify the source UV map.\n\n", "Setting/editing keyframes on multiple characters in pose mode doesn't work for Whole Character and Whole Character(selected Bones Only) options.\nOperating system: Windows 10 Home Edition\nGraphics card: GeForce RTX 2080\n\nBroken: 3.1.0 Master Build\nWorked: never\n\nSetting/editing keyframes on multiple characters in pose mode at the same time only affects the Active Character's Keys on certain insert key options.\nIdeally the insert key options work consistently for all selected characters bones.\n\n - Add two character skeletons to a new scene (Shift+a > Armature > single bone)\n - Select the first bone then Shift+Select the second one.\n - Switch to pose mode\n - hit \"a\" to select all. You should have both bones highlighted\n # hit \"i\" to insert a keyframe.\n - # if you choose Whole Character or Whole character \"Selected bones only\" it will only set a key on the bones for the Active Character.\n - # if you choose any other option it sets keys on both characters.\n\n\n\n\n\n\n**Additional comments:**\nIdeally Whole character would key all characters currently in pose mode.\nand Whole Character (Select bones only) would Key any currently selected bones regardless of which character they belong to.\n\nThe current functionality is confusing, if it is intended, the label should read \"Active Character\" and \"Active Character(selected Bones only)\"\n\nHowever when a user chooses to have multiple characters in pose mode it is because they are trying to pose and key them at the same time without having to switch back and forth with every adjustment. So having the ability to key multiple whole characters, or a random selection of bones belonging to multiple characters would be useful.\n\nThat said keying rotation/translation/scale&custom properties gets me what I need so this is not a blocker but it is something that could be improved for clarity as it will frustrate many animators unaware of this inconsistent behavior for the last two options.\nI was relying on \"Whole character\" and \"whole character (Selected bones only)\" to essentially key everything until I had to do multiple characters and I got confused for a while why my animations were looking janky until a dug into it." ]

[ "Vertex / Weight Paint. Blender does not show if vertex is selected.\nOperating system: Windows-7-6.1.7601-SP1 64 Bits\nGraphics card: GeForce GTX 1050 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 431.60\n\nBroken: version: 2.81 (sub 3)\nWorked: (optional)\n\n\nSelect All works fine, but for Select and Lasso Select there is no visual feedback. \n\n\n1. Enter Vertex Paint mode\n2. Turn On vertex selection\n3. Select anything\n\n[vertexSelect.blend](vertexSelect.blend)" ]

[ "Transform orientation undo is inconsistent.\n\nBroken: 3.0\nWorked: ?\n\nTransform orientation undo is inconsistent. It works in object mode, but doesn't work properly in edit mode. In edit mode it sometimes won't undo, or undoes together with another different step(like changing mode, or moving vertex).\n\n- Open default startup;\n- Change the Transform Orientation from Global to something else\n- Undo/redo - it works;\n\n---\n- Reopen default file\n- Go to edit mode\n- move vertex\n- Change the Transform Orientation\n- undo multiple times - Will never undo the Transform Orientation. (now deselect cube and perform undo operation: will revert selected transform orientation)\n\n", "Drag-to-toggle visible/selectable/renderable in the Group view.\nWindows 64\n\nBroken: 2.70a\n\nThe drag-to-toggle visible/selectable/renderable in the Outliner does not work in the Groups view.\n", "Edges are selected in face mode when deselecting vertices from vertex group Panel\nBlender Version\nBroken: Current main.\nWorked: N/A.\n\nShort description of error\nDeselect Button in Vertex Groups Panel will deselect vertices that are not assigned to any groups. Edges become selected while in Face Mode as well. Also the Select Button will Select Faces that are not assigned too but those Edges that are active in face mode must be Selected for the Bug to work. Pay attention to the Edge around the triangle Fan at the top of the uv Sphere While in face mode.\n\nBlend File Incuded for simplicity\nJust Press the Deselect and notice the Faces that Get removed. and the Edges are actively Selected.\n\n", "Modifiers disabled in viewport are not rendered if the object is in edit mode\nVersion 2.83\n\n\n - Open: [medieval_house_lights_camera.blend](medieval_house_lights_camera.blend)\n - Render with the object in edit mode and then in object mode.\nScene in object mode\n\nRender:\n\n\nScene in edit mode:\n\nRender:\n", "Mode locked and crash if objects become disabled in viewport while not in object mode\nOperating system: Linux-5.15.0-2-amd64-x86_64-with-glibc2.33 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 470.86\n\nBroken: version: 3.1.0 Alpha\nWorked: didn't crash but showed an error message in 2.91.2\n\nObjects that are in a mode which is not Edit Mode nor Object Mode, will lock the mode setting (so the mode options are greyed out and unusable) if they get disabled in the viewport. Hiding will not cause the this bug, it needs to be disabled in viewport using that display icon in the outliner, not the eye.\n\nThe solution for this is probably to make the disable in viewport action switch the mode to object mode if the active objects gets removed (much like delete from outliner). Another solution would be to still allow the mode switch.\n\nMuch worse though! Blender crashes sometimes when it gets into this state. I can reliably crash blender as described below.\n\nGo into weightpaint mode of the default cube and disable it in the viewport using the outliner (you need to enable the option first). Now you are stuck and have to enable it in the viewport first to be able to edit anything.\n\nNow restart blender etc. and try this:\n1. Go into Vertex Paint Mode of the default cube\n2. Hit Tab to go into Edit Mode\n3. Disable the cube in the viewport using the outliner as above\n4. Hit Tab to switch out of Edit Mode -> Crash", "Eyedropper in object selection box doesn't work\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce 920M/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 425.31\n\nBroken: version: 2.82 (sub 7)\nWorked: (optional)\n\nBroken: version: 2.81 (sub 16)\nWorked: (optional)\n\nBroken: version: 2.83 (sub 17)\n\nBroken: version: 2.90 (sub 2)\n\n\nI click on a specific eyedropper and then it doesn't work. Other eyedroppers that I know seem to work fine.\n\n3D Viewport > Sculpt Mode > Top Header Panel > Options > Orientation > Eyedropper in object box selection > Click on it\n\n[Based on the default startup]\n\n[2020-05-17 13-21-02.mp4](2020-05-17_13-21-02.mp4)\n\n", "Node editor: Allow editing geometry node tree without context\nLike the \"Material/World\" toggle in the shader editor, add an option to allow choosing any node group regardless of the context.\n\n", "Transform & Drag Action \nWhen switching the transform tools in the toolbar, there are times when you would still like to be able to box select while gizmos are active.\n\nWe can solve this by adding a Drag Action tool option to all the transform tools:\n\n\n\nWhen set to Box Select, you can still drag in the viewport to select, while using the gizmo to move, rotate or scale.\n\nDetails:\n - This option should ideally be shared between all tools with gizmos, I think. If you set one tool to work this way, you likely want them all to\n - This works best if we add back the Transform tool\n\n", "Lasso Select Broken if Select key is changed from Default Left to Right\nOperating system: Windows 10 (Home)\nGraphics card: Nvidia 1070 \n\n**Blender Version** \nBroken: 2.90.0 (Steam)\nWorked: Not sure, Has been like this since before 2.79b\n\nWhen the default select mode is changed from Left to Right Mouse button Lasso Select no longer works with Tweak + RMB Any. \n\nSplash Screen Change on initial set up or Select button or Edit - Preferences - KeyMap - Select With Mouse Button\n\nAffected: any View Mode that has the Lasso option. (Edit mode, Object Mode, Grease Pencil, etc)\n\n\nSee attached MP4.\n\nBased on the default startup or an attached .blend file (as simple as possible).\n\nStart Blender and use Start up Cube. [BlendBug_2.mp4](BlendBug_2.mp4)\n\nTemporary fix if the Select key is changed from Left to Right Mouse Button is to walk through the KeyBindings and set anything that uses Lasso to Tweak Right and anything that was using Tweak Left to Tweak Right where Lasso is was changed. ", "UV editor and 3d viewport selection sync per default\nIt would be very beneficial to have corresponding mesh component selection between the **UV editor** and the **3D viewport**. This would enable users to quickly identify which mesh components that they are working on. This design should not simply be implemented by setting \"sync selection\" as per default, since there are some drawbacks in the UV Editor with that mode such as\n- Select linked does not work with vertices and edges\n- Moving faces \"rips\" them from uv islands\n- Uv island selection mode is not available\n\nThis issue is addressed here as well : [BDcbbc/ ]()\n\nTask updated 2022-09-10 due to comments from Campbell\n---\nCampbelll expressed concerns about overhead 3D-viewport selection having to be aware of UV connectivity+selection. If it helps we could make sure selection sync to is an option so that selecting elements of high res meshes does not become slow.\n\n---\nAdded notes based on comments from Campbell (see comment May 4 2022,)\n\n\n- Show all visible geometry in the UV window\n- Selection actions in the UV editor would flush changes to the 3D-view. (where any selected UV would cause it's vertex to be selected)\n- De-selection actions in the 3D-view would flush changes to the UV-editor (where any fully de-selected vertex would have all it's UV's de-selected)\n- Selection actions in the 3D-view would be unable to pick a face corner (select vertex-group for e.g.) and select all the vertices UV's\n- When switching uv layers the uv selection is kept\n- Picking individual vertices/faces/edges can be supported. Individual vertices/faces/edges should be selected in the 3d view just as in the current state of blender. In the the UV editor, individual face corners can be selected. If the selected face corner(s) shares the same position as another face corner - all of the face corners that share location with the uv face corner selection will be selected in the uv editor. This selection will then be flushed to the 3d view, where any vertex that is connected to the selected face corner(s) will be selected\n- box/lasso-selection for e.g. that includes whole faces might reasonably be pick the UV's that are part of the newly selected faces. If a lasso tool selects multiple faces in the uv editor, any face corner that shares the same location with the selected face(s) face corner(s) will be selected. The selection will then be flushed to the 3d view as described above.\n\n\n---\nAdded notes after meeting with Julien 2023-07-14 \n\nWhen a face is selected in the uv editor and component selection type is changed to face corners, only the face corners should be selected - not the vertices. Changing from face selection to vertex selection in the 3d view would select all of the face corners connected to the same vertices.\n\nThe component selection types in the uv editor should be \n- face corner (\"vertex\" as it is currently called in the uv editor)\n- edge\n- face\n- island\n\n---\n\n\n||\n| -- |\n|Make sure UV is visible in edit mode even if vertices/edges/faces are not selected|\n\n||\n| -- |\n| Faces in the 3d view is selected based on the current selection in the UV editor and vice versa. This makes it easy for the user to understand which parts of the mesh she/he is working on|\n\n||\n| -- |\n|If faces needs to be hidden in the UV editor (in the case of overlapping UV’s etc), the user can simply hide the UV’s and the faces|\n---\n\n||\n| -- |\n|Desired result when selecting one vertex in the **3D VIEWPORT** - Note that the vertex is placed on a UV seam|\n\n||\n| -- |\n|Desired result when selecting one vertex in the **UV EDITOR**|", "if multiple splines on the same place... no selection is visable\nOperating system: Linuxmint 19.0\nGraphics card:Geforce GTX1050\n\nBlender Version\n81ea815dcb6, blender2.8, 2018-12-12\n\nI created a spline for modelling.\nThen I converted the spline into a mesh.\n\nI want to delete some vertices, but I could not select any.\nBy clicking an moving I saw there is 1 vertice selected.\n\nBut any other selection is blind.\n\nBy the way: I don't even know how these >10 splines were created...\nsuddenly they where there...\n\n\n[Bug-Multiple Objects.blend](Bug-Multiple_Objects.blend)\n", "Box select fails to select bones at certain zoom levels\nOperating system: Linux-5.19.0-43-generic-x86_64-with-glibc2.36 64 Bits, X11 UI\nGraphics card: NVIDIA GeForce GTX 1080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 530.41.03\n\nBroken: version: 4.0.0 Alpha\nAlso 3.5 and 3.6\n\nBox select fails when zoomed out\n\n\n- Open [Select.blend](attachment)\n- Box select some of the bones in the provided file.\n- It doesn't work for me unless you zoom in first\n- If you display the bone as \"Wire\" or \"BBone\" then everything is selecting correctly, otherwise not.\n\nIt fails in both pose and edit modes, perspective or orthographic.\n\n", "overlay.show_paint_wire doesn't work in vertex paint modes depending on edit mode selection\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 460.89\n\nBroken: version: 3.0.0 Alpha\nWorked: 2.80?\n\nThis overlay option seems to take into account edit-mode selection\n\n\nHere's what it's supposed to do:\n[viIeK7Tkv0.mp4](viIeK7Tkv0.mp4)\n\nBut on 2.93 and 3.0 it appears to do nothing in Weight/Vertex Paint mode if the geometry is selected in edit mode. This seems like a bug, unless it's some odd design choice?\n\n- Create a cube and select all its verts in edit mode.\n- Enter weight paint mode\n- Toggle the Show Wire setting under Viewport Overlays->Weight Paint (NOT \"Wireframe\" under Viewport Overlays->Geometry)\n- Nothing happens\n- Switch to edit mode\n- Deselect all with Alt+A\n- Go back to weight paint mode\n- Wireframe display now works as expected.\n\n\nNote: This now changed again in 44d2479dc3 (now it does not draw **at all** -- reported separately in #89880 (overlay.show_paint_wire doesn't work anymore))", "Handle Type selector in the Graph Editor's sidebar sets different types than selected\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 517.40\n\nBroken: version: 3.3.0\nWorked: Don't know if it ever worked before or when.\n\nWhen choosing the type of handle for a keyframe in the Graph Editor's sidebar, the options don't work as expected. Selecting \"Vector\" sets the handle to \"Free\", and selecting \"Automatic\" or \"Auto Clamped\" sets it to \"Aligned\".\n\nNot sure if this is expected behavior, but choosing an option to get a different one be set felt weird. Most of all, I couldn't set it as Vector from there, so I had to press V in the Graph Editor and do it from there. Interestingly, then it appears as \"Vector\" in the sidebar as well.\n\nI'm attaching a screenshot of the menu I refer to, and a .blend prepared to reproduce the issue (any file with an animation would do).\n\n\n\n1. Create an animation (already present in the attached .blend file)\n2. Open a Graph Editor\n3. Select a keyframe\n4. Open the Graph Editor's sidebar\n5. Try to change the handle type options in the F-Curve tab of the sidebar\n\n[Handles Issue.blend](Handles_Issue.blend)\n\n", "Greasepencil 3.0: Unable to switch to drawmode from sculptmode\nOperating system: Linux-5.15.0-73-generic-x86_64-with-glibc2.35 64 Bits, X11 UI\nGraphics card: GeForce GT 630/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 390.157\n\nBroken: version: 4.0.0 Alpha\n\nEnable GP 3.0\nOpen 2D session\nSwitch from the initial draw mode to sculpt mode via header button.\nTry to switch back to draw mode.\nAn error message appears:\n\n" ]

[ "Box selection or any selection method doesn´t work when I drag the mouse with move tool active (when using right click select)\nOperating system: Windows 10 Home 64 bits (10.0, compilación 19042)\nGraphics card: NVIDIA GeForce GTX 1060\n\nBroken: 3.0.0, f1cca3055776, master, 2021-12-04, as found on the splash screen)\nWorked: 2.93.6\n\n\nwhen i drag the mouse to select things in viewport with any transform tool selected box selection or any selection method doesn´t work. Instead what happens is that the action of the selected tool is executed\n\nReproduction\n1. Open blender\n2. Enable Right click select configuration\n3. Select move tool\n4. Try to do a box selection by dragging mouse" ]

[ "Multiple mix nodes lead to incorrect results\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 512.15\n\nBroken: version: 3.4.0\n\nI'm not good at English and don't know how to explain it, it's a stripped down part of my project. When two orange nodes have the same input value, they get a value of 0.5. But when the value is 5, sometimes 8 or 9, the correct result cannot be obtained, it seems that it is not refreshed.\n\nI provided the video and the .blend file, hope that helps.\n\n[bug.blend](bug.blend)\n[20221213_16.48_793.mp4](20221213_16.48_793.mp4)\n\n\n", "\"Material preview\" shows always eevee mat branch / never cycles\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 980/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 432.00\n\nBroken: version: 2.82 (sub 6)\nWorked: never, I think\n\n\"Material Preview\" shows always the material output assigned to eevee, but I want the cycles branch. Changing renderengine or the active material output node doesn't change anything.\n\nOpen the blendfile and try to preview the material with \"cycles\" image - without going to rendered viewport. The material preview seems to always use the node tree going to the Eevee material output node, even when the render enging is set to Cycles.\n[specific_material_output.blend](specific_material_output.blend)\n\n\n", "Add modifier error message for attribute name clashes\nIf user has a UV and a vertex color groups with the same name things behave in non predictable ways.\n\nThe nodes modifier should check for this and add an error message when this happens.", "Copy and Pasted framed nodes jitter around when grab and drag\nOperating system: Windows-10-10.0.18363-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 471.68\n\nBroken: version: 3.3.0\nWorked: N/A\n\n\nWhen copy pasting framed nodes with Ctrl + C then Ctrl + V, then G to move by mouse, many nodes appeared to be jittering for unknown reasons.\n\nThis behaviour doesn't exist when G to move existing framed nodes or when duplicating via Ctrl + D.\n\n(the video compared the behaviour \"ctrl + c / ctrl + v / g to drag\" with \"ctrl + d drag\")\n\n[blender_1n2MKEk5wo.mp4](blender_1n2MKEk5wo.mp4)\n\nThis is important because I have a reasonable doubt that copy pasting framed nodes from 1 node graph to another will sometimes erroneously create invalid socket link due to this jitter. This larger problem is very difficult to reproduce, BUT here is a case where I believe such behaviour result in a \"2 link in 1 socket\" invalid setup (when grab them over other frame nodes).\n\n\n\n- Download the blend,\n- Use Select Box to select the Frame and all nodes inside,\n- use {key Ctrl C} then {key Ctrl V} to duplicate framed nodes in GN view,\n- {key G} to move nodes to see the then swimming.\nTry varying the Zoom if the problem is not noticed.\n\nNOTE: The problem is more noticeable by setting the `Noodle Curving` in Theme to 0\n\n[frame-group-jitter.blend](frame-group-jitter.blend)\n", "Compositor/Stereoscopy with view switch and file output generate extra files\nBroken: 2.79, 2.81, ...\nWorked: ?\n\nwhen using Stereoscopy to render a scene from multiple cameras simultaneously, and using the compositor 'Switch View' node to save each camera output separately - unwanted files are saved with each output file.\n\n**Simpler version of the report**\n\nThe combination of the switch view node, mix RGB node and rgb node is creating the problem, disabling even just one of those three nodes produces normal results. \nThe bug is reproduced only when all mix nodes are connected to the same RGB node, but connecting each node the a duplicate of the same rgb node seems to resolve the problem (see screenshots attached)\n\nSample file:\n\n[Compositor bug.blend](Compositor_bug.blend)\n \n\n\n\n\nFor the records the issue also happens in 2.79.\n\n[stereo279.blend](stereo279.blend)\n\n**Original description of error**\n\nIn the attached .blend are 9 cameras set up with sterescopy. The compositor separates and saves each camera output to a different directory.\nFor some reason, with every file output from a certain camera there's another file output from a different camera (lets call it parasite camera) , only this one is black with no data (see attached screenshot). \na few charasteristics:\n- The parasite camera output is added to every output of the other cameras, but not to its own\n- Rendering again changes the parasite camera to a different one (in the screenshot it was 'Cam006', in other renders it was a different one)\n- It's not clear how the parasite camera is \"chosen\", it isn't selected or active, and it's not the scene's camera\n- There's no file output in the compositor that suggests this output\n\n \n[TemplateScenePopulated_001.blend](TemplateScenePopulated_001.blend)", "Crash when multiple materials are assigned to a mesh\nOperating system: Linux-5.17.15-76051715-generic-x86_64-with-glibc2.35 64 Bits\nGraphics card: NVIDIA GeForce RTX 3070 Laptop GPU/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 470.103.01\n\nBroken: version: 3.2.1 Release Candidate\nWorked: unknown (I haven't opened the .blend file in versions older than 3.2. It is broken in 3.2)\n\nWhen two materials are assigned to the mesh, random faces disappear and reappear when switching between Object and Edit modes. After switching back and forth several times, Blender crashes. In some instances it took significantly more switches from Edit mode to Object mode and back before Blender crashed. In some of these instances, Blender appeared to create new faces with one or more vertices at the object's origin.\n\nWhen transforming any vertex, edge, or face, Blender quickly generates new faces with at least one vertex at the object origin and crashed shortly thereafter.\n\nI posted a 30-second video of the issue (including two crashes) here: strange_face_artifacts_causing_crashes/\n\nThankfully, I have not reproduced this issue in any other .blend file.\n\n1) Open the attached file, \"Crash - random faces.blend\" and you will see the mesh has a number of holes in it.\n2) Press tab and the missing faces will reappear but other faces will disappear. Pressing tab additional times eventually results in Blender crashing.\n3) Select any vertex, edge, or face and transform the selected element using G, S, or R and new faces will appear. Blender will quickly crash.\n4) Reducing the number of material slots to one appears to fix the problem (all faces are visible as normal) but it reappears as soon as a material is assigned to the second slot.\n\nThank you very much for your time.\n\n[Crash - random face orientation.blend](Crash_-_random_face_orientation.blend)\n", "Quick favorites: Add previously removed behavior causes duplication\nOperating system: windows 10\nGraphics card: RTX 3090\n\nBroken: Current main.\nWorked: N/A.\n\nIf you remove one item from Quick favorites, the next time you will be able to add it again.\nThis cause duplication:\n\n\n1. Add some items to Quick Favorites.\n2. Remove some items.\n3. Sometimes you will able to add previously existing items back to the list again. (see attached video examples)", "Transform orientation undo is inconsistent.\n\nBroken: 3.0\nWorked: ?\n\nTransform orientation undo is inconsistent. It works in object mode, but doesn't work properly in edit mode. In edit mode it sometimes won't undo, or undoes together with another different step(like changing mode, or moving vertex).\n\n- Open default startup;\n- Change the Transform Orientation from Global to something else\n- Undo/redo - it works;\n\n---\n- Reopen default file\n- Go to edit mode\n- move vertex\n- Change the Transform Orientation\n- undo multiple times - Will never undo the Transform Orientation. (now deselect cube and perform undo operation: will revert selected transform orientation)\n\n", "Hidden sockets in Compositor don't stay hidden when we duplicate any node or render starts\nOperating system: Windows-10-10.0.18362 64 Bits\nGraphics card: GeForce RTX 2080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 436.15\n\nBroken: version: 2.81 (sub 11)\nWorked: (optional)\n\nThis seems more like a cosmetic problem and is more a really small bug but it is nevertheless \nvery annoying - especially with big compositing setups.\n\n - > Hidden Sockets - hidden with CTRL-H are showing up and are not hidden anymore\n[6vWCy01EPQ.mp4](6vWCy01EPQ.mp4)\n\n\n\n - - Create Renderpasses with more then 3 Outputs\n - - open up CompositorEditor \n - - Press CTRL - H twice to hide all unconnected sockets\n - - hit Render or duplicate any node\n\n", "Particle Texture influence broken when using two particle system on the same object\nOperating system: Ubuntu 18.04\nGraphics card: GeForce 970GTX\n\nBroken: 2.90.1\nWorked: Never (2.8+)\n\n\nWhen using two particles system on the same object with an influencing texture, one particle system has a wrong influence (e.g. Particle System shrink to quickly)\n\nBlend File: [particle_texture_bug.blend](particle_texture_bug.blend)\n\n0. Set the animation start and end to 100 - 200\n1. Create an Emitter and a Particle Object\n2. Add a Particle System to the Emitter and use the Particle Object for rendering\n3. Set the Start and End to 100 - 200\n4. Give the Particle System a Blend Texture that influences the size with \"Strand/Particle\" as Coordinates\n5. Add another System and Duplicate the previous one to get the exact same setting\n6. Change the Start and End of the new System to 0 - 100\n7. Bake all Dynamics\n8. Render Frame 199 and compare to rendering of 101. The particles in 199 are tiny in comparison (or invisible). They should look near identical\n\nOptional: \n9. Remove the 0-100 Particle and Redo Baking\n10. Frame 199 now has normal sized particles\n", "File Output nodes output gets overwritten by \"standard\" output (if the same filename is used)\nI have a scene with simple animation. \nI'm rendering muilti-layer open exr with an composite output node with the passes as in them. \nThe standard file 5 year old output bug is still here, where your file saves twice, once from the main file output tab and again on your composite output, normally I don't care whatever, it will just overwrite and I'll end up with one file. \n\nAfter setting up my render output with all the layers I want, normally I just let the overwrite button on regular output be set and it's fine. However all my passes in the exr file were changed to VeiwLayer.(pass), and I'm missing passes like image and alpha or others I might composite into the composite output node. \n\nMy work around is to make the file name something different than my composite output vs the normal file output name. \nBut now I'm saving twice and having to deal with deleting 100s of gigs of junk EXR files from the main file output tab. \n\nIf you render one frame from the file I uploaded, then drop it into a compositor like nuke, fusion or cough cough AE, you'll see the passes are different. \nP.S. You can kill the textures, sorry I didn't pack them for size sake\n\nThank you\n\nblender 3.6 \nFile SYS - (although it seems to be the case on multiple machines, so I thinks it's just an overlook bug): \nryzen 9 5950x\n128 gigs ddr4\nrtx 4090\n\nAnd \nM1max 64 gig integrated\n\n", "Broken material assigment after changing mesh-linking to object-linking\nWindows 11\nNvidia RTX 2060 super\n\nBroken: blender-3.2.0-stable+daily.e05e1e369187\n\n\nOn objects with multiple material slots, the currently selected slot gets bugged when switching from object-linking to mesh-linking. It allows the user to select a different material for each link type, which seems to trigger the bug and has lingering effects. For example, in geometry-nodes setting the material becomes unpredicatable, with material names seemingly swapped.\n\nFrom the provided blend file or startup scene:\n1. create a new material slot\n2. pick a material for this slot\n3. swap the material link type between object-link and mesh-link\n4. select a different material for the same slot\nThe material or slot is now bugged. Geo nodes 'set material' will become unpredictable.\n\n[bug_slots.blend](bug_slots.blend)", "Remove 'Use Nodes'\nWhen using Cycles and editing materials, the user is presented with a button called Use Nodes. This is confusing and adds an extra step before being able to set up materials.\n\nIt might be nicer to simply have the material use nodes always. Materials without nodes doesn't even really make sense when using Cycles as they are central to material editing and creation.\n\n", "ID Deduplication on Append - UX \nThis task is about which behavior should be default when linking/appending IDs, and assets, and how much/how users should be able to customize those behaviors.\n\nNOTE: Currently this is mostly listing topics to be discussed. This is also a first pass, there are likely many more use cases/topics to be discussed.\n\n# Problem\nCurrently, when appending a data-block, Blender appends (i.e. link and make local) the ID and all of its dependencies.\n\nThis means that if a single object in appended ten times, you will get ten copies of the object, but also of its meshes, materials, animations, etc.\n\nThis behavior may be fine in some cases, but not in others.\n\n# Default Behaviors\n## Generic Considerations\nTo avoid having many copies of the same ID, the first appended copy needs to be re-used for the subsequent appends.\n# Do we always re-use that ID, even if it has already been modified by the user locally?\n *Think first appended ID should always be re-used, reliably detecting modified IDs compared to linked one is possible, but tricky and not very efficient (RNA diffing)... There could be a mechanism though to break that reference such that next append creates a new copy of it.* **No diffing.** User has ways to make an existing local ID not re-usable anymore, and to force not re-using an ID when appending again. \n# ~~Could we use the UserPreferences settings for data copying, to decide which ID types we systematically create new copies for, and which ID types we try to re-use previously appended ID?~~\n\n## Specific Cases\nHere is a list of different use cases, possible behaviors being:\n* Link.\n* Append (Link + Make Local, aka new copy).\n* Re-use of previously appended local ID.\n\n### Append Material\n| Data type | DEFAULT | Alternative Behavior 1 |\n| --- | --- | --- |\n| Material itself: | re-use | *duplicate* |\n| Image: | NA | re-use |\n| Node groups: | NA | *duplicate* |\n| Actions: | NA | *duplicate* |\n\n### Append Object\n| Data type | DEFAULT | Alternative Behavior 1 |\n| --- | --- | --- |\n| Object itself: | *duplicate* | *duplicate* |\n| Obdata: | re-use | *duplicate* |\n| Actions: | re-use | *duplicate* |\n| Materials: | re-use | Apply Materials Rules ? |\n\n### Append Collection\n**Collection itself:*** In case of Empty-based instantiation:\n***In case of ViewLayer-based instantiation:** Object: *same as above for Object case?*\n\n# User Customization\n## Change behavior in UserSettings\nShould user be able to change default behavior in user settings, or only on a case-by-case basis when actually linking/appending IDs or assets?\n\n## Change Behavior in Asset/File browsers\nShould user be able to toggle options like \"Link everything\", or \"Never deduplicate\" ? etc.\n", "Vertex groups with identical names can exist, and when they do, a lot of things break.\nRenaming a bone of an armature also renames the matching vertex group of linked objects/obdata, if found.\n\nCurrently there is no check made for potential collisions in vgroups names, which can lead to two or more vgroups having the exact same name, which is considered as invalid data.\n\nThis could be solved in several ways:\n- Rename the existing/clashing vertex group *(and make sure all usages elsewhere are updated along)*\n- Change the specified bone name to be unique in a way that it does not conflict with any existing vertex group name.\n- Merge the (renamed) vertex group with the already existing one with that new name.\n- Refuse to rename the bone in case it would conflict with another existing vgroup name.\n\n--------------------------\n\n*Original Report*\n\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.71\n\nBroken: version: 2.93.5\nBroken: version: 2.83.0\nDidn't test earlier than that (thought 2.79 is too different to matter); didn't test intervening versions; 2.83.0 is just the oldest version since 2.80 that got saved in my Blender folder.\n\nRenaming bones in an armature renames vertex groups of child mesh objects without dealing with name collision. When that happens, a lot (most?) functions involving vertex group weights no longer work correctly-- it appears that they assume vertex group names are unique identifiers.\n\n1) Make a monkey. Give it a vertex group. It's named \"Group\".\n2) Create an armature. Parent the monkey to the armature with automatic weights. Monkey now has VGs \"Group\" and \"Bone\".\n3) Enter edit on the armature. In properties/bone/name, change the name of Bone to \"Group\".\n\nMonkey now has 2 VGs named \"Group\".\n\nAnd, when they get created, stuff breaks. I consider the real bug here to be, \"Vertex groups can acquire identical names\", but one could also consider the bug to be, \"Multiple operations assume VG names to be unique identifiers when they are not necessarily unique.\"\n\n\n\nHere's a quick screenshot of a bugged mesh (not one that I made.) Notice the sidebar list of vertex groups: we've got three different duplicated vertex groups in that list, by my count.\n\nWhat breaks when this happens? The first obvious thing, apparent from the screenshot, is interpolation of vertex groups from a subdivision modifier. We can see that the selected vertex group is no longer interpolating smoothly between verts.\n\nWhat else breaks? A \"normalize all\" operation no longer actually normalizes all vertex groups properly. Post normalization, these don't add up to 1:\n\n\n\nI'm sure that there are other operations that would stop working as well-- those are just the quickies that I noticed. If subdivision interpolation doesn't work right, I wouldn't expect poke faces interpolation or inset faces interpolation to work any better, for example. But I haven't tested all that.\n\nI'd consider this a really low priority bug. The problem doesn't actually show up that often (I'd estimate five times in five years for me.) And it's reasonably easy to work around, provided you're willing to throw your weights away-- you just delete all groups and start again. If unwilling to delete all groups, you can probably get what you want with some math on the VGs (with a VG mix modifier or with geo nodes) and then delete only the problem groups.\n\nUnfortunately, the one big problem with this bug is that it's not very apparent when it happens, which can lead to extra wasted work." ]

[ "Node editor -> Make new material. -> clear previous instead copying. Interface inconsistance\n2.80 (sub 45), 798d2cadc307, 2019-03-04 00:30\n\nSame element, same button, different behavior:\n[2019-03-04_10-04-59.mp4](2019-03-04_10-04-59.mp4)" ]

[ "Black artifacts on smooth surface in 2.79 64bit daily build\nOperating system: Windows 10\nGraphics card: RTX 2080Ti\n\nBroken: blender-2.79.0-git.d7cecc2ba3f6-windows64\nWorked: 2.79b\n\n\nThe software renders black artifacts on a smooth surface. This happens both in the rendered preview and in the final render. Artifacts disappear when smooth is not applied.\n\n\n\n\n\n\n\nJust go in rendered preview mode in the editor, or just F12 to see the artifacts in the render.\n\n[bug report.7z](bug_report.7z)", "Lag Adjusting Value Sliders & Navigating Nested Menus with a Wacom Tablet\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: Intel(R) UHD Graphics Intel 4.5.0 - Build 27.20.100.8681\n\nBroken: version: 3.0.0 Alpha\nWorked: 2.93.2\n\nTrying to change the value of shapekeys or shader sliders with my stylus doesn't get detected the first few seconds, then picks up, but in a non responsive way. Also while using my stylus, if I navigate to a nested menu like say \"Object > Transform\" There's a Lag when moving to another option in the \"Object\" Menu.\n\nThe steps above should apply. I did notice though, that if I change my Tablet API from 'Automatic' to 'wintab' in the the input preferences, it helps, but its still not as responsive as it is in 2.93\n\nAll these issues though, were non existent, if I tried using my laptop's trackpad. Also, my tablet is a Wacom Intuous Touch 5 with a stylus that supports up to 2048 levels if pressure sensitivity and has tilt support, just incase it could be any of these causing the issue. \n\nThanks.\n\n", "Keymap: custom key modifiers work inconsistently\nOperating system: Arch Linux, no DE, i3wm\nGraphics card: GTX 960\n\nBroken: 2.80\nWorked: 2.79\n\nI'm left-handed on the tablet, so I bind my sculpt hotkeys to the numberpad. Binding NumpadEnter+LMB to sculpt inverted (same as holding Ctrl) no longer works in 2.80. However, binding NumpadEnter+MMB to pan the view does work. Not sure where the conflict is. This setup did work in Blender 2.79.\n\nLoad factory settings\nImport .py keymap provided (or bind manually; sculpt.brush_stroke and view3d.move)\nEnter sculpt mode\nAttempt to sculpt inverted by holding NumpadEnter+LMB\n\n[system-info.txt](system-info.txt)\n[custom_modifier_bug.py](custom_modifier_bug.py)\n\nSimilar bugs from ages ago:\nT44349\nT34822", "Viewport display broken for fire- solid and rendered mode (Cycles & Eevee)\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1070 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 431.86\n\nBroken: version: 2.83 (sub 0)\nWorked: (optional)\n\n[Fire seems to get 'stuck' in the viewport (as in it does not disappear), it does however behave correctly when F12 rendered.\nThis happens in 2.83Alpha and 2.82Beta (both downloaded as of today, approx 16:00 Sydney time 15 Jan 2020)]\n\n[Bake the attached and observe.\nhere's what it looks like-\n[ViewportFire_2-82+2-83.mp4](ViewportFire_2-82_2-83.mp4)\n\nIt works Ok when F12 rendered as seen here-\n[Fire26Nov20190001-0050.mp4](Fire26Nov20190001-0050.mp4)\nI couldn't be bothered to render out the clip in Cycles but stills show the error is only in the viewport-\n\n\nHere's the .blend-\n[Fire2.83A.blend](Fire2.83A.blend)\n\nThis works properly in the 26November2019 build, here's what that looks like-\n{[F8279386](ViewportFireMantaflow_26-11-19Build.mp4)}]\n\n", "3D Texturing: Fix seam bleeding non-manifold meshes\n", "No way to edit baked cloth simulation with particle edit mode.\nOperating system: Windows 10\nGraphics card: NVIDIA GeForce GTX 680\n\nBroken: 2.82\n\nCan not edit the cloth cache inside of particle edit mode. I am aware of the fade time error inside of options, I am not using fade time I am purely just seeing if what the manual says is possible, editing a cloth sim cache\n\nBased on the default startup or an attached .blend file (as simple as possible).\n\n1. add a cloth sim\n2. Bake cloth sim\n3. go into particle edit mode\n4. no way to edit cloth sim\n\n[ClothEditTest.blend](ClothEditTest.blend) I have uploaded the file I am testing it with", "Reading from old 2.79 file. Mesh deform broken.\nOperating system: Windows-7-6.1.7601-SP1 64 Bits\nGraphics card: GeForce GTX 660 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.71\n\nBroken: version: 2.91.0 Beta\nBroken: 2.83\nBroken: 2.82\nBroken: 2.81\nBroken: 2.80\nWorked: 2.79b\n\n\nOriginal (2.79)\n\n\nAfter loading to 2.83.8+ versions\n\n\nFile\n[MD.blend](MD.blend)\n", "Modal Operator Breaks on Workspace Change\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 441.66\n\nBroken: version: 2.83.0\nBroken: 2.82.7\nBroken: 2.81.16\n\nWorked: 2.79b\n\nModal Operators appear to break when changing work spaces or fullscreening a window space.\n\nOpen attached file\nRun script (OperatorModalDraw.py)\nIn view3D run operator\nEither change workspace or RClick the header to maximize the view3D. \nDrawing broken and a console error is thrown persistently: \"internal error: modal gizmo-map handler has invalid area\". This message is only shown for 2.83.\n\n[bug_op_modal_fullscreen.mp4](bug_op_modal_fullscreen.mp4)\n\nBlend file: [modal_op_fullscreen.blend](modal_op_fullscreen.blend) \nIt's an empty scene. The script used is the template Operator Modal Draw with a few minor changes to allow events to pass through and to not end on L/RClick. \n\n", "Brushes Assets Viewport User Interface\n{[F13331829](image.png), size=full}\n\nAdditional Notes:\n\n{[F13331831](image.png), size=full}", "Particle Edit Breaks After Selecting a Mesh with Cloth Physics\nOperating system: macOS 10.13.6\nGraphics card: NVIDIA GeForce GT 650M 1024 MB\n\nBroken: at least 2.82 up to 2.92 experimental (I suspect it never worked since 2.80)\nWorked: 2.79\n\nParticle edit behaves like object edit mode, but the UI shows \"Particle Edit\" being active.\n- There is no circle typical for the comb tool around the mouse cursor.\n- No additional key points to select or black hair strands are rendered.\n- Instead of e.g. combing you will move the object around once you drag with the mouse.\nBasically everything except for the UI stays in \"Object Mode\".\n\nEffect On (broken):\n\n\nEffect Off (works as expected):\n\n\nBased on the default startup:\n1. Go to \"Particle Properties\"\n2. Add a particle system\n3. Set its type to \"Hair\"\n4. Add a second particle system\n5. Set its type to \"Hair\" too\n6. Add a new mesh e.g. \"Plane\"\n7. Go to \"Physics Properties\"\n8. Add \"Cloth\" physics\n9. Select the original \"Cube\" again\n10. Switch mode to \"Particle Edit\"\n\nWays to reverse the effect (make \"Particle Edit\" work properly again):\n- Either: Make a new file and append both objects of the old file\n- Or: Delete one of the particle systems on the \"Cube\" and undo the operation\n\nEven after reversing it you can restore the effect by selecting the \"Plane\" once.\nAt least two particle systems are required for the effect.\n\nThe effect can even be saved and stays the same after loading from the file again.\nThus I was able to create two files one with the effect and one without.\nThe file without the effect is a little bigger in size.\n\n[EffectOn.blend](EffectOn.blend)\n\n[EffectOff.blend](EffectOff.blend)\n\nIt took me a whole day to find the cause and boil it down so far.\nHope it helps and that it can be fixed with the provided info.", "different generation of graphic card won't utilise together.\nOperating system: Windows 10\nGraphics card: Radeon R9 295X2 (secondary chip) + Radeon RX560 4GB\nAMD Ryzen 1700X\nMSI X370 Gaming Pro Carbon\n3 * 8GB DDR4-2400\n\nBroken: 2.82.7 to 2.83.3 <--- where x is from 0 to current stable version\nWorked: 2.8, 2.83.4 (partly worked, may end up sudden wattman reset)\n\nRadeon Version: 20.7.2\n\nafter the 3rd step, the render did not show anything, switching back to material viewport, the whole program was frozen. not sure if that is graphic card problem, but that way i could not even render anything. on 2.82.7, this also happened and the render kernels failed to work.\n\n\n1. open up a file of as shown\n2. set the preferences that would use OpenCL to render with R9 295X2 (and RX560)\n3. preview in Cycles\n4. unable to load the preview, switch back to material viewport.\n\n[zeta school 1.blend](zeta_school_1.blend)", "Regression: GPU Subdivision: EEVEE doesn't refresh viewport correctly with Armature modifier\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.94\n\nBroken: version: 3.3LTS, master\nWorked: 3.2.2\n\nGPU Subdivision doestn't refresh viewport correctly an object when Armature and Subdivision Surface Modifiers are used.\n\n1) Start Blender.\n2) Add Armature object\n3) Switch to \"Shading\" workspace.\n4) Select the Cube and add `armature modifier` then `Subdivision Surface Modifier`.\n5) Switch to Properties Editor to \"Modifier Properties\"\n6) Add armature in armature modifier's object property. \n7) Switch between object-edit mode\nIt works fine when \"GPU Subdivion option\" is turned OFF (Prefrences > Viewport > Subdivision > GPU Subdivision)\n\nI added \"BeforeGPUSubDBug.blend\". You can reproduce this issue by loading the blend file and doing step 7)\n", "Loop deselect doesn't work in 3 button emulation mode\nOperating system: Windows 7\nGraphics card: irrelevant\n\nBroken: 2.80, 9a8b6d51c12, 2018-12-11 01:35, Branch: blender2.8\n\nWith three button emulation and left click select on in user preferences, no methods for deselecting edge loops work.\n", "Crash while using the Cloth Filter brush with collisions enabled\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: Radeon (TM) RX 480 Graphics ATI Technologies Inc. 4.5.14739 Core Profile Context 21.6.1 27.20.22001.14011\n\nBroken: version: 3.0.0 Alpha\n\nUsing the cloth filter brush with collisions enabled on *both* the deforming object and the non-deforming object causes Blender to crash.\n\n- Open the attached .blend file with a minimal scene where this error can be observed. Everything has been already set up.\n- With the pre-selected *Cloth Filter* brush set to *Gravity* and both *Use Collisions* and *Use Face Sets* enabled, `LMB` on the \"Skirt\" mesh, and start dragging to the left of the screen, causing the \"Skirt\" mesh to start flying upwards.\n- After a brief while, release `LMB` and repeat the above operation by dragging to the right. Note how the direction of the cloth filter brush was not reversed this time.\n- Almost as soon as this new sculpting operation is initiated, typically Blender crashes. Occasionally it might take a few more attempts.\n\n[facesets-crasher.blend](facesets-crasher.blend)\n\n**Additional notes**\nIf extensive changes are performed to the \"Collisions\" mesh, the bug will not be able to be reproduced anymore.", "Action editor key frames not displaying sometimes\nSometimes when working with multiply animations and switching through them in the action editor. The Keyframes stop displaying for certain action. \nWhen the Panel with the action editor View is removed and added again, the displaying is working again.\nThis bug was seen in Version 2.76 and daily Build from 10/30/2015.\n\nError:\n\nFixed after closing and opening action editor view\n\n\nMy Project without Mesh. Bug is seen in walk_back_cycle action.\n[animation_editor_error_project.blend](animation_editor_error_project.blend)\n\nThis bug is very annoying in my workflow.\n\nGreetings and thank you for your work!" ]

[ "Controllers Broken \nOperating system: Windows 10 Home Single Language 64-bit\nGraphics card: Geforce GTX 850m\n\nBroken: 2.80, 2018-12-27, 22.33 , 0661bcba3210\n(example: 2.79b release)\n(example: 2.80, edbf15d3c044, blender2.8, 2018-11-28, as found on the splash screen)\nWorked: (optional)\n\nControllers not working properly.... I have attached previews the bug found from previous beta build.... \n\n\n\nBased on the default startup or an attached .blend file (as simple as possible).", "Material Editor UI Issue\nOperating system: Windows 10 Pro\nGraphics card: GTX 970\n\nWorked: blender-2.80, 2018-12-28\n\nMaterial Editor UI Not showing properly\n" ]

[ "Decals crash Eevee on AMD graphics card\nOperating system: Linux 6.2.7 Manjaro \nGraphics card: AMD 7900XT\n\nBroken: 3.3LTS, 3.4.1\n\nUsing decals will crash blender in Eevee. cycles works\n\nas an example file please see the .blend that was linked here by the decalmachine creator: 66438\n\nThe issue there is unrelated but the file can serve to show the issue I am having. The moment I switch to rendered view or materials preview blender will crash. I see no error message in console. The log files states just: \n\n```\nbpy.context.space_data.shading.type = 'RENDERED' # Property\nbpy.context.space_data.context = 'RENDER' # Property\nbpy.context.scene.render.engine = 'BLENDER_EEVEE' # Property\n \n```\n\n", "With enabled Modifier On Cage option crashes on entering Edit Mode if in Render Viewport Shading\nOperating system: Windows-10-10.0.19043-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3070 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 522.30\n\nBroken: version: 3.3.0\n\nIn Eevee, in Render Viewport Shading, an object having a Geometry Nodes modifier with enabled On Cage Modifier Result, that affects a UV Attribute which is used in Shader, crashes upon entering Edit Mode.\n\n - Set Viewport Shading Mode to Render\n - Enter Edit Mode ->Crash\n[displace-preview.blend](displace-preview.blend)", "Crash on sculpt mask brush when Multiresolution modifier is copied from another object\nOperating system: Windows 11\nGraphics card: GTX 1060\n\nBroken: 3.3.0 / 3.2.0 / 2.93.10\nWorked: (First time trying multires hasn't worked on any of the versions I've tested on)\n\n\nCrash on sculpt mask brush when Multiresolution modifier is copied from another object.\n\n\nReplicate problem 1:\n\n - Add a Cube (#1)\n - Add Multiresolution modifier to Cube (#1)\n - Subdivide 4 times\n - Add a Cube (#2)\n - Copy Cube (#1)'s modifier onto Cube (#2) through the use of Link/Transfer Data > Copy Modifier\n - Select Cube (#2) and enter sculpt mode\n - Select Mask Brush\n - Use Mask Brush (Should crash blender)", "Crash creating \"position\" attribute with the wrong type while in Edit mode\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 531.41\n\nBroken: 3.5.0 and 3.6.0 Alpha\nWorked:\n\nWhile in Edit mode, attempting to create an attribute called `\"position\"` that has type not set to `\"FLOAT_VECTOR\"` will crash Blender.\n\nI couldn't get any other attributes to crash when added while in Edit mode.\nCreating the `\"position\"` attribute with the correct type, but incorrect domain does not cause a crash either.\n\nI can't imagine a use case for users wanting to do this, but it came up while I was determining the extent of 107228 and isn't really the same issue.\n\n1. Select a mesh object\n1. Enter Edit Mode\n1. Either:\n 1. Run `bpy.context.object.data.attributes.new(\"position\", \"FLOAT\", \"POINT\")` from the Python Console\n 1. Create a new attribute called \"position\" using the `bpy.ops.geometry.attribute_add` operator and set the `Data Type` to anything other than `Vector`: ", "Console warning \"ToolSettings\"/ \"snap_node_element\"\nOperating system: Windows 10\nGraphics card: Geforce GTX 970\n\nBroken:\n2.80, 8c8979067490, blender2.7, 2019-01-31\n\nI was importing a file from an older blender version and I noticed this error in the console:\n\nIt appears when you hover over the node editor and when toggling edit mode.\n\nThe reason I think it's not an issue which appears because of the different blender version is that I recently released an addon for managing nodes with layers and a few people noticed similar errors when editing a mesh with KitOps or similar addons and then trying to change the nodes.\nI also don't think it's an issue with my addon as I didn't change the file (included) with the addon and I tried it with a clean blender version as well.\n\n[warning_message.blend](warning_message.blend)\n- Open the file\n- Open blender console to see error messages\n- Hover over node editor with the cursor\n- > Warning message should appear", "Crash when scrubbing the timeline when using the line art modifier\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 531.41\n\nBroken: version: 3.5.0\nWorked: 3.4.1\n\nWhen scrubbing the timeline, blender will crash. This happens frequently when I have a project of sufficient complexity that uses the line art modifier. The provided blend file has been simplified as much as possible while still reproducing the crashing behavior.\n\n1. Open line_art_crash.blend\n2. Scrub the timeline. Jumping between frames makes it less likely to crash, but holding onto the playhead and slowly moving should have it crash in a few seconds.\n\n", "Toggle to Edit Mode call error when the Blender use linked a object data from another file\nOperating system: Linux 5.10.2\nGraphics card: Intel Haswel (4-th Gen ) HD Graphics 4600\n\nBroken: 2.92\n\n\nToggle to Edit Mode call error when the Blender use linked a object data from another file\n\n\n1) Create any object in a scene and save this file \n2) Create a new blend file and link the object from previous saved a file with `File` → `Link..` and then select this object.\n3) Then convert linked object to local object with `Object` → `Relations` → `Make Local...` → `Selected Objects`\n4) Select the object and press Tab that go to Edit Mode. \n5) Error - Unable to execute 'Toggle Editmode', error changing modes\n\nor download both files and then open linked.blend:\n\n[source.blend](source.blend)\n[linked.blend](linked.blend)\n\n", "Adding new hook while in Edit mode breaks when trying to undo/redo the operator\nWin7 (32bit)\n\nBroken: 2.78a\n\n\n- Add Path from Curve menu, add new Armature\n- Assign Bone to control point as Hook\n- Blank Hook modifier appears along with real one\n\nScreencast: [screencast0001-0250.mp4](screencast0001-0250.mp4)\n", "Rigidbody crash undoing while animation plays (which was started at frame 1 or below)\n[blender_system_info.txt](blender_system_info.txt)\n\n[blender_debug_output.txt](blender_debug_output.txt)\n\n[animation_john2.blend](animation_john2.blend)\n\nOperating system: Windows 10\nGraphics card: Nvidia 2070 Super\n\nWorked: -\n\nBlender crashes when object is rotated to a steep vertical angle and animation player is run.\n\nFor animation_john2.blend\n- Play Animation\n- Select a bone (could be any other operator as well I think)\n- Undo (this will also jump back to frame 0 -- where we started -- which is the important part, starting animation at frame 2 or higher does not result in the crash)\n- Redo\n\nRepeat until crash. Occurs between 2-5 repeats.\n\nAlso occurs at other similar high angles.", "bpy.ops.constraint.apply crashes when owner='BONE' and in object mode\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2060 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.94\n\nBroken: version: 3.4.1\nWorked: unknown\n\nThe bpy.ops.constraint.apply operator crashes blender when targeting a bone constraint while in object mode. The attached blender file reliably causes the crash. It also seems to crash in pose mode at times, but I've so far been unable to recreate this in a small example file. See also [#97640](T97640) where a crash relating to the operator being used on objects has seemingly been fixed between 3.1.2 and 3.2\n\nto recreate simply open the blender file and run the script.\n[apply_bone_constraint_crash.blend](apply_bone_constraint_crash.blend)\n[apply_bone_constraint_crash.crash.txt](apply_bone_constraint_crash.crash.txt)\n\n", "Trim tools doesn't work with Dyntopo enabled\nBroken: blender-2.92.0-ef5d6e9c457a-windows64\n\nEnable the experimental \"tools with missing icons\"\nFile ´/ New / Sculpting\nEnable Dyntopo\nChoose the box or trim lasso tool and try to trim the mesh\nDoesn't work", "Geometry Attributes panel in Edit Mode (design pending)\nThis task is going to adding a unified panel for editing attributes of the selected element by value in edit mode. It makes users able to pass arbitrary per-element data to geometry nodes, opening up more possibilities for procedural modeling.\n\nAll data types (int, color etc.) should be supported. \nDomain supports should match selection modes in edit mode (eg. no face corner support for meshes). \n\n## Basic\nAttributes on the active element is listed in the side panel. \nItems are grouped by domain.\n\n\n## Multiple Selection\nfor float types (float, vector, 2d vector): \n* Show a mean value computed from all selected elements. \n* Button offsets values of all selected elements.\nfor other types: \n* If an active element with a matching domain exists, show the value of it.\n* Button only affects the value of the active element.\nFill operators are provided for batch editing. It fills current button value into all selected elements.\n\n", "Working on a mesh that has a high multires subdivision causes blender to build Up memory even on lower levels.\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 536.99\nRAM: 16GB\nCPU: AMD Ryzen 7 3800\n\nBroken: version: 3.6.0\n\nWhen editing a mesh that has been subdivided a few times will cause blender to increasingly start to consume more and more memory aggressively even if the modifier is turned off.\n\nSubdivide the mesh a few times using the multiresolution modifier, noted on 5-up levels on a mesh with originally 9 272 faces. then either switch the modifier Viewport Level to either 1 or disable It entirelly, start editing the mesh in edit mode the memory consumption stats will start to skyrocket with no signs of going back down until blender starts slowing down the computer and/or crashes entirely. (Got to about 32 GBs some times about where the computer would stop handling It anymore) I haven't tried doing it in workbench I think so I'd suggest trying to do It in Eevee mode too.\n", "Region overlap crashes when interacting with action zones\nEnable region overlap user preference.\n\nIn the default 3D viewport, press NKey, then drag on the corner widget. -> Blender crashes.\n\nAttached asan output: [region_overlap_azone.txt](region_overlap_azone.txt), [region_overlap_azone_2020_01_28_update.txt](region_overlap_azone_2020_01_28_update.txt) (found while investigating #46764).\n\nUpdated asan output\n\n\n----\n\nNote: Blender internals aren't well suited of handling regions being manipulated like this while the rest of Blender is being used. We could either prevent interaction during the fade, or disable it since its not really needed.", "Blender crashing when rendering with eevee, not with cycles however.\nOperating system: Linux-5.15.82-1-lts-x86_64-with-glibc2.36 64 Bits\nGraphics card: AMD Radeon RX Vega (vega10, LLVM 14.0.6, DRM 3.42, 5.15.82-1-lts) AMD 4.6 (Core Profile) Mesa 23.0.0-devel (git-e20564cfdb)\n\nBroken: version: 3.3.2\nWorked: None, I've tried 3.3-LTS, 3.4 (both the default version provided through steam and manually selecting 3.4 LTS) and the daily-alpha build (with the daily-alpha build I reset settings all to default as well) All of these versions are the ones provided through Steam.\n\nWhen rendering in eevee some of my meshes completely bug out as if only part of them are rendering (I've looked at my normals and they're all fine) and after that my entire PC crashes, forcing a power cycle. The viewport display, when in rendered mode, looks fine in eevee but after rendering and before my PC crashes the viewport also messes up with my meshes looking bizarre. This issue does not occur in cycles when using CPU or GPU compute and other blender files work fine in eevee / cycles both in the viewport and when rendering. I also tried applying all modifiers despite needing a displacement on my torches to animate them but no luck. My blend file also becomes corrupt sometimes after the crash.\n\nSimply just hitting the render button with my current .blend file when in eevee. Sometimes the render goes through with broken meshes (which then I can close the render window and switch to eevee in my viewport display but my meshes will be all messed up and then my PC will crash, sometimes it crashes before the render output. \n\n\n[Dungeon Course.blend](Dungeon_Course.blend)\n\n[Dungeon Course.blend1](Dungeon_Course.blend1)\n\n[3.4 config.7z](3.4_config.7z)" ]

[ "Seg fault crash with various edit functions\nOperating system: Linux-5.10.55-1-MANJARO-x86_64-with-glibc2.33 64 Bits\nGraphics card: Mesa DRI Intel(R) HD Graphics 4600 (HSW GT2) Intel Open Source Technology Center 4.5 (Core Profile) Mesa 21.1.6\n\nBroken: version: 3.0.0 Alpha\nWorked: An older daily build from about a week ago, also 2.93\nCaused by 647a8bff06\n\nNewest Blender master daily build immediately crashes when trying to edit the attached test mesh. Also the shading on it in edit mode appears broken. None of these problems are present in Blender 2.93 from distro repo.\n\n1. Open the attached Blend-file edit crash 1.blend\n2. In edit mode, select 2 or more vertices on the mesh\n3. Press J to connect them through an edge\n\nAlternatively:\n2. In edit mode, select all vertices / faces\n3. Use Mesh > Normals > Flip\n\nAlternatively also:\n1. Import the attached OBJ-file edit crash 1.obj into a new document\n2. Apply the same editing steps\n\n**Additional info**\nThe problematic mesh was extracted from a bigger document, but even after appending only the mesh data or exporting it as OBJ or FBX, importing and saving into a new document all through Blender 2.93, the crash is still reproducible in 3.0. Strangely, I'm not able to reproduce a crash or the shading issues if I start from a new document and edit a cube or poly plane into the same shape.\n\nI disabled all add-ons that could have interfered, tried the Cycles-X branch build from yesterday and ran Blender both on the integrated and on the dedicated GPU through `prime-run` on this laptop, but the outcome is always the same.\n\nThe generated crash log is also attached.\n\n[edit crash 1.crash.txt](edit_crash_1.crash.txt)\n\n[edit crash 1.obj](edit_crash_1.obj)\n\n[edit crash 1.blend](edit_crash_1.blend)" ]

[ "Render problems\nOperating system: Windows 10 Pro Ameliorated\nGraphics card: RTX 3070 driver version 472.12\n\nBroken: v3.6 and v3.61\nWorked: v3.51 <- Using the same file created in v3.61\n\nWhile rendering a frames of an animated character, the current rendered animation frame usually shows in the 3D viewport in v3.51. However, in the broken version, it fails to show the current frame being rendered, nothing gets updated visually. Then eventually in the 3D viewport the character goes into its T pose for a few frames then the character starts losing visibility of its layers, at that point Blender crashes. \nI will create a .blend file if I am the only one getting this bug.\n\n", "Cycles (and sims?): Support Intel 12th Gen CPUs \nOperating system: Windows 10 and 11\n\n3.2.1\n\nOn Windows, Cycles does not use all Cores on a Intel 12th Gen CPU (\"Alder Lake\", contains Efficiency and Performance cores).\n\nCycles has to use the Windows API function **SetThreadInformation()**.\nOr Blender **SetProcessInformation()**\n\nThe effect is most seen if if you are rendering and minimize Blender to e.g. browse with Firefox.\nOr if you use a render manager that renders in the background.\n\n\n\nI am not sure if there are other functions in blender that might use all cores like simulations?\n\n\n\n**Test results:**\n\n```\n\nWindows 10:\n-\tBlender UI in Foreground: All 4 EFF cores and all 16 PERF cores.\n-\tBlender UI minimzed/background: All 4 EFF cores and 0 of 16 PERF cores.\n-\tConsole/service render All 4 EFF cores and 0 of 16 PERF cores.\n\nWindows 11:\n-\tBlender UI in Foreground: All 4 EFF cores and all 16 PERF cores.\n-\tBlender UI minimzed/background: All 4 EFF cores and 8 of 16 PERF cores.\n-\tConsole/service render All 4 EFF cores and 2-8 of 16 PERF cores.\n```\n", "Inaccuracy in 'Info' editor\nThis bug is difficult to describe specifically, since I have no idea of *all* of the affected properties. All I noticed is that *some* are reported inaccurately (i.e. the reported rna path does not exist or does not point to the changed property).\n\nA few I found are on the View Layer tab in the Properties editor:\n`bpy.context.scene.view_layers- [x].cycles.denoising_store_passes` is reported as `bpy.context.scene.denoising_store_passes`\n`bpy.context.scene.view_layers- [x].cycles.pass_debug_render_time` is reported as `bpy.context.scene.pass_debug_render_time`\n`bpy.context.scene.view_layers- [x].cycles.use_pass_volume_direct` is reported as `bpy.context.scene.use_pass_volume_direct`\n`bpy.context.scene.view_layers- [x].cycles.use_pass_volume_indirect` is reported as `bpy.context.scene.use_pass_volume_indirect`\nPlus all of the Cryptomatte and Denoising properties.\n\nI have no idea how the Info editor receives this information, but it looks as if the message is partially hard-coded and not up to date?", "Slow EEVEE timeline scrubbing when using Scene World material\nOperating system: Arch Linux\nGraphics card: Intel UHD 620 (`iris` with `INTEL_DEBUG=reemit`)\n\nBroken: `0781c22ceedc`, `2.90`\nWorked: \n\n\nScrubbing through the timeline on an empty scene is slow when viewing through EEVEE and Scene World enabled, even though the node graph for it is a simple Background Shader node.\n\nThis also happens when *nothing* is connected to World's material output.\n\n[slow_scrub.mp4](slow_scrub.mp4)\n\nYou can see in the first part of the video, when using LookDev and have Scene World disabled, the scrubbing is relatively fluid and very close to the cursor. After enabling Scene World you can see that the scrubbing is lagging behind the cursor.\n\nBased on the default startup or an attached .blend file (as simple as possible).\n\n1. Create a new scene\n2. Delete everything (not really necessary)\n3. Try scrubbing the timeline without Scene World in LookDev\n4. Try scrubbing the timeline *with* Scene World in LookDev\n5. Notice that the latter is less responsive\n\n**Note**\n\nI have only experienced this on this GPU (probably due to its low-performance), however I still feel this is a weird bug.", "Blender 3.5 crash on render with Intel 4600 integrated GPU\nOperating system: WIndows 10 x64\nGraphics card: Intel 4600\n\nBroken: 3.5 \nWorked: 3.4.1\n\n\nBlender crashes when attempting to render in Evee. Cycles seems to work.\n\n", "Command Line Rendering with EEVEE is slow\nWin 7 Pro 64Bit \nGTX 1070\nDriver: 417.22\n\nBroken:\nblender-2.80-606223f6a61-win64\n\n\nIf I render with EEVEE via Command Line, the GPU Load is around 40-50% (4min 33sec).\nRendered the same scene in a normal way via Gui, the GPU Load is around 90-98% (3min 20sec).\n\nIf I render the same scene with Cycles via Command Line, the GPU Load is around 90-98%\n\nblender -b PathToBlend -a\n\n\nI've tried several scenes with the same result all the time.\n\n[cube-eevee.blend](cube-eevee.blend)\n\n[cube-cycles.blend](cube-cycles.blend)\n", "UI: Annotate Tool - Missing Layer and Color Parameters\nOperating system: Windows 10\n\nBroken: 3.5.0, 3.6.0 (2023-04-06)\n\nIn some places, layer and color parameters are missing:\n\n\n", "race_spaceship.blend demo file renders empty result (no renderlayer selected in Render Layers node dropdown)\nOperating system: Windows 11\nGraphics card: Radeon Graphic\n\nBroken: 3.6\n\nScene not display well as final render (demo file).\n\n1. Start Blender File.\n2. Check if Viewport working (all great).\n3. Render (seems going well).\n4. Final image (no image, trasparent background).\n\n", "Blender 3.5.0 Win11 RTX3090 Fatal Crash on activating Cycles\n@deadpin thanks for replying.\n\nDriver Version: 31.0.15.3161 (NVIDA 528.24) DCH Win1164 installed\n\nCycles now working without issue thank u.\n\n_Originally posted by @Jaymi-McManus in /blender/blender/issues/106953#issuecomment-923094_\n\n>>>>>\n\nUpdate: \n\n1. Closed the ticket too soon, after adding some simple color, specular, metallic, changes in BSDF and then navigating between layout and other tabs, fatal crash occurs again\n\n2. Updated NVIDIA drivers to 31.0.15.3161 (NVIDIA 531.61) DHC / Win11 64 the issue still persists\n\nUpdate logs attached.", "bpy.context.temp_override - TypeError: Area not found in screen\nBlender 3.5.1\n\nWhen I try to run `bpy.context.temp_override` with area that's not present on current screen I get the error below. \nIt's useful feature that helps to run operators without switching screen.\n\nTried the same thing with old context override workflow like `bpy.ops.node.clipboard_copy(temp_override)` and it works without errors.\n\n```\n# Error: Python: Traceback (most recent call last):\n# File \"test_nodes_copy.blend\\error\", line 37, in <module>\n# File \"test_nodes_copy.blend\\error\", line 25, in copy_node_graph_to_active_object\n# TypeError: Area not found in screen\n```\n\n## ways to reproduce\n\nI've attached .blend file with the setup and code below that's causing the error - to reproduce you can open it and just play the \"error\" script from \"Scripting\". There is also \"working\" script that using old context override and it works without errors.\n\nCode that causes error (the code itself is copying node graph from one material to another - haven't found other method to do so besides calling operators on areas from other screen):\n\n```python\nimport bpy\n\ndef get_shader_editor_context():\n for screen in bpy.data.screens:\n for area in screen.areas:\n if area.type == \"NODE_EDITOR\":\n for space in area.spaces:\n if space.tree_type == \"ShaderNodeTree\":\n context_override = {\"area\": area, \"space\": space, \"screen\": screen}\n return context_override\n\ndef copy_node_graph_to_active_object(material):\n temp_override = get_shader_editor_context()\n old_material = bpy.context.object.active_material\n new_material = material\n \n # remove all nodes from the current material\n for n in old_material.node_tree.nodes[:]:\n old_material.node_tree.nodes.remove(n)\n \n # change current material and make other window's shader editor is updated\n bpy.context.object.active_material = new_material\n temp_override['space'].node_tree = new_material.node_tree\n \n with bpy.context.temp_override(**temp_override):\n # select all nodes and copy them to clipboard\n for node in new_material.node_tree.nodes:\n node.select = True\n bpy.ops.node.clipboard_copy()\n\n # back to original material\n bpy.context.object.active_material = old_material\n temp_override['space'].node_tree = old_material.node_tree\n bpy.ops.node.clipboard_paste(offset=(0,0))\n\n\ncopy_node_graph_to_active_object(bpy.data.materials['test'])\n```\n\nDeprecated temp context override method that's working:\n```python\nimport bpy\n\ndef get_shader_editor_context():\n for screen in bpy.data.screens:\n for area in screen.areas:\n if area.type == \"NODE_EDITOR\":\n for space in area.spaces:\n if space.tree_type == \"ShaderNodeTree\":\n context_override = {\"area\": area, \"space\": space, \"screen\": screen}\n return context_override\n\ndef copy_node_graph_to_active_object(material):\n temp_override = get_shader_editor_context()\n old_material = bpy.context.object.active_material\n new_material = material\n \n # remove all nodes from the current material\n for n in old_material.node_tree.nodes[:]:\n old_material.node_tree.nodes.remove(n)\n \n # change current material and make other window's shader editor is updated\n bpy.context.object.active_material = new_material\n temp_override['space'].node_tree = new_material.node_tree\n \n # select all nodes and copy them to clipboard\n for node in new_material.node_tree.nodes:\n node.select = True\n bpy.ops.node.clipboard_copy(temp_override)\n\n # back to original material\n bpy.context.object.active_material = old_material\n temp_override['space'].node_tree = old_material.node_tree\n bpy.ops.node.clipboard_paste(temp_override, offset=(0,0))\n\n\ncopy_node_graph_to_active_object(bpy.data.materials['test'])\n```\n\n", "EEVEE AOV of full transparency is bad compared to Cycles\nOperating system: win10\nGraphics card: 1050ti\n\nBroken: 3.4\n\nAOV of full transparent shader is barely noticeable. Cycles works fine.\n\n\n[eevee_aov_transparent.blend](eevee_aov_transparent.blend)\nFor default scene:\n\n - Set fully transparent shader (alpha clip or hashed) and add AOV layer for it\n - Render EEVEE -> AOV layer is bad\n\n", "EEVEE-Next: Known Issues\nThis task lists the minor missing features that are known for not yet working with EEVEE-Next.\n\nPriority of each task needs to be assessed and assigned.\n\n# LookDev\n- [ ] camera world alignement isn't working.\n- [ ] blur background isn't working.\n- [ ] diffuse contribution isn't taken into account.\n\n# Shadow\n- [ ] big performance bottleneck caused by rendering method\n\n# Hardware support\n- [ ] Metal backend doesn't support shadow rendering and GI baking yet", "Hairs not rendered in cycles when edit mode or particle edit mode is active\nOperating system: macOS-11.5.2-x86_64-i386-64bit 64 Bits\nGraphics card: AMD Radeon Pro 5500M OpenGL Engine ATI Technologies Inc. 4.1 ATI-4.6.20\n\nBroken: version: 3.0.0 Alpha\nWorked: unknown\n\nWith the attached file, when you render with F12 while edit mode or particle edit mode is active in the viewport, the hairs are not rendered. In Eevee this does not occur.\n\nOpen attached file, select the bezier curve\nPress f12 to render while in object mode -> hairs are rendered\nSwitch to edit mode or particle edit mode and press F12 -> hairs are not rendered\n\n[Feather hair bug.blend](Feather_hair_bug.blend)\n\n", "Bug with environment texture\nOperating system: Windows 10\nGraphics card: Nvidia GeForce RTX 3050 Laptop GPU\n\n3.6.2\n\nAfter I checked and unchecked \"Open Shading Language\" checkbox Blender crashed, after that I opened my project and turned rendered viewport the environment texture become a one solid color, and as I was rotating my camera it was changing the color.\n\nBased on an attached .blend file you can go to the Render viewport and take a look at this error.\n\n", "Cycles XML, light node does not react to transform\nCycles XML, light node does not react to transform (it was ok before migration to graph/node* node parsing). Let me know if this is a bug or it is by design.\n\nSee attached sample XML.\n\n```\n<cycles>\n<transform translate=\"0 0 -4\"><camera type=\"perspective\" /></transform>\n<integrator max_volume_bounce=\"0\" method=\"branched_path\" sample_all_lights_direct=\"true\" sample_all_lights_indirect=\"true\"/>\n<background heterogeneous_volume=\"false\" volume_sampling_method=\"equiangular\" ><scatter_volume name=\"scatter\" color=\"0.8, 0.8, 0.8\" density=\"0.3\" /><connect from=\"scatter volume\" to=\"output volume\" /></background>\n<shader name=\"point_shader\"><emission name=\"emission\" color=\"0.8 0.1 0.1\" strength=\"100\" /><connect from=\"emission emission\" to=\"output surface\" /></shader>\n\n<state shader=\"point_shader\"><light type=\"point\" co=\"0 0 0\" size=\"0.01\" /></state>\n\n<!-- Light does not react to transform node -->\n<transform translate=\"0 0.5 0.0\">\n\t<state shader=\"point_shader\"><light type=\"point\" co=\"0 0 0\" size=\"0.01\" /></state>\n</transform>\n\n</cycles>\n```" ]

[ "Particles not visible in render\nOperating system: Windows 10 Home 21H1 v. 19043.1466\nGraphics card: PowerColor RX 6600 XT OC\nCPU: AMD Ryzen 7 2700 Pro\n\nBroken: version: 3.0.0 LTS\nWorked: 2.93\n\n**Short description of the error**\nWhile rendering Spring demo with Blender 3.0, the grass is missing for some reason, only flowers is left\n\n**Exact steps to reproduce:**\n\n1. Download and run Blender 3.0 LTS: download/\n2. Download and open Spring demo (Blender 2.8): 5d2607ee6b3001d49bab798a\n3. Render as-is (this impacts both CPU and GPU rendering)\n\nPossibly this can be fixed by manipulating the original file's node system, I just want to know if this is a bug or just some hair/particle systems are no longer supported.\n\n" ]

[ "Blender 3.5 Image editor does not fully refresh on some zoom levels while painting\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 526.86\n\nBroken: version: 3.5.1\nWorked: 3.3.2\n\nBlender 3.5 Image editor does not fully refresh on some zoom levels while painting\n\n\n- Open attached file\n- Scribble on the top left image editor\n- On 3.4.1, 3.5.0 and 3.5.1, the image editor being painted doesn't seem to update the view fully, both only below that uv seam line, and 3.5.0 had more portions that aren't updated compared to 3.5.1.\n- Zoom in/out, it refreshes back to normal.\n\nNote:\n\n- This only happens on some particular zoom levels, sometimes I can get a lower zoom level to produce this error too, by zooming in and out and draw, you could probably find some more zoom levels where the image update region isn't complete.\n- When you zoomed out to some level and it works, and then you could zoom to some other level it could fail again.\n\n| 3.3.2 (correct) | 3.4.1 | 3.5.0 | 3.5.1 |\n|--|--|--|--|\n|  |  |  |  |\n\n------\n\nOriginal description:\n\nWhen zoomed in at a particular level the real time properties of the image editor stop working, In the short video i sent I am painting a texture using the Image editor window, it updates the texture on the model and the UV editor below, but the image editor window doesnt update until I either zoom in or zoom out, At first I thought it was random but then I found it can be easily replicated by simply being zoomed in at the level that I was when the bug was happening.\n", "Making a linked asset datablock local does not clear the asset status\nOperating system: Linux-5.17.15-76051715-generic-x86_64-with-glibc2.35 64 Bits, X11 UI\nGraphics card: Quadro RTX 6000/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 515.48.07\n\nBroken: version: 3.6.0 Alpha\n\nCurrently when linking an asset datablock from a library, making the datablock local does not clear the asset status, creating a new asset implicitly.\nNot so much a bug, but I just talked with Julian Eisel about this and he agrees that this should behave differently.\n\n- link to material asset from a library file\n- localize the material datablock\n\n", "Asset System & UI: Improve memory management for assets and previews\nLoading asset libraries can lead to significant memory use, since all contained assets need to be represented in memory. This may be thousands of assets. Further, the UI may request previews for assets, where each preview has a notable memory footprint (around 70kB for each preview at the default 256x256 image size).\n\nAs it stands, Blender is not mindful about this at all. For example simply calling the \"Add\" menu in geometry nodes editors loads all asset libraries into memory (!), including all previews (!) and keeps all this data in memory until another .blend is read (!). This is bad and wasteful design that users have to pay for.\n\n## Status Quo\n\nMemory and asset lifetime is managed rather differently in the asset browser and the rest of the UI (asset shelves, add menus, etc.).\n\n**Current Asset Browser Memory Management:**\n\nThe situation for the asset browser is not too bad (because it's simpler).\n\n- **Good**: Only the currently shown asset library is loaded into memory.\n- **Good**: Previews are lazy loaded into a cache based on the assets in view.\n- **Good**: Previews are removed if the cache is filled with a resonably small amount of previews.\n- **Bad**: Storage is per asset browser instance, so when multiple are open and showing the same asset libraries, storage is duplicated (especially wasteful for the \"All\" asset library).\n- **Bad**: The preview cache is cleared and reloaded often, causing flickering on common interactions (e.g. undo/redo).\n- **Bad**: The preview cache fails miserably with large asset libraries and causes continuous freezes, making the asset browser unusable with large asset libraries.\n\n**Rest of the UI:**\n\n(Asset views, the node add menu, the node search menu, asset-view template, etc.)\n\n- Challenge: Usually use the \"All\" asset library, meaning all assets are loaded when showing such UIs.\n- **Good**: Memory is shared between these UIs (static storage), so libraries are only loaded and stored once. (Asset browser storage is still separate).\n- **Bad**: No lazy loading of previews, all previews are loaded into memory\n- **Bad**: Only freed when closing the file.\n- **Bad**: Previews are also loaded when none are displayed (e.g. node \"Add\" menu)\n\n---\n\n## How to solve it?\n\n### Good solution characteristics\n\nA good solution should give us the following:\n- Only hold asset previews in storaged that are in view of the user\n- Don't keep asset libraries in storage that the user isn't interacting with (at least not regularly, see next point)\n- Provide fast repetetive access to asset libraries (e.g. assets shouldn't be reloaded every time the \"Add\" menu is called in the geometry nodes editor)\n- Support shared storage. All UIs giving access to assets should be able to access the same static asset library and preview image storage.\n\n(Further improvements are possible still, see below)\n\n### Lazy Loading Previews\n\nPreviews should always be lazy loaded based on which assets the user sees, and only if the previews are actually displayed (e.g. not for the node \"Add\" menu). #109234 does most of the necessary work for this.\n\n### Garbage Collector?\n\nA garbage collector could be implemented to remove unused asset libraries and asset previews from storage. Some kind of user-counting will be needed to detect the unused state, even when multiple parts of the UI display the same asset library (and thus may share storage). By being time based rather than user-count based only, the garbage collector still gives enough time for repetitive access to asset libraries. For example the asset library may still be in storage for multiple calls to the node \"Add\" menu, but be freed from storage if it hasn't been called for at least 5 minutes.\n\nNote that this doesn't need to be asset system specific. There are other UI previews that we load from disk that could make use of this. So this could be a more general UI garbage collector.\n\n---\n\n## Possible Further Improvements\n\n## Partial asset library loading?\n\nUsers rarely ever look at all assets of a library. They typically look at assets of a certain type (e.g. in the node \"Add\" menu), a specific catalog, or the view simply only has space for a subset of the available assets. So this is another case where we can avoid keeping data in storage, e.g. by lazy loading assets or deleting them once they are recognized as irrelevant.\n\nKeeping track of which assets are loaded and which have to be made available can be finicky. Plus we may run into issues when asset library contents were updated in-between partial reads.\n\n## Lazy load extended metadata\n\nIn most cases, all the user sees from an asset is the name, and potentially the preview. Keeping all the asset metadata in storage is a bit of a waste. We could lazy load this as well, although this can become finicky too, since some metadata needs to be accessed faster or more frequently (e.g. basic type information, tags for search, etc.) than other.\n\n## Compressed preview buffers\n\nPreview buffers could be compressed to reduce the memory footprint of each, potentially leading to faster load times too.\n\n", "Asset Browser doesn't show assets when drop-down set to \"All\"\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.94\n\nBroken: version: 3.6.0 Alpha\nAlso broken in 3.5 stable version!\nWorked: This feature is new as of 3.5 and I'm seeing the behavior in both. \n\nAssets from other files are not visible when asset browser preference is set to \"All\" (when assets are part of catalogs)\n\n- Extract uploaded file\n - Open general blend file\n- Set it as asset library path\n- Open asset browser and set preference to \"All' [notice every asset is visible (i.e. engine parts and sculpted sphere)\n- Now open `open-file.blend`\n- In asset browser, set preference to \"All\" (notice engine assets are missing)\nIf switched to \"user library\" then engine assets are visible\n", "Able to paint/select \"through\" disabled UI items.\nBroken: 2.80, 2.81...\n\nWith region overlap, if a UI element is diabled, clicks are passed through it, which can result in paint strokes.\n\n\n1. Load the default scene and enter Texture Paint mode on the cube.\n2. Add a paint slot and make a stroke.\n3. In the UV/Image editor, select the painted image, open the sidebar and switch to Image tab. The UI items should be disabled there now.\n4. Make sure that sidebar overlaps the image.\n5. Click on sidebar background.\n6. Click on any of the disabled items.\n\nIn (5), nothing happens. In (6), the click passes through and leaves a paint stroke on the image.\n\n", "Occasional render glitches when using simulation zones on an object without a material.\nThis bug has an easy workaround, mostly posting this so anyone running into the same issue knows the solution.\n\nOperating system: Linux-5.15.0-79-generic-x86_64-with-glibc2.35 64 Bits, X11 UI\nGraphics card: NVIDIA GeForce RTX 3080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 530.30.02\n\nBroken: version: 3.6.2\nWorked: N/A, 3.6 is the first version with simulation nodes.\n\nIn Eevee, rendering an object with a simple simulation zone applied in geometry nodes and no material assigned to the object results in some rendered frames being blank or displaying odd glitches.\n\nThe workaround is to simply assign a material to the object.\n\nCheck out the attached .blend file for reproduction.\n\n", "Eevee: Occasional render glitches when using attribute node\nOperating system: Linux, specifically KUbuntu with X11\nGraphics card: RTX 3080 (Driver Version: 530.30.02, CUDA Version: 12.1)\n\nBroken: 2.93 - 4.0.0 Alpha, 2023-06-29, f47eed749eaf\nWorked: 2.92 (as available on Steam)\n\nUsing an attribute in an Eevee material causes occasional but severe render glitches. This does not happen consistently on any particular frame. It does not happen when rendering from the command line with `--background`.\n\nTo reproduce, render an animation with the attached Blender file. One in every 50 frames or so are corrupted. Occasionally it gets most of the way through the animation before ever producing a bad frame.\n\nThis file is basically the startup file with two changes:\n1. An attribute node with the name set to \"Color\" is plugged into the base color on the default material.\n2. The sample count is lowered to one. (The issue persists without this, but with a higher sample count, some samples glitch and others do not. The final image is a mix of the glitched and regular samples.)\n\nI have ensured that this is not a problem with my particular Blender settings by deleting everything in `~/.config/blender/4.0/config` and selecting \"Save New Settings\" in the splash screen before opening and rendering the file in the 4.0.0 alpha.\n\nNormal frame:\n\n\nCorrupted frame (as rendered in 2.93, some other versions simply display a transparent hole where the cube was without affecting the background):\n\n\n", "Asset System: cache temp-loaded datablocks\nThis is a follow-up task of #86183 (Pose Library: temporarily load Action asset).\n\nGoal: to be able to repeatedly apply the same pose, without loading the same datablock from the same blend file over and over again.\n", "Regression: GPU Subdivision: EEVEE doesn't refresh viewport correctly with Armature modifier\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.94\n\nBroken: version: 3.3LTS, master\nWorked: 3.2.2\n\nGPU Subdivision doestn't refresh viewport correctly an object when Armature and Subdivision Surface Modifiers are used.\n\n1) Start Blender.\n2) Add Armature object\n3) Switch to \"Shading\" workspace.\n4) Select the Cube and add `armature modifier` then `Subdivision Surface Modifier`.\n5) Switch to Properties Editor to \"Modifier Properties\"\n6) Add armature in armature modifier's object property. \n7) Switch between object-edit mode\nIt works fine when \"GPU Subdivion option\" is turned OFF (Prefrences > Viewport > Subdivision > GPU Subdivision)\n\nI added \"BeforeGPUSubDBug.blend\". You can reproduce this issue by loading the blend file and doing step 7)\n", "Wrong editor type saved in workspace\nOperating system: Win11\nGraphics card: AMD Radeon (TM) Graphics (Included Graphics)\n\nBroken: 3.5.1\nWorked: N/A.\n\nMy properties editor is glitched and keeps resetting to nonlinear animation editor.\n(When saving file and reopening)\n\nIdk how to repreduce it, but here are steps for my blender file:\n1. Switch editor type to properties editor.\n2. Save file.\n3. Close blender .\n4. Open the saved file.\n5. Properties editor switched to nonlinear animation editor.\n6. (When repeating 1-4 with all objects deleted, then its working normally without the switch at restart).\n7. Viedeo: EDgFd6MIzjM.\n\n", "Re-render after saving a render result as image with 'Copy' option turned OFF makes blender unresponsive\nOperating system: Windows-10-10.0.18362 64 Bits\nGraphics card: GeForce RTX 2080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 431.86\n\nBroken: versions 2.79 & 2.81 (sub 16)\nWorked: (optional)\n\nRender with F12, save the rendered image with 'Copy' turned off, and render again. This makes (parts of) Blender unresponsive.\n\n - Render the default cube.\n - Image - Save as, and uncheck Copy in the options.\n - Click Save as image.\n - Re-render the image.\n - Press {key Escape} to close the render window (it doesn't).\n - Press {key F12} to re-render (it doesn't).\n\nStrangely enough some parts of Blender are still responsive. For example {key Shift F12} still switches to the timeline.\n", "Design issue: Icon next to overridden property does not perform the function it strongly suggests\n3.0.0\n\nWhen a propertry is overridden, the \"animate property\" button next to it is replaced by this \"reset\" icon:\n\nThese really seem to suggest that clicking them will remove the override. Instead it just adds a keyframe as usual. This seems misleading.\n\n - Override a property of some library override asset\n - Click the button to the right of that property\n\n", "Weird VSE crash\nOperating system: Win11\nGraphics card: RX580\n\nBroken: (example: 2.80, edbf15d3c044, master, 2018-11-28, as found on the splash screen) blender 4.0.0 alpha\nWorked: (newest version of Blender that worked as expected) no clue\n\nBlender crashes on opening file. Was trying to make an alternative caption that gets around the Multiply color issue #112267 but came across a new bug.\n\nBased on the default startup or an attached .blend file (as simple as possible).\nOpen the file I've attached\n\n", "Crash when playing animation in rendered view (cycles)\nOperating system: macOS-12.0.1-arm64-arm-64bit 64 Bits\nGraphics card: Apple M1 Max Apple 4.1 Metal - 76.1\n\nBroken: version: 3.0.0, 3.5\nWorked: unknown\n\nPlaying animations in cycles rendered view frequently crash blender. With the attached file I consistently get a crash within 1-2 minutes of starting the animation.\n\nThe crashes do not occur when running the intel version through rosetta. They occur much faster/more often when using the CPU than when using the GPU.\n\nOpen attached file\nSwitch to rendered view (Crashes with and without Metal preference)\nPress spacebar to play\n\n[Boat boating.blend](Boat_boating.blend)\n[Animation crash.txt](Animation_crash.txt)\n```\nThread 17 Crashed:\n0 Blender \t 0x10319354c tbb::internal::custom_scheduler<tbb::internal::IntelSchedulerTraits>::process_bypass_loop(tbb::internal::context_guard_helper<false>&, tbb::task*, long) + 340\n1 Blender \t 0x103193644 tbb::internal::custom_scheduler<tbb::internal::IntelSchedulerTraits>::process_bypass_loop(tbb::internal::context_guard_helper<false>&, tbb::task*, long) + 588\n2 Blender \t 0x103192c60 tbb::internal::custom_scheduler<tbb::internal::IntelSchedulerTraits>::local_wait_for_all(tbb::task&, tbb::task*) + 184\n3 Blender \t 0x103183300 tbb::internal::arena::process(tbb::internal::generic_scheduler&) + 252\n4 Blender \t 0x10318c3b8 tbb::internal::market::process(rml::job&) + 40\n5 Blender \t 0x10318d6b0 tbb::internal::rml::private_worker::run() + 288\n6 Blender \t 0x10318d584 tbb::internal::rml::private_worker::thread_routine(void*) + 12\n7 libsystem_pthread.dylib \t 0x1b39cd4ec _pthread_start + 148\n8 libsystem_pthread.dylib \t 0x1b39c82d0 thread_start + 8```", "UI: Discussion Topics & Possible To-Do's for 3.0\n3.0 is the next chance to do breaking changes which should not go into normal releases. That may be: higher level design changes, muscle memory breaking, documentation breaking (e.g. changed operator names) or non-trivial visual changes. There should still be no .blend file compatibility breaking changes.\n\nIn this task we can collect such changes for consideration. This is not a \"wish a feature\" list. These are things we discussed already but couldn't do earlier because they would be too disruptive.\n\n* Active vs. selected (#82359 (Stronger binding of Active and Selected))\n* Keymap polish\n\n - Consistent selection with modifier keys #78253 (Formalize Default Selection Behavior and Key-map)\n - {key 1}/{key 2}/{key 3} for changing into vertex/edge/face edit mode #88071 (Keymap: use number keys for mode switching (2nd Iteration))\n - {key D} for view-navigation sub-menu #88092 (Keymap: use D-key for view navigation pie-menu (2nd iteration)).\n - {key Alt} usage #83689 (Default Keymap Update: Improve tool access using Alt-LMB)\n\n* Toolsystem wrap-up (change toolsystem defaults if needed)\n* Theme polish (#74358)\n* General defaults (Preferences, startup configuration, bundled workspaces, etc)\n - [ ] Set default subdivision modifier levels to be 1 for viewport and render.\n* More dynamic file selection (e.g. show import/export settings based on file selection)\n* Reevaluate closing popups on mouse moves.\n* Improve naming:\n**Avoid \"link\" for anything unrelated to link/append (e.g. *Link Materials* -> *Share Materials*).** Correct usages of \"copy\" (e.g. *Copy Modifiers* -> *Replace Modifiers*).\n* 3D cursor default orientation method to \"Surface\"\n* #89266 (Investigate renaming the \"Video Sequencer\" editor)\n* #83612 (UI: Tracking: Migrate to active tools based toolbar)" ]

[ "Marking objects as assets makes the top bar disappear\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce RTX 2060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.71\n\nBroken: version: 3.0.0 Alpha\n\n\nThe top bar (containing \"file, edit, render\" etc and the workspaces presets) disappears when marking an object as an asset in the outliner\n\n\n* Open Blender in fullscreen (as in covering the whole screen, not with \"Window > Toggle Window Fullscreen\")\n* Add any object to the scene (the bug worked with meshes, curves, cameras, etc)\n* In the outliner, right-click the object > ID Data > Mark Asset\n\nThe bug only works when in fullscreen, not when the window size is smaller. This can be quickly fixed by moving the window around or disabling fullscreen (when enabled again the top bar is there again)\n\n{[F10208187](image.png), size=full}\n\nThank you! The asset browser is a very nice new feature!\n", "Topbar dissapears when asset is assigned to asset browser\nOperating system: Windows 10\nGraphics card: AMD 6700XT\n\n**Blender Version** 3.0\nBroken: (both 3.0.0 and 3.10)\nWorked: \n\nWhen I select multiple objects and try to set them as assets in the asset browser, all fonts start changing size and the menu bar at the top of the Blender completely disappears. It is back when you restart your blender, but it is an annoying issue probably caused by GPU, It doesn't happen on NVIDIA RTX 2060. \n\nI have the newest beta driver for HIP installed.\n\n\n\n", "Sometimes, mark as Asset will cause the menu bar to disappear\nOperating system: Windows-10-10.0.19042-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 471.41\n\nBroken: version: 3.0.0 Beta\n\nSometimes, mark as Asset will cause the menu bar to disappear, This problem does not always happen, but it will happen with a high probability\n\n1. open blender\n2. just mark as asset, than you will, then you will find the menu bar disappeared\n\n[20211122_16.28_165.mp4](20211122_16.28_165.mp4)\n" ]

[ "Move the transform info away from the header, into the viewport.\nCurrently, when you transform any item in Blender, we take over the header while the transform is happening:\n\n\n\nThis makes for a vey flashing/blinking UI, where both the header and the viewport gizmo disappears and re-appears in a jarring way. \n\nInstead, the header and viewport gizmos should stay untouched, and the transform info can be moved to the viewport along the bottom.", "It says when I try to put my render into the downloads file permission denied can somebody help?\nOperating system: ?\nGraphics card: ?\n\nBroken: ?\nWorked: ?\n\n?\n\n?", "Low unit scale values show wrong Values and Units in sliders\nOperating system: Linux-5.8.0-2-amd64-x86_64-with-debian-bullseye-sid 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 450.66\n\nBroken: version: 2.91.0 Alpha\nWorked: not in 2.80 or 2.83\n\nWhen the unit scale is too low (< 0.0001 e.g. 0.00009), the values in all sliders will jump up to the value they would have with a unitscale of 1.0. This causes problems with the clipping planes of the viewport camera (didn't do this in 2.80). This seems to be independend of the unit system (metric/imperial).\n\nIf this is a known limitation then the shown precision of the unit scale slider should reflect that. Also a fix for the clipping plane issue is needed as that did work in 2.80.\n\nBased on default startup.\n1. In Scene Settings: enable seperate units\n2. set unitscale of 0.00009 or less (metric r imperial)\n3. observe what the dimensions sliders tell you about the dimension of the default cube.\n4. observe the buggy floor grid in the viewport probably due to the same bug in the clipping sliders (wasn't there in 2.80)\n", "Make the “redo” panel also appear in the tool settings tab\nWe should make the 'redo' panel also appear in the Tool Settings tab in Properties:\n\n", "Knife Tool produces bad geometry in certain cases (disconnected edges/vertices, overlapping faces)\nif i use tools knife i see this bug.\n\n\n\n", "Lost hotkey with save preset\n2.80\n\n\nI finally can replicate problem from 2.7x and it is still problem in 2.80. When you change some combination of hotkeys and save it to the preset. You lose some of them. On my GIF example Numpad 1 dissapear.\n\nSometimes it come back after restart Blender, but sometimes it is just gone.\n\n\n\n", "GPencil: Randomize brush's 'Position' attribute doesn't work when camera is too close to Grease Pencil object\nOperating system: Win10 Home 64\nGraphics card: rtx3070\n\nBlender Version\nBroken: 2.93, 3.1.0\nWorked: N/A\n\nExact steps to reproduce the error:\n\n- Open file\n- try to use Randomize brush\n- Strength and Thickness work, but not Position \n- Switch to the other camera that's farther away from the GP object, and randomizing the position works fine.\n\n[randomize_not_working_close_NEW.blend](randomize_not_working_close_NEW.blend)\n\n\nwatch?v=4pTrEm5HqVU", "Rim on objects when shadow catcher is in play\n开启阴影捕捉后边缘有硬边BUG \n\n\n\n", "Radius in place of Size at Empty viewport menu\nI{[F6596506](RIGHT.JPG)}\n\n am new to blender and not know where to post this but plz check it.It seem wrong to me\nthanks", "shrink/flatten not working right way?\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 445.87\n\nBroken: version: 2.91.0 Alpha\n\ni think shrink/flatten not working the right way. only top works fine goes up down.\nbut bottom and sides are backwards and a bit counterintuitive.\n\ni dont know if its bug but wanted to report anyway.\n\n\ncheck video\n\nthanks\n\n\n[2020-09-27 04-57-36.mp4](2020-09-27_04-57-36.mp4)\n\n[blender.blend](blender.blend)", "Blender Wont run.\nOperating system: Windows\nGraphics card: AMD Radeon HD 8570\n\nBroken: 2.92\nWorked: Nothing has worked. But blender worked before I did this. \n\nWhen I open blender it closes immediately.\nAnd I get this \n\nwatch?v=3rugIsYwl7s).\nSo Out internet was down, And I wanted to get blender On my other pc account, So I went to the main account And Thinking that the installer required internet, I took the system files, and copied them to a thumb drive and pasted them Into the other account when I did that this first started happening, so when The internet came back On. I just got rid of the files I pasted in and then Installed it. With the same result as last time. But this time the bug/glitch was on all accounts. and I tried uninstalling and reinstalling it from the apps and features and from the control panel. I also installed the app from the windows app store with no luck. If you guys know how to fix this issue then please email me at [email protected]\n Ps, My CPU wasn't hot.\n- - -\nAble to run blender from `blender_debug_gpu_glitchworkaround.cmd` batch file", "Transform cursor (patch included)\nRecently LMB-Drag was made to move the cursor *(edit, this has since been reverted)*.\n\nHowever this just moves the cursor under the mouse,\n\nit _could_ be useful to use transform here, it allows.\n\n- Drag, XKey to lock to an axis\n- ... Shift+Y to constrain to a plane.\n- ... Hold Ctrl to snap (snap to verts for example is handy)\n- ... Shift for more precise placement...\n... most other transform keys.\n\n\nOne reservation I have about this, is that you probably would often want to be able to snap to vertices and edges, and not have to depend on the current snap options in the 3d view header.\n\nSo we could override the snap options that transform uses.... but this seems a bit crappy too, since you may want to snap to faces in some cases, or even snap to grid.\n\nAs an alternative, we could just not bother attempting to integrate this with transform, and add snapping to the current modal operator (similar to how we have with the Ruler operator, which shares low level code with transform snapping)\nBut in that case we miss the axis constraints which seem like they would be nice to support.\n\n\nHere is a test patch to show how it works: [P87](P87.txt)\nNote, it crashes if you try to scale and its hard coded to LMB for cursor place, so its mostly a proof of concept.", "Switching to a mode that has a tool set which is not available for that object causes error [e.g Select Box tool is available if paint mask is enabled, if it is disabled it causes error]\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 442.92\n\nBroken: version: 2.83.0\n\n[when i try open vertex paint or weight paint , brush and color bar not coming to viewport]\n\n[after basic modeling something i apply all modifer and try to create uv but i cant do properly. So after unsuccessful attempt i tried to paint weight paint but color and other bar not coming to screen. But if i create new blend file and append it. Also work and i can create but after some paint also broke vertex paint file and mine locak vertex paint parts gone. btw i also cant open in new file as link...\n\n[try_007_gun.blend](try_007_gun.blend)\n\n\n", "GP: Time offset and Armature modifier causes glitches\nOperating system:WINDOWS10\nGraphics card:GTX1060\n\nBroken: 2.93LTS\n\n\n[#89433.blend](T89433.blend)\n\n- Open file\n- Move selected bone to left until it hits constraint, and then a bit randomly\n\nGP object will keep rotating in one direction and can be reset only by changing frame.\nSame happens when scrubbing in negative frame range.\n", "Adjust modifier values when using \"Apply scale\"\nThis feature request was discussed with Campbell Barton during meeting 2021-02-23\n\nWhen using \"Apply scale\", the modifier values are not adjusted if the object scale is not [1,1,1]. I would like to suggest that the operator \"Apply scale\" iterates through all modifiers of the selected objects and adjusts all values related to scale\n\nFor example\nBevel modifier - Amount\nDisplace - Strength\nData transfer - max distance, ray radius\nUV project - Scale X, Scale Y, \netc\n\n---\n\nCurrent behaviour\n{[F9849514](image.png), height = 700}\n\n---\n\nDesired behaviour\n{[F9849529](image.png), height = 700}\n\n---\n\nsuggested Operator mockup\n\n{[F9852183](image.png), height = 300}" ]

[ "Shrink/Fatten 'Offset Even' option fails\nOperating system: macOS 10.14\nGraphics card: AMD Radeon R9 M395X 4096 MB\n\nBroken: Blender 2.80 beta\nWorked: Blender 2.8 from a few months ago, plus 2.7x\n\nShrink/Fatten Offset Even option doesn't work, and cannot be enabled\n\n\n - Select some faces. \n - Use Shrink/Fatten\n - Try and enable Offset Even\n\nIt doesn't work.\n", "Extrude along normals, Shrink/fatten etc tools issue (Offset Even option is unclickable)\nWin 10 64 bits\n\n2.8 bb0b4b008ed\n\n**Description of error**\nCan't click the Offset Even checkbox in the redo panel when using the Extrude along normals or Extrude individual or the Shrink/fatten tools. (not sure if there are more tools affected)\n\n\nAlso, clicking on the flip normals option seems to be activating the offset even at the same time.\n\n\nAnd lastly, the offset even option in the topbar seems to be working in an inverted manner, like, when the option is checked it works like if it was unchecked, and vice versa.\n" ]

[ "Crash after undo \"Mask slice to new object\"\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 1050/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 461.92\n\nBroken: version: 2.92.0\nWorked: This problem happens to me with 2.92, the pc that I use now I formatted it, unfortunately before I used version 2.82\n\nMy problem is that when I use the option \"mask slice to new object\" and then I use ctrl + Z the blender closes !! I formatted the PC a few weeks ago, unfortunately I didn't use version 2.92 before. What I realized is that when I use the \"mask slice to new object\" option and then go to edit mode, I can use control + Z without any problem, that is, it does not close.]\n\nThis happens to me in any file:\n\n1. Sculpture mode.\n2.Use the \"mask\" or \"box mask\" brushes\n3. I paint the mask.\n4.I go to the \"MASK\" menu.\n5. I click on \"mask slice to new object\".\n6. Press CTRL + Z.\n7. Blender closes.\n", "Additional unwanted objects being created when asset browser asset is used\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA RTX A4500/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 512.15\n\nBroken: version: 3.5.0\n\nAdditional duplicate (unwanted) objects being created when using assets from the asset browser.\n\nI haven't managed to consistently reproduce how I get this issue to happen. However, I have attached files which are guaranteed to demonstrate it.\n\n1. The oxford-straps-demo.blend file contains the source meshes which are placed in a collection and marked as asset called \"Source Asset\"\n\n2. The test-asset.blend file contains the \"Source Asset\" dropped in as Linked then Library Overridden. That itself is then marked again as an asset called \"Test Asset\". (This is my workflow - I would do this, then add materials at this stage).\n\n3. Open a new blend file and drop in the \"Test Asset\" (with Append). You will see the appended asset plus two other unwanted objects which are duplicates of the objects contained within the sealed asset.\n\nI have to delete these every time the asset is used.\n\n", "Collection usability\n- [ ] Viewport UI\n- [ ] Shortcuts to toggle visibility?\n[ ] . . .\n\nNeed a clear design on what to do.", "Design issue: Icon next to overridden property does not perform the function it strongly suggests\n3.0.0\n\nWhen a propertry is overridden, the \"animate property\" button next to it is replaced by this \"reset\" icon:\n\nThese really seem to suggest that clicking them will remove the override. Instead it just adds a keyframe as usual. This seems misleading.\n\n - Override a property of some library override asset\n - Click the button to the right of that property\n\n", "Make Local feature is undocumented\nBroken: version: 2.83.0\nWorked: Never\n\n**Description of error**\nMake Local feature is undocumented\n\nRight click on collection -> \"ID Data\" -> \"Make Local\" <---- no action (not intuitively obvious what the action would be)\n\n", "Materials linked to the object's data are lost when removing attributes from instances in geometry nodes\nOperating system: Windows-10-10.0.19045-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 4090/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 528.24\n\nBroken: version: 3.5.0 Alpha\n\nMaterials linked to the object's data are lost when removing attributes from instances in geometry nodes.\n\n1.Instance a collection using geometry nodes.\n2.Remove an attribute from the instances (the attribute doesn't need to exist).\n3.Materials stored on the object data are now lost.\n\n\n[BugExample.blend](BugExample.blend)\n\n\n", "NLA tracks become zombies (can't be killed nor edited) if you append data from a file where these Tracks were overides on a linked armature.\nOperating system: Windows-10-10.0.19043-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3090/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 531.79\n\nBroken: version: 3.5.1\n\n\nSteps: \n1 - have a file where a character is imported as linked, (link its collection) create library overide of content (classic way of animating a character in production, not having it locally in your anim scene)\n2 - Add NLA tracks to animate it, e.g. a walk action, another track to control the root/position in world ect...\n3 - In a second blender file, append this (scene or collection ect..) the character is still linked, the NLA tracks considered library overides, but they are \"zombies\", you can't delete them, you can't edit them, the clips inside refuse to slide to other tracks, if you duplicate the clips, they get sent to a new NLA track, and cannot slide to these zombie tracks. \n\nThe only option is to un-tick these tracks, lock them, and hope to not get confused later if you or someone else open the file. This is not a show stopper then, as you can duplicate linked all these clips, which just restores your animation cut.\n\n", "Pinning Bones/Constraint Tabs in Properties Editor doesn't work\nBroken: 2.76 - master (1283c66)\nWorked: Unknown\n\nThe context pinning functionality in the Properties Editor doesn't work for Bones and Constraints. There may be a few others too that I haven't noticed yet, but these are the biggest ones.\n", "Crash when delete linked object in scenes outliner\nOperating system: Windows 10\n\nBroken: 3.6, 3.5, 3.3 LTS, 2.93 LTS\nBroken: 2.91.0\nWorked: 2.90.1\n\nCaused by b0741e1dcbc5e4549e95745b1f1b501f8cd33add\n\nBlender is crash when delete linked object in scenes outliner\n\n1. Open start scene\n2. Create new scene \"Scene.001\"\n3. Click on Cube, Ctrl+L, Link objects to scene \"Scene.001\"\n4. In Outliner switch to \"Scenes\"\n5. Select \"Cube\" object in Scenes Outliner in \"Scene.001\"\n6. Move cursor to Outliner, press key \"X\" or \"Delete\"\n7. Blender is crashing\n\n", "New Curves data block\nThis is a follow up task based on discussions we had about #94193.\n\nCurrently, the curve object contains three distinct object types: curves, surfaces and text. We want to split those up into separate types, because there is no real reason for why they are combined into a single type. \nThis task focuses on introducing a new `Curves` data type which we will also need for hair and geometry nodes processing of curves.\n\nInitial steps:\n* - [x] Rename existing (experimental) `Hair` data block to `Curves`. In the ui it will be called `Hair Curves` for now to distinguish it from the old curve type. \n* - [x] Decide on and implement exact data structure used for curves.\n * Should be stored compactly in continuous arrays.\n * Should support arbitrary attributes on points and splines.\n * Ideally, the actual curve data structure is embedded into the `Curves` data block as a separate struct, so that the same struct could e.g. be used for grease pencil.\n* - [x] Refactor geometry nodes to use the `Curves` data block instead of `CurveEval`. (#95443)\n* - [x] Remove the runtime-only `CurveEval` data structure added for geometry nodes.\n* - [ ] Render geometry nodes curves as with the new type in Cycles and Eevee (#96455)\n* - [x] Curves sculpt mode that is mostly focused on hair use cases\n* - [ ] Curves edit mode, similar to existing curve edit mode\n\nFurther in the future:\n* - [ ] Add existing necessary features from the legacy curve object (features like shape keys may add complexity)\n* - [ ] Use versioning to convert legacy curve objects to the new curves object\n* - [ ] Add support for simulations through a generic simulation system involving geometry nodes\n* - [ ] Use versioning to convert hair particle systems to the new curves object\n\n", "Running scripts on append/link\n(I mentioned this at lunch and @dr.sybren said I should make a design task here c: )\n\n# Problem\nMost production rigs come with their own scripts, whether it's just for UI purposes or have actual functionality. It's standard practice for riggers to make these text datablocks a dependency of the rig to make sure that it's brought along when the rig is appended or linked somewhere. However, such scripts will not run on append or link.\n\nThis introduces a piece of poor UX for animators, which in my experience results in a huge amount of confusion. The animator would have to save and reload the file, or open a text editor and run the script file, but they have no reason to know that they have to do either of these, unless the rigger tells them so. @Hjalti and @PabloFournier probably remember running into this on many occasions.\n\n# Proposal\nMy idea is to add a piece of functionality to the append and link operators; A \"Run Scripts\" bool param, enabled by default, with a tooltip like \"Execute scripts which are set to register on file load after appending/linking\". Can probably come up with better wording, but I guess you get the idea; Appending or linking(directly or indirectly) a text datablock with the register checkbox enabled(`use_module==True`), will also execute it.\n\nIf the user has Auto-Run Python Scripts user preference disabled, it should throw up the usual pop-up to request permission to run the script.\n\nAn example file with a rig and rig UI script, all ready to be linked/appended:\n[custom_rig_ui.blend](custom_rig_ui.blend)\n\nThoughts? Are there other use cases? Are there any scenarios where this would cause problems?", "Multi-object-Mode: EditMode transform properties\nEdit-mode buttons currently only show median values for the active edit-object.\n\nThe Mean Crease value of the N panel doesn't work for multiple objects:\nHere i have 2 objects selected\n\nIn Edit Mode changing the Mean Crease value does change its effect only on the selected object:\n", "Allow menus to be added referenced to another menu \n## Motivation\nAddon developers cannot add menus on specified place. \nThey are always added at the end (append), or at begin (prepend)\n\n## Analysis\nAfter a look at current code, i think a good approach should be allow to define a reference label on append function.\n\nWhen drawing the menu layout, a new RNA call should be done on a new function class, that should then execute the referenced drawing calls on py scope.\n\nIssue with this development is that for any but menu drawn two python calls are performed to allow addons draw before and after itself. Maybe this could be solved checking in C scope if it has post/pre functions added.\n\n## Development steps\n* {icon check-square-o color=green} Draw after label\n* {icon check-square-o color=green} Draw before label\n* {icon check-square-o color=green} Give warning if not used (label not found)\n\n## Know Issues\n* Issues with enums, if shown as submenu it's not working. Is shown on self menu, the label is the operator name, and considered a block\n\nFeedback is welcome!\n\n\n\n\n", "Outliner: Object selected/deselected when clicking with RMB on it's object data [`Sync Selection` option causes this]\nBroken: version: 3.0.0\nBroken: version: 2.91.2\nBroken (differently): 2.90.1, 2.83\nWorked: 2.82\n\nWhen clicking with RMB on *some* object data in outliner (can confirm for materials), parent object is selected/deselected (both in outliner and in 3D viewport). That can lead to undesired actions. For example, when click RMB on material object data from a cube and choosing `Delete` you'll get one of the following:\n- Delete only material object data if parent object is not selected\n- Delete cube object itself **and** material object data if parent object is selected\nMesh/ligh/camera object data also have this \"parent object selection/deselection\" behavior but for this object data it's not so apparent in terms of functionality since in both cases mesh/light/camera object data is deleted and object can't exist without those so it's deleted too.\nModifiers and constraints are handled consistent with context menu - deleting only modifier - although it's something to be discussed since it shows that parent object is selected.\nFor shortcut behavior there is already #94062.\n\nIn 2.90.1 core issue is same but behave slightly different - it'll not always change selection/deselection, can provide details if requested.\n\nFor materials:\n1. Open attached file. You'll see two objects with same mesh object data.\n2. Repeatedly click with RMB on material object mata in outliner to see parent object selected/deselected.\n3. Choose `Delete` from context menu when parent object is selected - one of the objects is deleted and one material that was clicked on (you can see it on other objest mesh object data since they share it).\n4. Undo\n5. Choose `Delete` from context menu when parent object is not selected - only one material is deleted.\n\nFor showcasing mesh/light/camera data behavior:\n1. Open attached file.\n2. RMB on `Sphere` mesh object data to a point where parent object is either selected or deselected.\n3. Choose `Delete` from context menu - in both cases all objects are deleted.\n[RMB outliner selection bug.blend](RMB_outliner_selection_bug.blend)", "Blender editing performance with many datablocks\nEditing scenes with many datablocks can be slow, for example adding, deleting or duplicating many objects.\n\nAdding or deleting a single object has time complexity O(N), where N is the number of objects. This is due to unique name tests, looping over all datablocks to clear all pointers, and other operations that require accessing many datablocks. In general all such operations must become O(log(N)) or O(1), along with possible access of just the other datablocks linking to it.\n\nUndo push (#60695) and dependency graph rebuild are also O(N). We are not trying to improve those as part of this task, and note these can be run once after making many edits.\n\nPractical steps to reduce time complexity would be:\n\n* Store two-way links for all (non-COW) data-blocks, so that iterating over all datablocks that link to a given datablock is fast. There are transient data structures for this, but it should become permanent. Every `id_us_min()` and `id_us_plus()` should be turned into an operation that sets the datablock pointer and creates a link in the other direction as well.\n* Extend or modify datablock lists in `Main` to support quick testing for duplicate names.\n* Change view layer collection syncing to not require looping over all objects." ]

[ "Append already linked Data\nBroken: 2.8 e17285a3816\n\nWhen appending data that is already linked into the file, it will also be linked instead of being made local.", "Append grabs previous linked datablocks\nOperating system: Windows-7-6.1.7601-SP1 64 Bits\nGraphics card: GeForce GTX 660 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.71\n\nBroken: version: 2.83.9\n\nLink collection from another file, delete it, then append it. you will have linked datablocks instead editable.\n[2021-01-20_16-38-29.mp4](2021-01-20_16-38-29.mp4)\n\nYes, I understand, I can purge linked datablocks before append, but what if I want to use both?", "Appending object that is part of linked collection imports object as link\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 445.87\n\nBroken: version: 2.83.3\nBroken: version: 2.91.0 Alpha\nBroken: 2.79b\n\nWhen going File > Append, the object that is selected in the dialog is brought in as a linked object.\n\n1. Link in collection as instance that contains multiple objects\n2. Try to append one of the objects contained in the linked collection. \n", "Append and Link file conflict\nOperating system: Download Linux Mint 20.1\nGraphics card: AMD Radeon RX 580\n\nBroken: 2.93 Alpha\nWorked: Has never worked - to my knowledge\n\n\nAfter Linking a collection from a Blender scene, anything from that scene, e.g. a material will not append correctly.\n\n\nStart Blender. Select all the objects and move them (for clarity later). Select the cube and change its material to a blue color and change the material name to 'Blue'. Save As: Blue.blend.\n\nOpen a new scene and from the menu select File/Link. Select the scene Blue.blend and navigate to its collections, select the Scene Collection. The collection is linked - the collection can be moved, scaled or animated, but not edited, for example the Blue material cannot be changed - all is well.\n\nNow from the menu select File/Append, and select the same Blue.blend scene, but this time navigate to materials and select the Blue material. The Blue material should have been appended - but all is not well. If you now select the cube (from the open scene, not the linked one) and try to put the Blue material on it, the settings for the material cannot be edited. If you try and append a material from any other Blender file, this does append correctly and can be edited. \n\nMy understanding is that an appended item should be 'in' the Blender scene and can be edited. But it appears that with an item from a scene that has already been linked something is preventing this from happening. \n", "It is not possible to append objects which are part of linked collection\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 472.12\n\nBroken: version: 2.93.4\n\nIt is not possible to append objects which are part of the collection that has been linked in the file. Only way to append the file is to remove the linked collection and the reference to the linked file.\n\n1. Create a new blend file A, create a collection inside it, and create a few mesh objects in the collection. \n2. Save the file A.\n3. Create a file B. Link a collection from the file A by using File->Link operator.\n4. Append one of the objects from the collection in the file A by using File->Append.\nResult: The object is not appended, but linked instead, and can not be edited.\nExpected: The append operator actually appends the object, creating a local copy of it inside of the file B.\n" ]

[ "It says when I try to put my render into the downloads file permission denied can somebody help?\nOperating system: ?\nGraphics card: ?\n\nBroken: ?\nWorked: ?\n\n?\n\n?", "particle system collisions stop working outside 255 blender units\nOperating system: windows 10\nGraphics card: gtx 1080\n\nBroken: 2.79b 2018-3-22, \nparticle systems do not interact with faces which are oriented in the axis where the face is more than 255 blender units away from the origin\n[name ]([particleproblem.blend](particleproblem.blend))", "Adding texture to particle emission density breaks particle simulation\nOperating system: Windows-10-10.0.17763 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 430.86\n\n\nBroken: version: 2.80 (sub 74)\nWorked: (optional)\n\n\nPlease see attached file, after adding cloud texture and enabling Density in Texture influences, the simulated particles either stick to the surface or fly of into infinity as a group. \n\nOpen attached file and play animation, note some particles either sticking or a large group of particles incorrectly flying of.[particle_system_broken.blend](particle_system_broken.blend)\n\n", "Compositor artifacts from image wrapping\nOperating system: Linux-5.15.13-arch1-1-x86_64-with-glibc2.33 64 Bits\nGraphics card: NVIDIA GeForce GTX 1060 6GB/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 495.46\n\nBroken: version: 3.1.0 Alpha\nBroken: 2.80 upwards\nWorked: 2.79\n\nIn the compositor, when using a Scale node with a value of < 1.0 after a Translate node with wrapping enabled, a strange pattern can be sometimes seen in the final render (sort of like Z-fighting). Also, re-rendering with the same settings doesn't always yield the same results with this setup.\n\nChanging the chunksize to anything other than 256 seems to fix this for me.\n\n\n\nI've created and attached a .blend file that has the compositor setup in the right way to display this bug. So all that's required is to render the scene - possibly refreshing the compositor a few times as the bug doesn't always occur.\n\n[blender-bug-report.blend](blender-bug-report.blend)", "Blender 3.0 Denoising on Transparency Creates Artifacting (Blend included)\nOperating system: Win 10 (latest update)\nGraphics card: RTX 3090\n\nBroken: (3.0.0 and 3.0.1 dc2d18018171 Branch: Master)\nWorked: (2.9.0)\n\nWhen enabling transparency under \"Film\" settings, artifacts occurs in the transparency channel\n\nOpen attached .blend in 2.9.0 and 3.0.0 and hit render to see. I've set-up a compositor node setup to highlight and visualize the artificating found in 3.0.0. Try rendering with both Denoising methods to see variations in artifacts.\n\n**Other notes**\nI've played with as many render settings as possible and found that the denoiser is responsible. OptiX is worst, OpenImage is not as bad in artifacting.\n\n[alpha-transparency-denoising-issue-sample.blend](alpha-transparency-denoising-issue-sample.blend)\n\nNotice the clean colors in the greys (background composited in Compositor to help see issue)\n\n\nNotice the vertical and horizontal line noise\n\n\n[system-info.txt](system-info.txt)", "Keyframing off Animated property of rigid body on second frame of rigid body simulation\nOS: Ubuntu 13.10\nLinux kernel: 3.11.0-18-generic\n\nBroken: 2.70-b64bdb2-linux-glibc211-x86_64\n\n\nNormally, keyframing off the Animated property of a rigid body will cause rigid body to maintain its velocity and rotation that it has at the last animated keyframe. However, if the Animated property is keyframed off on the second keyframe of the rigid body simulation, the object will continue the animation from a stand-still, i.e. it loses its initial velocity / rotation.\n\n\n\nCreate bug:\n\n - Open new Blender file\n - Keyframe location of the default cube\n - Go to frame 20, move cube along the y-axis by 20 units and keyframe location.\n - Enable Rigid Body for the cube\n - Go to frame 10, enable Animated (Physics context -> Rigid Body panel) and keyframe it\n - Go to frame 11, disable Animated and keyframe it\n - Go to Scene context -> Rigid Body Cache panel, and set Start to 10 (and say, End to 50)\n - Go to 3D View and hit ALT A to start animation. Notice that when the cube reaches frame 10, it loses its y-velocity and just falls straight down.\n\nResolution:\n\n - Set the start of Rigid Body Cache to 9. \n - Update cache by unlinking the RigidBodyWorld group from Rigid Body World panel, then choose it again.\n # Play animation again, and it's fine.\n\nAlthough, I have figured out how to get it working, I think this is unexpected behaviour (which took me a long time to figure out).\n\nSteps 1-7 have already been completed in this file: [rigidBody_bug.blend](rigidBody_bug.blend)\n", "Preview Texture Blend\nOperating system: Arch Linux\nGraphics card: GTX 1060 6GB\n\nBroken:\n(example: 2.79b release)\n(example: 2.80, edbf15d3c044, blender2.8, 2018-11-28, as found on the splash screen)\nWorked: (optional)\n\n\nTexture gradient is not displayed correctly. Texture Blend.\nWhen you configure the texture, the preview displays the texture view.\nThe preview is more white than black.\n\n\n\n\nIf the texture to use as brush, you can see that the boundary particle moves wrong.\n\nIf in 2.8 open the file created in 2.7.\nThe texture is displayed differently.\nBut it works the same way as in 2.7.\n\n\n\nRun the attached files, see how the particles behave.\n[Blend.zip](Blend.zip)\n", "3D Texturing: When changing the texture resolution, the pixel data isn't recreated\n* Start painting\n* Change image resolution\n* Continue painting.\n\nWe have a safe guard for this, but it doesn't seem to work anymore.\n", "Cycles: Keyframing motion blur shutter value fails to apply it properly when generating animation\nOperating system: Windows 10\nGraphics card: Nvidia RTX 3090\n\nBroken: 3.6, 4.0\n\n\nIf we keyframe shutter property of motion blur, the updated shutter value is not considered during the final render.\nSuppose shutter value is 0 at the start of render, it will remain 0 thorught the animation even if we've keyframed the value.\n\n- Open attached file\n- select cycles render Engine\n- Start animation render (Ctrl-F12)\n", "Cycles: Motion blur artifacts\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 472.88\n\nBroken: version: 3.2.2\n\n\n**description of error**\nWhen rendering an animated rigged model on cycle **with motion blur** ; if the model has **auto smooth enable AND a subdivision modifier**, then it is possible that strange artefact will appear during the render. (doesn't happen in the viewport) \n- This bug do not happen if auto smooth is disable or doesn't have a subdivision modifier. \n- The bug also only appear during some specfic poses. So this is mainly noticiable when rendering an animation.\n- Those artefact often appear near sharp edge.\n- Only happen on Cycle, Eevee render seem fine.\n- [Edit] changing the shutter values of the blur will sometime remove the bug on some frames OR make the bug appear on frame that was fine before.\n\nHere 2 Render exemple with and without autosmooth , both has motion blur enabled :\n \nAnd here an animation where you can notice the bug.\n[Animated0035-0318.mp4](Animated0035-0318.mp4)\n\n\n[MotionBlurCycleArtefact.blend](MotionBlurCycleArtefact.blend)\nIn this blender file, rendering the frame 302 will make the bug happen. \n- If you disable auto smooth on the model (the main body and not the cloth) then the render will be fine.\n- If you disable the subdivision modifier on the model then the render will be fine.\n- If you disable the subdivision modifier AND the auto smooth on the model then the render will be fine.\n\nThank\n", "Perfomance drop if Texture Limit Size enabled \nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce RTX 2060 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 461.40\n\nBroken: version: 2.92.0\n\n[2021-04-02 12-29-11.mp4](2021-04-02_12-29-11.mp4)\n\n1. Add 4k cheker map to default cube\n2. Go to material preview\n3. Go to UV edit, manipulate UVs\n4. Toggle Texture Size Limit \n\n[#87133.blend](T87133.blend)\n", "Regression: Image Editor Texture Draw Stops Updating\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1060 3GB/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 511.79\n\nBroken: version: 3.1.0 (probably introduced by {f663a1dc09}?)\nWorked: 3.0\n\nthis was actually exposed with 507a4deef1 (but we have seen before that that commit is not the real cause for a couple of issues -- @lichtwerk)\n\nImage on Image Editor -> Paint does not update when being drawn on at certain close view levels until the viewport is moved \n\nLoad default scene\nSwitch to Texture Paint tab\nCreate new image\nZoom in 7 or 8 times (for reference I am using a scroll wheel that has steps)\nDraw on the image\nIt will sometimes update fine, sometimes stop updating when it passes over a what looks like a UV boundry, that extends off the screen\n\nSee gif where I added a material to it\nSorry if this has already been reported, haven't been able to find a public bug tracker\n[image_paint_bug.blend](image_paint_bug.blend)\n\nThis is the file I (-- @lichtwerk) used on linux to bisect/reproduce on a 1920x1080 display (since this seems to depend on monitor/zoom etc):\n[image_paint_bug_repro.blend](image_paint_bug_repro.blend) ", "Texturing with anchored type stroke\nOperating system: Linux-5.4.0-47-generic-x86_64-with-debian-bullseye-sid 64 Bits\nGraphics card: GeForce RTX 2060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 450.51.06\n\nBroken: version: 2.91.0 Alpha\n\nTexture paint with stroke set to anchored while using texture significantly slows down blender up to non responsive state and crash while paint texture is larger than 2k.\n\n[Texture pain stroke ty[e anchored.blend](Texture_pain_stroke_ty_e_anchored.blend)\n", " LineArt: (Collection) Issue for Cameras in Orthographic View\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2080 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.94\n\nBroken: version: 3.3.1\nChecked in Blender version 3.4.0 Apha, commit date: 2022-10-08 02:26 - same issue\n\nFor orthographic Camera without any shifts - Line art modifier (for a Collection) showing lines for objects behind other objects - which otherwise should not be visible. And some lines are not visible which should be.\nHappens with contours option in the Line Art modifier. Slight rotation on z axis solves the issue.\n\nNoticed issue with some specific frames on my complete blend file.\n\nOpen attached blend file and go to front orthographic view.\n\n[BugReport01.blend](BugReport01.blend)\n\n", "No Vector Displacement On Instances\nWindows 10 x64\n\n2.79.4\n\nVector displacement doesnt work with instances, It mirrors the displacement on the first object, Im sure this is working as intended but there are many cases where you need to displace actual instances as well.\n\nHaving a checkbox somewhere to enable instance displacement would be amazing\n\n\n\n\n\n[instances.blend](instances.blend)\n" ]

[ "Persistent data, motion blur, and geometry displacement results in glitchy meshes during rendering\n**System Information:**\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 1070 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 461.92\n\n**Blender Version:**\nBroken: version: 2.93.1\nWorked: N/A\n\n**Short description of error:**\nI don't know if this is a bug or a limitation, when rendering with Cycles with `Persistent Data` and `Motion Blur` are enabled and there is an object with `Displacement`, the mesh will start to glitch. I tried also with a fixed camera and the result is the same.\n\n| `Persistent Data` off | `Persistent Data` on | \n| -- | -- |\n|[A0001-0050.mp4](A0001-0050.mp4)|[0001-0050.mp4](https://archive.blender.org/developer/F10217141/0001-0050.mp4)|\n\n[persistent data test.blend](persistent_data_test.blend)\n\n**Exact steps for others to reproduce the error:**\nopen the attached .blend file and render the animation\n\nor\n\n1. Create a new file.\n2. Change the render engine to Cycles and enable `Motion Blur` and the `Persistent Data` option.\n3. Create an object and give it a material, making sure you change the `Displacement` method for this material to `Displacement Only`\n4. In that material, attach something to the `Displacement` output of the node tree. A noise texture works.\n5. Ensure the object with the material created in the precious steps has enough geometry for the displacement to be noticeable.\n6. Render an animation with this setup and observe the mesh changing as the animation progresses, this should not be happened." ]

[ "No \"affect alpha\" in brush settings in the Image Editor\nOperating system: Linux-5.3.0-40-generic-x86_64-with-debian-buster-sid 64 Bits\nGraphics card: GeForce GTX 980 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 435.21\n\nBroken: version: 2.82 (sub 7)\n(as well as the latest alpha: version: 2.83 (sub 5))\nWorked: 2.80 (Checkbox existed in Image Editor brush settings, but had NO effect on brush behavior)\n\nIn the 3D Viewport as well as the Properties Editor while in Texture Paint mode, all brushes offer the \"affect alpha\" option in the \"Advanced\" section of the Brush Settings. In the Image Editor that option is missing from the \"Advanced\" section for no apparent reason.\n\n\nOpen Blender with factory settings. Switch to \"Texture Paint\" tab. Compare the \"Advanced\" section of the active brush's tool settings in the Image Editor to that in the 3D Viewport or Properties Editor.\n\nSurely this has to be a well-known issue and exists only due to technical limitations of some kind.\nWhat should I tell my students why this inconvenience exists and when it will be addressed?", "Missing red border for render border in Image Editor\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 1060 6GB/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 460.89\n\nBroken: version: 2.93.0 Alpha\n\nIn the 3D view you have a red border around an active render border. In the Image editor not.\n\nThis can lead to hours of lost render times since you cannot see if a render border is active or not. And is inconsistent. Two equal functionalities should work equal.\n\nOpen Blender, activate Render border in the 3D view and render a portion with viewport shading rendered.\nNow render the image, and render a portion of the image with enabled render border. You will notice that in the 3D view you have a red rectangle that indicates that the render border is active and what area to render. \n\nBut in the image editor you just see that just a portion is rendered. The next render can be with or without render border. There is no visual hint if the tool is active or not. And no hint if you have selected the correct area for preview rendering.\n\n\n\n\n\n\n", "Some particles are not being rendered during the first render\nOperating system: Windows 10\nGraphics card: I tried on two different computers with different GPUs - Nvidia RTX3070 and Mobile Nvidia RTX2060, same behaviour.\n\nBroken: 2.93.6, 3.0.0 Release Candidate, 3.1.0 Alpha\n\n\nBlender is not rendering some particles (in my case this is a grass) during the first render.\nIt does after re-rendering the scene, but this is quite confusing and inconvenient, especially in the moments when I have to resume render of animation.\nApparently this only happens when we add a force field to control the particles.\n\nOn top of this, 3x does not render trees (particles).\n\nOpen the blend file and press F12 to render. You should have similar result as on the screenshot:\n\n\n\nRender again, and the grass will appear:\n\n\n\nI attach results also from 3.0:\n\n\n\n\n\nand 3.1:\n\n\n\n\n\n- Open attached file\n- Render (F12) - See the particles issue\n- Render again - Now it is ok\n\n[turbulence_particle_eevee_bug.blend](turbulence_particle_eevee_bug.blend)\n\n---\nOriginal File:\nBecause file is big (260M) I have decided to share the link: view?usp=sharing <- if you would prefer me to upload it anyway here, please just let me know.", "Cycles viewport unneccessary update calls\nOperating system: Ubuntu 20.04\nGraphics card: Nvidia MX150\n\nBroken: 2.83.17, 8a8c61190b06, Beta and 2.90.2, 8f21f867e715, Alpha\nWorked: 2.82a official\n\n[#76970.blend](T76970.blend)\n\n* set scene renderer to Cycles\n* configure two viewports and a shader editor in a single screen\n* set one viewport to cycles preview render\n* set second viewport to workbench texture color\n* add two image texture nodes and load them with different textures\n* by clicking on the texture node both viewports are redrawn.", "Texture Paint - Clone tool not working when clone source area out of viewport\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 1070 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 457.30\n\nBroken: version: 2.91.0\n\nTexture Paint - Clone tool not working when clone source area out of viewport\n\nImport image as plane- select part of image to clone in bottom part of image. Scroll up to the top part of image so the bottom is not visible = clone tool doesnt draw anything.\n\n\n\n", "Viewport render gives black background when overlays are turned off.\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce RTX 2070 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 457.51\n\nBroken: version: \n - 2.92.0 Beta\n - 2.92.0 Beta \nWorked: \n - 2.92.0 Beta\n - 2.91\n\nviewport render gives black background, this happens only when the overlays are turned off.\n\nopen the file, go to view/viewport render image.\n\nthank you\nMax\n\n\n[black_bg.blend](black_bg.blend)\n\n\n", "Image of the viewport can no longer be obtained via glReadPixels in Python. Gives strange results since 2.83\nOperating system: Arch Linux 5.8.5-arch1-1\nGraphics card: Intel i7-8665U\n\nBroken: since 2.83\nWorked: till 2.82\n\nIn Python, we get an image of the viewport from within an registered draw handler routine using:\n\nviewport = bgl.Buffer(bgl.GL_INT, 4)\nbgl.glGetIntegerv(bgl.GL_VIEWPORT, viewport)\nwidth = int(viewport[2])\nheight = int(viewport[3])\nimage = np.zeros([height, width, 4],dtype=np.uint8)\nbuffer = bgl.Buffer(bgl.GL_BYTE, image.shape, image)\nbgl.glReadPixels(0, 0, width, height, bgl.GL_RGBA, bgl.GL_UNSIGNED_BYTE, buffer)\n\nIt worked fine till 2.82, as shown here\n\n\nHowever, it gives blackish colors of the objects in the viewport since 2.83:\n", "Asset Browser doesn't show assets when drop-down set to \"All\"\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.94\n\nBroken: version: 3.6.0 Alpha\nAlso broken in 3.5 stable version!\nWorked: This feature is new as of 3.5 and I'm seeing the behavior in both. \n\nAssets from other files are not visible when asset browser preference is set to \"All\" (when assets are part of catalogs)\n\n- Extract uploaded file\n - Open general blend file\n- Set it as asset library path\n- Open asset browser and set preference to \"All' [notice every asset is visible (i.e. engine parts and sculpted sphere)\n- Now open `open-file.blend`\n- In asset browser, set preference to \"All\" (notice engine assets are missing)\nIf switched to \"user library\" then engine assets are visible\n", "Missing progress indicator for 'Render Audio'\nOperating system: Darwin-19.6.0-x86_64-i386-64bit 64 Bits\nGraphics card: AMD Radeon Pro 570 OpenGL Engine ATI Technologies Inc. 4.1 ATI-3.10.16\n\nBroken: version: 2.83.4\n\nWhen we use the 'Render Audio' option from Render menu, after the rendering starts, there is no indicator about the rendering progress, how much time is left.\n\nIn any scene with audio in the Sequence editor, try using Render>Render audio command. There is no progress indicator. It is very vital specially for long length tracks.\n", "Missing viewport update after creating texture with canvas menu (sculpt mode, paint tool)\nOperating system: Linux-5.13.0-0.rc6.45.fc35.x86_64-x86_64-with-glibc2.34.9000 64 Bits\nGraphics card: NVIDIA GeForce GTX 970M/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 495.44\n\nBroken: version: 3.3.0 Alpha\nWorked: -\n\nMissing viewport update after creating texture with canvas menu (sculpt mode, paint tool)\n\n- set viewport shading to `Material Preview`\n- Select cube\n- go to sculpt mode\n- select paint tool\n- open canvas menu\n\n- create new material slot texture with a certain color\n- result: \"Compiling Shaders 1\" in the viewport statistics top left, but this never finishes, update (incl. seeing the actual texture on the cube) only takes place after interaction with the viewport (rotating, zooming, doing an actual paint stroke)\n\n", "Viewport animation render crashes when collapsing the viewport.\nBroken: version: 2.91.0 Beta (and 2.83, 2.80, ...)\nWorked: never\n\nViewport animation render crashes when collapsing the viewport.\n\n[0001-0485.mp4](0001-0485.mp4)\n[Viewport render crash.blend](Viewport_render_crash.blend)\n\n", "Cycles viewport render doesn't resize if an animation is playing.\nOperating system: macOS-13.0-arm64-arm-64bit 64 Bits\nGraphics card: Apple `M1` Pro Apple 4.1 Metal - 83\n\nBroken version: 3.5.0 Alpha\nBroken version: 3.3.1\n\nIf a .blend file contains a animation, and the animation is playing, and you're in the Cycles rendered viewport mode, then the viewport resolution/size will not update as you adjust it's size.\n\n1. Create a scene that has a colour that differs from the standard background colour of the 3D viewport.\n2. Animate something in the scene (E.G. Animate the movement of an object)\n3. Change the render engine to Cycles (This issue only appears in Cycles for me.)\n4. Enter the rendered viewport mode so you're now seeing a Cycles viewport.\n5. Start playing the animation.\n6. While the animation is playing, start adjusting the size of the 3D viewport editor and notice how the Cycles rendered view isn't updating with the change in size. This is the issue. Once you stop playing the animation, it will update.\n\nHere is a video demonstrating the issue:\n\n[Cycles preview doesn't resize.mp4](Cycles_preview_doesn_t_resize.mp4)\n", "Render problems\nOperating system: Windows 10 Pro Ameliorated\nGraphics card: RTX 3070 driver version 472.12\n\nBroken: v3.6 and v3.61\nWorked: v3.51 <- Using the same file created in v3.61\n\nWhile rendering a frames of an animated character, the current rendered animation frame usually shows in the 3D viewport in v3.51. However, in the broken version, it fails to show the current frame being rendered, nothing gets updated visually. Then eventually in the 3D viewport the character goes into its T pose for a few frames then the character starts losing visibility of its layers, at that point Blender crashes. \nI will create a .blend file if I am the only one getting this bug.\n\n", "GPencil: Texture Image Alpha pass through grease pencil strokes\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 441.87\n\nBroken: version: 2.83.1\n\nContext: Using texture image to make some kind of 2d fur on top of a GPencil Object\n\nThe Image Texture used to make some fur is mostly transparent and the brush used to make the colored part is a texture paint brush with Cloud Mask (it was made entirely in Blender with Blender tools).\n\nWhen on top of the GPencil object, the part of the Image Texture that is only half transparent(between 99%-1% transparent), makes the GPencil color disappear, even when moving the center point behind the one of the fur images.\n\nOn the contrary, when on top of another Image Texture (imported to be used as camera mapping), the Image Texture of the fur doesn't impact the transparency of that one.\nNotice that the Image texture boundary, even from the imported one, has problems with clipping and always makes a white line appear.\n\n- Open attached .blend file\n- Zoom in and out to notice the alpha transparency influencing the GP object\n\nThanks for your help!\n\n\n\n\n\n\n\n[bug_alpha.blend](bug_alpha.blend)", "Viewport Render (and Render Animation) color looks completely off from the Viewport\nOperating system: Darwin-19.0.0-x86_64-i386-64bit 64 Bits\nGraphics card: AMD Radeon Pro 560X OpenGL Engine ATI Technologies Inc. 4.1 ATI-3.2.24\n\nBroken: version: 2.82 (sub 6)\n\nRendering the viewport animation gives a very different result from the actual viewport.\n(Also: Viewport Anti-Aliasing isn't taken into consideration on the Viewport Render).\n\n||\n| -- | -- |\n\nThe problem is more noticeable when the Viewport Shading is set to \"Flat\", even though it also shows in \"Studio\". It delivers very dull colors with less contrast and a blue-ish tone.\nChanging between Cycles and Eevee doesn't affect the Viewport Render (from my tests).\n\n||\n| -- | -- |\nflat:\n||\n| -- | -- |\n\n- Open attached file.\n- Go to View -> Viewport Render Image\n[Viewport Colors.blend](Viewport_Colors.blend)" ]

[ "EEVEE: Viewport Render shows transparent image\nOperating system: GENTOO Linux/rolling release/last update: this morning\nGraphics card:NVIDIA Corporation TU106 [GeForce RTX 2060 SUPER] (rev a1)\nNVIDIA drivers version: 510.47.03\nOptix and experimental features are activated.\n\nBroken: blender-3.2.0-alpha+master.35c261dfcf61-linux.x86_64-release,blender-3.2.0-alpha+master.467c16eab1ec-linux.x86_64-release\nWorked:blender-3.2.0-alpha+master.f021d467526a-linux.x86_64-release\n\nCaused by 507a4deef1\n\n**EEVEE: Viewport Render Animation defunc**\nViewport Render Animation with EEVEE displays nothing but a grid while rendering. The resulting video is ok, though.\n\n- Open default scene\n- {nav View > Viewport Render Image/ Animation}\n" ]

[ "Geometry Nodes related memory leak and exception access violation\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 4070 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 535.98\n\nBroken: version: 3.5.1\nWorked: \n\nBlender freezes, memory usage spikes and eventually you get exception access violation crash.\n\nOpen attached file and enable Layer 3. Blender will freeze, memory usage will spike and eventually lead to exception access violation crash. Crash log is generated.\n\n[gnodes bug1.blend](attachment)\n\nTo stop crashing disconnect Twig Object from Object Info node. I think this started happening after I deleted the \"fir_twig\" object, but there was still a hidden reference in the Object info. However, I was not able to replicate this in a new file and the crash stops happening if I delete / modify some of the node groups, even in unrelated ways. I am pushing Blender to the limits and this is a heavily reduced file.\n\n", "Selecting (or connecting) nodes in Compositor is malfunctioning when when monitors have different dpi\nOperating system: Windows-10-10.0.16299-SP0 64 Bits\nGraphics card: GeForce GTX 1080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 431.86\n\nBroken: version: 2.83 (sub 5)\n\nSelecting nodes in Compositor is glitching when when monitors have different dpi. This happens when Compositor is opened on both monitors.\n\n - In Windows 10 set custom scale on the second monitor in Display Settings > Scale and Layout, for example to 175%, so that each monitor has different scale.\n - Put one Blender window on each monitor (same Blender session).\n - Open Compositor in both windows. Enable nodes.\n - In the window with scale equal to 100% try clicking on the Composite node couple times. Selection will switch to the other node.\n[custom_dpi_compositor_bug.blend](custom_dpi_compositor_bug.blend)\n[custom_dpi_compositor_bug_02.03.2020.mp4](custom_dpi_compositor_bug_02.03.2020.mp4)", "Intersect Boolean does not work properly when mesh has a degenerate face\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: Radeon (TM) RX 480 Graphics ATI Technologies Inc. 4.5.14760 Core Profile Context 20.45.37.01 27.20.14537.1001\n\nBroken: version: 2.91 - 3.0.0\nWorked: n/a\n\nThe `Intersect Boolean` Exact mode does not work correctly when the mesh has at least one degenerate face.\nIn this case, it was tested on a mesh with a triangle whose vertices are collinear\n\n- Open attached file - (The edited mesh has a degenerate triangle)\n- Go to `Face` -> `Intersect (Boolean)` - (The operation will not work because of that triangle)\n\n- Undo the operation\n- Move the triangle vertex a little so that the face is no longer degenerate.\n- Reselect that loose geometry ({key L})\n- Redo Boolean operation (now it works)\n[intersect_w_degenerated_tri.blend](intersect_w_degenerated_tri.blend)\n\n", "GN: Delete Geometry node hides edges in some cases\nOperating system: Windows-10-10.0.19043-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1060 6GB/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 511.23\n\nBroken: version: 3.2.0 Alpha\n\nHard to describe this one more abstract..\n\nI start with a mesh line primitive (0,0,0) (0,1,0) and extrude it **n** times along **x** to get a grid.\nNow I try to delete all points with a distance from the **y** axis greater than **d** \nWhenever **d** lies between the first and the last extrusion, the edges of the starting primitive are not displayed.\nIt shows up in the spreadsheet however and reappears if I apply the modifier and tab into edit mode and back \n\n[2022-02-05 19-27-50.mp4](2022-02-05_19-27-50.mp4)\n[del_geo_node.blend](del_geo_node.blend)\n\n", "Graph Editor - Multiple Windows Problem\nOperating system: Windows 10 - 64\nGraphics card: Radeon RX 5500 XT\n\nBroken: 2.93.0\n\nI split my single Blender window to two Graph Editor/Node Editor & 3D Viewport for my multiple screens. If my mouse hovering anything but node editor(Include Graph editor or 3D Viewport) and I press play that timeline cursor like starts twitching.\n\nCreate two different windows. Graph Editor and 3D Viewport. Than press play on 3D Viewport window selected.\n\nTest file:\n[Bugfile.blend](Bugfile.blend)\n\nVideo:\n[2021-06-10 15-35-41.mp4](2021-06-10_15-35-41.mp4)", "Trackpad stops working for a while\nOperating system: MacOS 12.6.5\nGraphics card: AMD Radeon Pro 5500M\n\nBroken: 3.6.2\nWorked: don't know\n\n\nSometimes, Blender stops responding to trackpad gestures such as two-finger scroll or pinch zoom. This is not just in the viewport but everywhere - I can't scroll menus or the node view, for example.\n\nThe UI is otherwise responsive when this happens - I can scroll menus by dragging the slider at the edge, just not with a two-finger scroll.\n\nIf I go and do something else in another app for a few minutes, it usually fixes itself.\n\nThis happens sporadically and I don't have a reliable way to reproduce it, but it seems to happen often after I render an image - after the image has rendered, I try to use pinch-zoom to look more closely and find I can't. I am not sure if it happens in other situations as well - if it does I will update this report.\n\n\nI'm afraid that I can't reproduce this reliably - it just happens every now and then, in any .blend file. I am happy to take any advice on how to look for a way to reproduce it.\n\n", "Blender shuts down when playing a Rigid Body animation in a linked scene \nOperating system: Windows-10-10.0.17763 64 Bits\nGraphics card: AMD Radeon R7 200 Series ATI Technologies Inc. 4.5.13492 Core Profile Context 22.19.677.257\n\n\nBroken: version: 2.80 (sub 74)\nWorked: (optional)\n\n\nBlender shuts down when playin the animation of a rigid body object in a linked scene\n\nLoad Factory Settings\nAdd the cube as Rigid Body\nCreate a new scene, Linked Copy\nPlay the animation in any of the two scenes\nWhen switching to the other scene, Blender shuts down\n\n", "Tangent node used as a vector on Texture Nodes yields nothing through a Bump Node\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 680/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 472.12\n\nBroken: version: 3.0.0 Alpha\n\nIf you plug a tangent node into any procedural noise texture as the vector, it will work, however once you plug the result into a vector bump node it will draw as an empty bump map. \n\n1. In the shading workspace, grab a noise texture\n2. Grab the tangent node and plug it into the noise textures vector input. Any tangent mode will do fine (x,y,z)\n3. Grab a vector bump node and plug the noise texture into height. \n4. View the resulting bump, nothing will be yielded (however if plugged into the normal input it will)\n[Example.blend](Example.blend)\n", "Space Mouse not working properly\nOperating system: Mac OS Ventura 13.3.1 (a)\nGraphics card: Apple M1 Max\n\nBroken: 3.5.0\nWorked: 3.0.0\n\nNo buttons of the Spacemouse work anymore.\nEverything worked with Blender 3.0.0 and 3D Connexion Driver 10.7.0.\n\n1. Install latest 3D Connexion Driver 10.8.0.\n2. Set up some buttons in 3DxWare 10 like:\n a. \"Button 1\" = Keystroke \"Tab\".\n b. \"Button 2\" = Keystroke \"1\".\n c. \"Button 3\" = Keystroke \"2\".\n3. Start Blender.\n4. Try to switch between Edit and Object mode with \"Button 1\":\n a. No Function.\n5. Try to switch between select Modes with \"Buttons 2-3\":\n a. No Function.\n6. Try to select more Objects with holding \"shift\" on Space Mouse:\n a. No Function.\n\nI miss the combination so hard, it's impossible to work without my space mouse, I'm used to it for a decade. :(\nThe Space Mouse works fine in all other Programms, its an Blender Issue exclusively.\n\nAdditional Infos:\nSometimes restarting Blender multiple Times, re-installing it, fixes temporary the function of the basic Keys \"shift, cmd, etc\". The other buttons never work. They worked always in the past and still work in all other programs where I use the mouse. ", "Bevel modifier disconnects vertices in the UV map\n{[F8713789](2KmuzHs.png)}Operating system: Windows 10 x64\nGraphics card: GTX 970\n\nBroken: 2.83.1, 47b6c332587ca45eddc98e75b3f563e2f17a0d47, probably all 2.8+ versions\n[SM_UVTest.blend](SM_UVTest.blend)\n\nWhen using an angle or weight limit on the Bevel modifier, vertices connecting beveled and unbeveled edges are split in the UV map, producing distorsions.\n\nAdded a blend file. To reproduce the issue, apply the Bevel modifier, and select all faces to see the problematic vertex in the UV editor. \n\nSome context is here : 1241831", "Grease pencil edit mode proportional edit \"connected\" bug\nOperating system:\nGraphics card:\n\nBroken: (example: 2.80, edbf15d3c044, master, 2018-11-28, as found on the splash screen)\n\nProportional edited vertices from a mesh behave different when editing in \"only connected\" than GP points do. Mesh points (vertices) calculate the influence by the length of the path, while GP strokes react as soon as they get into the influence circle.\nTheleft rectangle in the picture is a mesh, the rectangle on the right is a grease pencil object. Notice how the points react differently.\n\n\n\n[ProportE.bug.blend](ProportE.bug.blend)", "Scaled Rigid Body causes issue (moving origin or mesh breaks collision, double transforms on negative scale)\nOperating system: Ubuntu 18.04 LTS\nGraphics card: Intel HD 4000\n\nBroken: 3.0\nBroken: 2.90.1\nWorked: I don't know if this works with any other version, I created the file using 2.90.1.\n\n\nSee the attached .blend file. This file contains two errors with Rigid body simulation. I'm filing both errors in one bug because these may be related (both happen in one example simulation).\n\nError 1: object origin distance from another object changes interaction between objects\n\n1. Open attached example case (.blend file)\n2. Play the animation to verify that \"Ball\" is attached to the chain.\n3. Stop the animation and jump to first frame.\n4. Select object \"Ball\" and from menu Object - set Origin - Origin to Center of Mass (Volume)\n5. Play the animation to verify that \"Ball\" is no longer attached to the chain.\n\nError 2: Rigid body type \"Active\" changes orientation of the object\n\n1. Open attached example case (.blend file)\n2. Play the animation to verify that chain is attached to the object \"Cube\".\n3. Stop the animation and jump to first frame.\n4. Change \"Cube\" Physics properties - Rigid body - Type: Active and Settings: Dynamic\n5. Play the animation and notice that Cube is now rotated 180 degress around the X axis. Obviously this results in chain no longer connected to the cube as expected and chain and Ball drop directly down during the animation.\n\nExample .blend file (cannot simplify more because Blender seems to be unstable with this file):\n[physics-test.blend](physics-test.blend)\n\n**Additional information**\nIt seems that I can work around Error 1 by selecting all items and applying \"Rotation & Scale\". I think this should not be required to get two objects to interact during simulation so I'm reporting the problematic file as-is. I cannot figure out the reason nor workaround for Error 2.\n\n$ apt policy blender\nblender:\n```\nInstalled: 2.90.1--release~98526-3e85bb34d0d-0thomas~bionic0\nCandidate: 2.90.1--release~98526-3e85bb34d0d-0thomas~bionic0\nVersion table:\n```\n *** 2.90.1--release~98526-3e85bb34d0d-0thomas~bionic0 500\n 500 ubuntu bionic/main amd64 Packages\n", "Nodes: Changing Input position with a Driver\nBroken: 3.1 and 3.2 Alpha\n\nIn Geometry Nodes editor, shift an input socket above a socket having a Driver and it will switch the Driver.\n[2022-03-12 15-58-02.mp4](2022-03-12_15-58-02.mp4)\n\n- Enter Group Node\n- Shift \"Vertices Y\" input above\n[Test.blend](Test.blend)", "Some Cycles nodes are active when muted (RGB To BW, Math, vector Math, Separate RGB)\nWin10 64bit\n\nBroken: 2.79\n\nRGB To BW node is active when muted\n\nOpen file and render\n[RGB_BW.blend](RGB_BW.blend)\n\n", "EEVEE Volume draw distance broken with emission shader\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce RTX 2070 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 432.00\n\nBroken: version: 2.83.4\nWorked: none\n\nWhen using an Emission shader as Volume shader in EEVEE, draw distances smaller than 100 produce unecpected behaviour.\nUsing Principled Volume works fine as long as Density is not 0, if it is 0 the behavior breaks in the same way.\n\n- Open attached blend (Cube with emission shader as Volume)\n- Change Volumetrics End distance to 70m\n- Zoom in and out in EEVEE viewport, the cube does disappear at a distance range inbetween the set clipping range\n[eevee_volume_drawdistance_bug.mp4](eevee_volume_drawdistance_bug.mp4)\n[eevee_volume_drawdistance_bug.blend](eevee_volume_drawdistance_bug.blend)\n\n" ]

[ "Noodles connected to vector math node may become invalid but remain connected after mode change\nOperating system: Windows 10 20H2\nGraphics card: NVIDIA RTX 2070 Super\n(Other systems have exhibited identical behavior.)\n\nBroken: 2.91.0\n\nWhen the vector output of a vector math node set to cross product (or any other mode offering a vector output type) is connected to another node's input, and the mode of the vector math node is changed to one such as dot product offering only a scalar output (or vice versa), the connection between nodes is not completely severed, though the noodle seems to disappear. The invalid connection can cause visual artifacts in the render. If the user clicks on the input of the node that the vector math node was plugged into, and drags, the noodle once again appears, and it is possible to sever the connection or plug it into another input.\n\nThe issue can be replicated simply as follows:\nStart Blender, open the shader editor (change to world shader or create a material to edit), add a vector math node and plug the default vector-type output into any input (such as \"surface\" of the world output). Now change the vector math node to a mode with scalar output type, such as dot product. The noodle will disappear. Now click on the input dot where the noodle had ended, and drag. The noodle will reappear.\n\nAnother interesting observation is that if the vector math node is moved after its mode is changed, and the invalidated noodle is made to reappear (by dragging its endpoint), its starting point remains at the original location of the vector math node.\n\nIn addition, the receiving socket remains \"occupied\" (in that it will not display the input field as if it was unconnected).\n\nThe included .blend contains a vector math node that has both invisible inputs and outputs. To see the noodles reappear, switch the vector math node to \"Scale\" mode.\n\n[phantom_noodles.blend](phantom_noodles.blend)", "Logics of nodes breaks under certain circumstances\nOperating system: Windows 10\nGraphics card: GTX 1060 6GB\n\nBroken: 2.83LTS - 3.1.0 alpha (tested only in those versions, so it doesn't means that it doesn't appears in version before 2.83)\n\nNodes become unpredictable, and logic brokes once user does some unusual actions.\n\n1. Open shader editor, add Vector Math node.\n2. Connect output of Vector Math node to any other input, for example to the Math node input. (Mode of Vector Math node by default is \"Add\", and \"Add\" is returns a vector)\n3. Change mode of Vector Math node to Length, Distance, or Dot Product. (Those 3 modes return Value Instead of Vector)\n4. Now you see what i mean.\n\n\n**My thoughts about whats happening here**\nAs you might notice, this happens when existing connection changes its return type from one type to another, example on above shows \"Vector To Value\" change, but same will happen when you'll try \"Value To Vector\".\nSo my hypothesis is that blender doesn't expects that connections between nodes might change from one type to another. So if this hypothesis is correct, then it means that it might happen not only with Vector Math node, but any other node that can have its output type to be changed.\n\n**What i mean by \"logics of nodes brokes\"?**\n1. Visual connection between nodes disappear, but they still connected in some sense, for example Math node still shows that something is connected to one of its inputs, so user can't change value.\n1.2. You might try to use Ctrl + RMB shortcut to cut connection between nodes, but blender can't register any connection.\n1.3. You can manually detach output of Vector Math node from the input of Math node, and you will success, however, if before doing so, you will move Vector Math node to other \nplace, and then perform note 1.3, you can see that blender didn't updated position of Vector Math node.\n\n(Since screenshot doesn't capture cursor, to help you undestand what happens on the screen, there is cursor drawn by hand)\n2. Computation of nodes brokes. Example:\nSimple node setup, 2 node groups that use same 3 nodes, difference between those groups is that group on top builded with Vector Math node at first setted to \"Add\" mode, then when building node group setup is done, Vector Math node mode changed to \"Distance\", causing bug to appear.\nNode group on bottom was initially builded with Vector Math node mode setted to \"Distance\".\n**Output of top node group**\n\n**Output of bottom node group**\n\nAt first i was expecting both node groups to perform exactly the same, but as you can see it doesn't.\n\n**Interesting finding**\nI noticed if i disconnect Vector Math node from Math node in bottom node group, and manually set value to 0 in Math node, it seems to perform same as broken top node group.\n\n**Last thing that you need to know and try:**\n1. Open shader editor, add Vector Math and Math nodes.\n2. Connect output of Vector Math node to first input of Math node.\n3. Change mode of Vector Math node to Length, Distance, or Dot Product.\n4. Connect output of Vector Math node to second input of Math node.\n5. Using Ctrl + MMB shortcut scroll through different modes of Vector Math node.\nHard to explain result with words, just try and see it." ]

[ "Add visual warning for links between incompatible node sockets\n# Add visual warning for links between incompatible node sockets\n\n## Motivation\nCurrently, if a node link is between two sockets that are not compatible with each other, the link has no effect without making the user aware of it.\n\n## Proposal\nDraw node links between incompatible sockets red. Additionally, a warning could appear after connecting the node.\n\n\n\n----\nBasic implementation for the red colored node links: [P244](P244.txt)", "Annotations shouldn't be shared among different Geometry Nodes\nOperating system: ubuntu 20.04\nGraphics card: cpu\n\nBroken: 2.93.0\n\nWhen you ad an annotation to a geometry node, by default the same annotations can be seen in all the other geometry nodes layout.\nThis is unexpected, since each geometry node layout should have its own annotations.\n\nAdd a geometry node modifier.\nWrite something with the annotate tool.\nDuplicate that geometry node modifier, erase the old nodes, the annotations, and you'll find out that you erased the annotation also from the first geometry node layout.\n", "Eevee: Image Texture Node: Switching the source from \"Generated\" to \"Single Image\" and back to \"Generated\" leaves the shader result as pink color\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce RTX 2070 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.38\n\nBroken: version: 2.91.0 Alpha\nWorked: Unknown\n\nSwitching the source of an Image Texture shader node from \"Generated\" to \"Single Image\" and back to \"Generated\" leaves the shader result as pink color in Eevee. If it is working correctly, it should display once more the original texture as seen initially with the \"Generated\" setting. The problem is present only when Eevee is used as renderer. With Cycles, the texture is switched correctly to undefined (pink color) when switching to \"Single Image\" and back to the original one if changed back to \"Generated\".\n\n\nLoad the attached .blend file, change the image source on the material texture node to \"Single Image\" and then back to \"Generated\" while Eevee is used as renderer. \n\nFor comparison: Change to Rendered Viewport and Cycles as renderer. Repeat the changing of the image source to observe the correct behavior.\n\n[generated-image.blend](generated-image.blend)\n\n\n\n", "Geometry Nodes: Crashes when the curve resampling node length is 0\nOperating system: macOS-13.4.1-arm64-arm-64bit 64 Bits\nGraphics card: Metal API Apple M2 Pro 1.2\n\nBroken: version: 3.6.0\n\n\n\n\nThis is an irregular operation, but I hope it is not a crash but a warning.\nI don't know if this is a bug or a feature request. \nThere is also a possibility that if the length is close to zero, it will cause the computer to have 100% CPU.\n\n", "Output UVs from mesh primitive nodes with anonymous attributes\nInstead of creating a named `uv_map` attribute on the geometry, they should be anonymous attribute field outputs. This brings up a few complications.\n- Currently we have no 2D vector socket. We probably want to avoid adding this for 3.0, so a 3D vector can be used for now. Automatic conversions will make versioning simple when we add a 2D vector socket, hopefully for 3.1.\n- The BMesh mesh primitive icosphere creates an `MLoopUV` attribute, this should be converted into a 3D vector anonymous attribute.\n\nThe output should be called \"UV Map\"", "Camera data node \"View Vector\" differs between Cycles and Eevee in the viewport when used as world output\nOperating system: Windows 10\nGraphics card: RX580\n\nBroken: 3.0.0\n\nThe camera data node's \"View Vector\" output does not work under Cycles in the viewport when used in World shaders.\nNote that it _does_ work as expected if there's an active camera (although it's not quite as bright as Eevee).\n\nEevee viewport rendering:\n\n\nCycles viewport rendering:\n\n\nCycles viewport (with active camera) rendering:\n\n\nConnect the camera data \"View Vector\" output directly to the \"Surface\" input of the \"World Output\" node.\nSwitch to rendered view.\nObserve that the view vector is incorrect when navigating the viewport in Cycles, but correct when in \"camera view\" (numpad 0) or in Eevee.", "Viewport Z-Bias & retopology overlay related issues\nSince this is being reported frequently and retopology overly has caused some weird issue with mask display as well previously, it's best we keep track of them in a central place so we could check if any particular fix is working for all situations:\n\n------\n\nProblem still presents:\n\n- Retopology overlay affecting other parts of the drawing/picking operation: #109115 #107768 ~~#109277~~ ~~#107411~~\n- (Potentially related) Selecting when X-Ray turned on is not visually effective and then switching to face mode crashes: #110136\n\n------\n\nFixed:\n\n- ~~Wireframe visible in close distances: #110048 #109040 #109514 #109499 #109685~~\n- ~~Retopology overlay 0 z offset problem: #109662~~\n\n-------\n\nProbable cause:\n\nAs suggested in 107768#issuecomment-937545 , I believe the problem here is that\n\n1. `DRW_object_is_in_edit_mode` does not exclusively check if object is in edit mode, but it combines other logics that affects drawing and stuff.\n2. `XRAY_ENABLED` marco is not always accompanied with `RETOPOLOGY_ENABLED`, this leads to inconsistencies in the user interaction and the display.", "\"Z\" changed to \"depth\" breaks consistency in compositing and in Render layers\n\nwin7 gtx 550ti\nOperating system and graphics card\n\nBroken by: 4cf7fc3b3a\n\n\"Z\" changed to \"depth\" breaks consistency in compositing and in Render layers\n\n\n- Make sure that the **z**-pass is checked in the Render Layers tab\n- Open node editor, notice that the**z**-pass is now **depth**-pass\n- Notice as well that everywhere else it is still consider as**z**-pass even in the manual z_combine.html\n\n\n\nAlthough **depth** is more meaningful name, it is breaking consistency. Suggestion: we bring back the old behavior or we change everywhere to z-depth or depth.\n", "Changing outputs from Separate RGB node into Displacement (using microdisplacement) does not trigger geometry update.\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 446.14\n\nBroken: version: 2.83.0\n\nChanging outputs from Separate RGB node into Displacement (using microdisplacement) does not trigger geometry update. Have to tab into edit mode to force mesh update.\n[rgb_disp_bug.blend](rgb_disp_bug.blend)\n\nAdd plane with microdisplacement enabled.\nConnect Image node to Separate RGB node and one of the channels into Displacement node.\nChange view to Rendered mode.\nChange channels from R to G to B to R and notice no geometry updates happen in the view port. (Make sure to use an image with different channels)", "Linked group breaks Object Mapping in Texture Coordinates\nWindows 10\n\nBroken: 2.79b official release\n\nLinked group breaks Object Mapping in Texture Coordinates, when:\n- a texture (on a plane) uses another object (empty) as texture coordinates\n- all objects are in the same group\n\nFurther explanation in the blend file.\n\n1) Link the group \"object_mapping\" from the blend file\n2) move the group in any direction\nDespite the gradient texture on the plane using the empty's coordinates, this is completely ignored once linked as a group\n[object_mapping.blend](object_mapping.blend)", "Geometry Nodes: Only re-evaluated log do update available attributes in drop-down menu\nBroken: version: 4.0.0 Alpha\n\nThis is a fairly insignificant UI bug. If the GeoNodes modifier is disabled, the list of available attributes to select in one of the inputs is not updated. For example, when a vertex group is renamed, that won't be reflected in the list, until the modifier is enabled again. Can result in a bit of confusion.\n\n1. Open attached file.\n2. Notice the GeoNode modifier is disabled.\n3. Rename the vertex group.\n4. Try assigning the renamed vertex group as the modifier's Factor input; You can't, because it won't appear in the list.\n5. Enabling the modifier updates the list again, and then all is well.\n\n", "object display as wire , invisible when overlay is turned off \nwin 10\nintel i5\namd radeon \n\nblender 2.8 \n6b39dc7672eb\n\n\nNot sure if it's abug or intended behaviour \n\nWhen an object is set to wireframe ( object properties , viewport display ) , when overlays is turned off , the object is not shown ", "Node Group Texture slot problem\nOperating system: Windows 10\nGraphics card: gtx1060\n\nBroken: 2.9\nWorked: N/A\n\nIn the Texture Paint mode, when all textures are inside a node group, if you change brush radius it jumps to the top texture slot every time.\n\n- Open attached file\n- In the property editor, select `test` texure slot\n- Change brush radius.\nThe top texture slot (`test`) was selected automatically\n[Node Group Texture Slot Problem.blend](Node_Group_Texture_Slot_Problem.blend)", "Texture Node Editor Using nodes causes weird issues in sculpt mode.\nOperating system: Linux-5.15.0-48-generic-x86_64-with-glibc2.31 64 Bits\nGraphics card: NVIDIA GeForce RTX 2060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 515.76\n\nBroken: version: 3.3.1\nWorked: perhaps never? (can redo in 2.83)\n\nAdding an image from the nodes of the texture editor for example, a sculpting brush, causes really weird issues. Disabling nodes and using the texture properties does not have this issue.\n\n- Check the blend file. Try sculpting, you will see weird noise being displaced with the texture.\n- Now disable the nodes and set the texture type from the texture properties to the same image. Sculpting again you see the issue is not happening.\n[bug.blend](bug.blend)\n\n\n\n", "Edit mode. Wire become non-selectable dependent on resolution (regression)\nOperating system: Windows-7-6.1.7601-SP1 64 Bits\nGraphics card: GeForce GTX 660 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 430.86\n\n\nBroken: version: 2.80 (sub 75)\nWorked: 2.79\n\n\nWithout multisampling wire just disappear and with multisampling it is still visible but not selectable.\n\n[2019-09-14_23-07-53.mp4](2019-09-14_23-07-53.mp4)\n[2019-09-14_23-09-58.mp4](2019-09-14_23-09-58.mp4)\n\n\n\nTry to select loop and zoom in-out with and without viewport multisampling. Here the file, but bug will appear even with UV-sphere.\n[b.blend](b.blend)\n\n" ]

[ "BUG: Blender nodes don't disconnect properly\nOperating system: macOS-10.14.5-x86_64-i386-64bit 64 Bits\nGraphics card: Intel(R) Iris(TM) Graphics 6100 Intel Inc. 4.1 INTEL-12.9.22\n\nBroken: version: 3.0.0\nWorked: N/A\n\nWhen changing a Blender node in a way which disconnects from another, Blender doesn't handle this disconnection properly. This leads to a \"phantom\" connection which appears disconnected, but Blender still thinks it's connected.\n\nFor example, make two Vector Math nodes. Set both to \"Add\". Connect the first to the second. Then set the first to \"Length\". It appears visually disconnected, but Blender still treats it as if it's connected. It won't show the inputs, and you can even drag to disconnect the node despite it not being connected.\n\n1. Make two Vector Math nodes.\n2. Set both to \"Add\".\n3. Connect the first to the second (whichever input doesn't matter)\n4. Set the first node to \"Length\".\n5. Notice how the second behaves connected even though it's not.\n[bugdemo.blend](bugdemo.blend)\n\n\n\n" ]

[ "Compositor : improve Viewer node behavior\nSome issue with the current design:\n* The backdrop only shows the output of the viewer node. This means that for simple cases, users need to manage multiple nodes.\n* It's not obvious when seeing the results of the viewer node that this is not what final rendering will actually output, which leads to mistakes and having to re-render.\n* Viewer nodes work differently in geometry, shading and compositing nodes, mainly for historical rather than good design reasons.\n\nTo address these problems, I propose to make the Viewer node behave more like the one in geometry nodes.\n* If there are no Viewer nodes, then image editor, 3D viewport and node editor backdrop and will show the (active) Composite output.\n* Like geometry nodes, the Viewer node will have a button to enable and disable it. When enabling, editors by default will show the viewer output with some visual indicator that it is a viewer.\n* The editors should have an overlay option to disable showing the viewer node, so it's possible to configure an editor to always show the final output with others showing the viewer output.\n* The \"Viewer Node\" image datablock will be removed.\n\n", "compositor performance issue when larger empty area surrounding the image\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.6.0 NVIDIA 536.99\n\nBroken: version: 4.0.0 Alpha\nWorked: much faster in 2.93\n\ncompositor performance is severely degraded if there is a larger empty area surrounding the pixels. Also becomes more and more sluggish the longer it plays back, despite it only updating a single image block. Task manager doesn't show any increase in memory usage as it becomes slower.\n\n<video src=\"attachment\" controls></video>\n\nGet an image sequence to playback in the compositor's backdrop, either with a image sequence node, or by updating a single image via python and calling render.render to force the playhead to wait for the compositor to finish calculating. Try playing back with the compositor contracted and expanded. See video for demonstration.\n\nIssue affects the tiled, full frame, and gpu compositor modes. Also happens when experimental compositor is disabled completely in the preferences.\n\nExample video attached.\n\n", "Node Operators: Add Selection input node\n_No response_\n\n", "Objects referenced in the `UVProject` modifier property are not snapped\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1060 3GB/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 441.66\n\nBroken: version: 2.90 (sub 0)\n\nWhen i select an object as projector transform in the UVProject modifier it doesn't snap to the object I added the modifier to.\n\n- add cube\n- add an empty\n- add uvproject modifier to the cube\n- set the empty as projector transform\n- enable snaping to surface\nor\n- open attached file\nWhen moving the empty it does not snap to the cube\n[bug.blend](bug.blend)", "OpenEXR DWA compression should not apply to non-color data passes\nLinux Mint 17.3 \n\n2.78c\n\nI write you because i've been testing alot with DWAA and DWAB. But there are major features missing. First of all. Blender need a Compression Tab kind of Setup button. I dont find anything in the release build 2.78c.\nI uploaded a screenshot from the compositing app Natron to explain what i mean. \n\nSecond important thing is to make sure which passes get compressed and which are not. So it would be a good idea to create checkerboxes on each layer with a boolean description of \"i want to compress this layer or not\" \n\nLayers with Capital letters not gonna get compressed. please read more in the foundry forum about that topic.: 125476\n\nCheers.", "Cycles destroys an existing texture image while baking to a new image\nOperating system: Windows 11\nGraphics card: rtx 2070\n\nBroken: 51bc44420b89\n\n\nBlender seems to destroy or empty an existing texture in a multi material setup while baking to a new image.\n\n\n\n\nload the attached file\n\nselect the BAKE material\n\nselect the BAKE image node\n\nRender Props -> Bake ( it is already setup)\n\nPay attention to the viewport to see that one of the textures turn black, and the BAKE texture is still black\n\nYou can try variations as in putting the BAKE node in other materials however the result seems to be the same failed bake.\n\n", "Object Info node random output doesn't work properly\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: AMD Radeon HD 7600M Series ATI Technologies Inc. 4.5.13399 Core Profile Context 0\n\nBroken: version: 3.6.2\nWorked:\n\n**Short description of error and steps for others to reproduce the error**\nI have a scene with many collection instances of a collection containing a car and its wheels. The car has a material on it with its base color being controlled by an object info node's random output:\n[![enter image description here][1]][1]\n\nEverything looks good as long as I am in material preview with the overlays turned off:\n[![enter image description here][2]][2]\n\nBut with the overlays turned on(or sometimes the other way around) or if rendered using the EEVEE render engine, all the cars will have the same color.\n[![enter image description here][3]][3]\n\nHere is the .blend file:\n4whyzTYK#LYLMOGpAT3VvUi4j73Jg63W2p8sG_HRQ8L3E_A107z0\n\n [1]: Api9y.png\n [2]: XZ0Gz.png\n [3]: OeeZJ.jpg\n\n", "Materials animation doesnt show up in animation actions editor\nOperating system: Linux-5.8.18-300.fc33.x86_64-x86_64-with-fedora-33-Thirty_Three 64 Bits\nGraphics card: GeForce GTX 970M/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 450.66\n\nBroken: version: 2.92.0 Alpha\nWorked: never\n\nMaterials animation doesnt show up in animation actions editor\n\n[#82522.blend](T82522.blend)\n- Open blend\n- keyframes for nodetree actions will be visible in the Graph Editor\n- keyframes for nodetree actions will be visible in the DopeSheet\n- keyframes for nodetree actions will be visible in the NLA Editor\n- keyframes for nodetree actions will **not** be visible in the Action Editor and appropriate Action cannot be selected from the dropdown\n\njust make a base color of material animated and then see that its not in action editor\n\n", "Occasional render glitches when using simulation zones on an object without a material.\nThis bug has an easy workaround, mostly posting this so anyone running into the same issue knows the solution.\n\nOperating system: Linux-5.15.0-79-generic-x86_64-with-glibc2.35 64 Bits, X11 UI\nGraphics card: NVIDIA GeForce RTX 3080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 530.30.02\n\nBroken: version: 3.6.2\nWorked: N/A, 3.6 is the first version with simulation nodes.\n\nIn Eevee, rendering an object with a simple simulation zone applied in geometry nodes and no material assigned to the object results in some rendered frames being blank or displaying odd glitches.\n\nThe workaround is to simply assign a material to the object.\n\nCheck out the attached .blend file for reproduction.\n\n", "Contents of \"Viewer Node\" image block do not get updated \nThis might not necessarily be a bug, but may be a feature request. I apologize in advance for being very unprofessional here. I don't think my problem is necessarily relevant to the way I build Blender of the system hardware I use. I've seen the same problem over and over on different systems. Anyways, I compile Blender from source as follow on a freshly-installed Ubuntu:\n\nHere's how I set up Ubuntu after installing it:\n\n```\napt-get -y update && apt-get -y install locales \\\n make \\\n\t\t\t\t\t dpkg \\\n wget \\\n bzip2 \\\n libglib2.0-0 \\\n libxext6 \\\n libsm6 \\\n libxrender1 \\\n g++ \\\n gcc \\\n xvfb \\\n libyaml-cpp-dev \\\n\t\t\t\t\t git \\\n cmake \\\n\t\t\t\t\t vim \\\n\t\t\t\t\t curl \\\n\t\t\t\t\t ca-certificates \\\n software-properties-common \\\n\t\t\t\t\t python3 \\\n python3-pip\n```\n\n\n```\ngit clone blender.git\ncd blender\ngit checkout 8ef39d5c882896bd75e0d4d17fb3e3d4710fc768 # Blender 2.79\ngit submodule update --init --recursive\ngit submodule foreach git checkout master\ngit submodule foreach git pull --rebase origin master\n\nblender/build_files/build_environment/install_deps.sh --source=/blender-git/\n\nPPATH=\"$(which python3)\"\ncmake blender \\\n -DCMAKE_INSTALL_PREFIX=/usr/lib/python3/dist-packages \\\n -DWITH_INSTALL_PORTABLE=OFF \\\n -DWITH_PYTHON_INSTALL=OFF \\\n -DWITH_PYTHON_MODULE=ON \\\n -DPYTHON_SITE_PACKAGES=/usr/lib/python3/dist-packages \\\n -DPYTHON_VERSION=3.5 \\\n -DWITH_OPENAL=ON \\\n -DWITH_CODEC_AVI=ON \\\n -DWITH_MOD_OCEANSIM=ON \\\n -DWITH_CODEC_FFMPEG=ON \\\n -DWITH_SYSTEM_GLEW=ON \\\n -DWITH_FFTW3=ON \\\n -DWITH_OPENCOLORIO=ON \\\n -DWITH_GAMEENGINE=OFF \\\n -DWITH_PLAYER=OFF\nmake -j 7\nmake install\n```\n\nI want to do a very simple thing but it seems that it's becoming very complicated to do it. I have described what my problem is [here ]]. I do not necessarily want to simulate clicking but it turns out that I cannot update the contents of 'Viewer Node' image block. Finally, I thought the best way would be to automatically have backdrop enabled. So I followed the instructions given [[ https:*blender.stackexchange.com/questions/72742/depth-data-works-but-not-in-background-mode | here ](https:*blender.stackexchange.com/questions/102790/how-to-trigger-lmb-click-events-to-activate-a-specific-view-node-and-update-cont) by *Sebastian Koch* thinking that I should be able to dynamically add/remove view nodes and update the contents of the image block automatically. Although I could access to the pixels, but But this does not happen!\nI wrote the right code to do exactly what I show in [this video ]] but the pixels in \"Viewer Node\" image block do not get updated, as they do in the video. Not only that, I tried activating newly-added Viewer nodes manually (as described [[ https:*blender.stackexchange.com/questions/102790/how-to-trigger-lmb-click-events-to-activate-a-specific-view-node-and-update-cont | here ](https:*youtu.be/YCv5qdKLkZo)) but I am still unsuccessful in updating the pixels in the image block. I also tried removing the image block in Python and then continue adding new Viewer nodes. But from the time I remove the image block its contents will always be zero even if I redo the rendering.\n\nThe only way I could successfully update the contents of the image block was after, first, I followed *Sebastian Koch*'s solution and then wrote some code to remove/add all nodes and their links to get either depth map or surface Normals and did the rendering again. But the whole point of me trying to store pixels into Numpy arrays was to avoid doing re-rendering to get around the IO overhead. I could have instead stored two files (for both depth map and Normal maps) throught two Output nodes with rendering only once.\n\nFirst, if this behavior does not look like a bug, I would appreciate the community provide me an easy solution to do what I want. Second, It would be great if you can improve Blender's Python API and add ways of easily accessing rendering results in memory so that people like me who want to do millions of rendering do not need to deal with the delays cause by IO to transfer rendering results from memory to hard drive and can easily store the results into Numpy arrays and store them on disk in chunks. \n\n", "Can't see wireframe (vertex, selected components) through semi-transparent object\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce RTX 2070 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 442.19\n\nBroken: version: 2.83 (sub 10)\nWorked: c26f470cfeea is last I have by hand\n\nIn the viewport, you can't see wireframe (vertex, selected components) through a semi-transparent object\n\n[75106.blend](75106.blend)\n1 - Create 2 objects one behind the other \n2 - Make front object semi-transparent (via Object properties > Viewport Display > Color > Alpha\n3 - Select the second object in Edit mode\n\n", "compositor blur node set to fast gaussian produces black output when fed a colour output from an EXR image.\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.6.0 NVIDIA 536.99\n\nBroken: version: 4.0.0 Alpha\nWorked: (4.0 alpha from a few weeks ago not sure of version as deleted now)\n\nCompositor blur node set to fast gaussian produces black output when fed a colour output from an EXR image.\n\n1. Open a new blend file.\n2. Go to the compositor.\n3. Drag in the image file attached below.\n4. Connect the resulting image node's 0_Image output to a viewer node and turn on the backdrop.\n5. Image will be shown.\n6. Now place a blur node between the image node and the viewer node.\n7. Set the x and y values of the blur node to 1 and the size to 1.\n8. Image will blur.\n9. Now change the blur mode to fast gaussian.\n10. Image will go completely black.\n\n[test-image](https://projects.blender.orgattachment)", "blender 3.4 and 3.5 hangs during compositing when rendering an animation\nOperating system: Linux-5.4.0-132-generic-x86_64-with-glibc2.31 64 Bits\nGraphics card: NVIDIA GeForce RTX 3080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 470.141.03\n\nBroken: version: 3.4.0 Beta\n\nwhen rendering a longer animation the beta and alpha versions tend to hang on the \"initializing execution\" part of the compositing. No compositing happens on the render in the render window, you see frame n while the playhead already is a frame n+1.\n\nWhen rendering the same animation it happens randomly after 5 frames or after 150 frames, but mostly with 200+ samples and at least HD resolution, smaller res and less samples went through mostly. I compared the same animation file with 3.3.1 and it just worked and the very same file with 3.4 got stuck.\n\n", "Scalar Group Node Output is One Pixel Large\nuname -r: 4.3.0-1-amd64\nDebian 8 testing (sid)\n\nBroken: 2.77a (official tar'ed build from blender.org)\n\nMy compositing node system relies extensively on nodegroups, sometimes nested several layers deep. Sometimes, these nodegroups need to process not a bitmap, but simply an individual scalar value - which might be inputted using the RGB node, or the Value node. It's when outputting a scalar from a nodegroup that it breaks; the output value is a bitmap that is one pixel large.\n\nI suspect that any person or studio interested in creating a nested node system for their own ease of use will run into this issue!\n\n1. Load compositing scene.\n2. Create 2 Math nodes. Put each in its own nodegroup.\n3. Let each nodegroup have one input and one output, passing through the Math node's first input within each group. In both cases set the Math node to Multiply, and the second input to 1.\n4. Pass in the value 1 for the first nodegroup, and let the output of nodegroup 1 become the input of nodegroup 2.\n5. Find any image, and plug it into the first input of a color Multiply node. In input 2, plug in the output of nodegroup 2.\n\n6. Notice as a single pixel in the image remains unchanged, as the rest of the image becomes black.\n\nI attached screenshots to better illustrate the setup:\n\n\n\n", "Intermittent Segfault When Programmatically Creating Node Group\nThere have been intermittent crashes plaguing some of my addons. I think I have narrowed it down to creating node groups using the Python API. Specifically, to connecting something to a new node output.\n\nEnclosed [ldo_node_crash.py](ldo_node_crash.py) is a simple script that seems to reliably reproduce the problem for me; I hope it does for you, too. Open it in, or paste it into, Blender’s text editor, and run with ALT-P. I set the loop to run for 10 iterations, but for me, so far, it has always crashed before that.\n\nI am currently normally running Blender 2.82 on Debian Unstable AMD64; I tried a Blender build from the latest master branch, and the crash happens there, too. The backtrace looks like this:\n\n```\nblender-build-latest-debug/bin/blender(BLI_system_backtrace+0x26) [0x8257e79]\nblender-build-latest-debug/bin/blender() [0x2a09cfc]\nblender-build-latest-debug/bin/blender() [0x2a09efc]\n/lib/x86_64-linux-gnu/libc.so.6(+0x3b7e0) [0x7f5c97ff07e0]\nblender-build-latest-debug/bin/blender() [0x3919d1c]\nblender-build-latest-debug/bin/blender() [0x3919e13]\nblender-build-latest-debug/bin/blender(node_connected_to_output+0xf6) [0x391a0a6]\nblender-build-latest-debug/bin/blender(ED_node_tag_update_nodetree+0x41) [0x390df93]\nblender-build-latest-debug/bin/blender() [0x321f7c6]\nblender-build-latest-debug/bin/blender(NodeLinks_new_call+0x98) [0x323b5c9]\nblender-build-latest-debug/bin/blender(RNA_function_call+0x47) [0x3159a88]\nblender-build-latest-debug/bin/blender() [0x334374e]\n/usr/lib/x86_64-linux-gnu/libpython3.7m.so.1.0(_PyObject_FastCallKeywords+0xd7) [0x7f5c9890c887]\n/usr/lib/x86_64-linux-gnu/libpython3.7m.so.1.0(+0x6dfc3) [0x7f5c9871bfc3]\n/usr/lib/x86_64-linux-gnu/libpython3.7m.so.1.0(_PyEval_EvalFrameDefault+0x68a2) [0x7f5c98722c42]\n/usr/lib/x86_64-linux-gnu/libpython3.7m.so.1.0(_PyEval_EvalCodeWithName+0x92d) [0x7f5c988443dd]\n/usr/lib/x86_64-linux-gnu/libpython3.7m.so.1.0(PyEval_EvalCodeEx+0x3e) [0x7f5c9884465e]\n/usr/lib/x86_64-linux-gnu/libpython3.7m.so.1.0(PyEval_EvalCode+0x1b) [0x7f5c98844e8b]\nblender-build-latest-debug/bin/blender() [0x333463c]\nblender-build-latest-debug/bin/blender(BPY_execute_text+0x39) [0x33348cc]\nblender-build-latest-debug/bin/blender() [0x3969d26]\nblender-build-latest-debug/bin/blender() [0x3969dfb]\nblender-build-latest-debug/bin/blender() [0x2ce7313]\n```" ]

[ "Even when you use compositor ouput, Blender simultaneously rendering garbage frames from \"default\" ouput from Output Properties tab\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce RTX 2080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 451.77\n\nBroken: version: 2.83.4\n\nI want to render animation and save specific passes from layers and scenes. I have everything set up in the compositor and i am oupting it to multilayer exr via compositor ouput node. But Blender simultaneously, although i have ouput in compositor, still rendering gargabe frames in format and directory set in Output Properties tab. It cant be dissable, when i put empty path here render just wont start. This make no sense for me becasue it takes space and write disk memory . It just left garbage frames to delete in default /tmp\\.\nAlso, compositor dont save metadata(for example to use cryptomattes in Fusion or Nuke) but Output Properties tab saves.\n\nJust use compositor and set output file node, after clicking \"render animation\" Blender will save render at the same time to directory in compositor and in Output Properties Tab.\nThere is no way to save only from compositor output.\n", "Doubled Animation Render Output using only File Output Nodes\nWindows 7\n\nBroken: 2.79. 5bd8ac9,\nWorked: presumably never?\n\n**Short Explanation**\nWhen rendering an animation while using compositing output nodes, you get doubled output because of the \"write_still=True\", when the \".base_path\" of the output nodes and \"bpy.data.scenes[\"Scene\"].render.filepath\" differ.\n\nIn a new scene, go into the node editor, tick on \"use nodes\" and switch to the \"compositing nodes\" view.\nReplace the \"Composite\" node with a \"File Output\" node.\nChange its base path to \"[...]\\folder1\"\n\nGo into the \"properties\" area, into the render tab.\nUnder \"output\", change the folder to \"[...]\\folder2\"\n\nRender an animation, you will get an output in both folders.\n\nI would suggest turning off the properties panel output path when \"use nodes\" is True and output nodes are present?", "Rendering with a file output node instead of composite still writes the files twice\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3070 Laptop GPU/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 522.30\n\nBroken: version: 3.3.1\nWorked: ?\n\nI'm pretty sure I did it in the past, but maybe I'm wrong... In which case: sorry.\n\nI'm pretty sure that it used to be possible to basically ignore the scene's file output settings (what's in *Properties* Editor > *Output Properties* Tab > *Output* Panel) via the compositor by having no composite node and run the node tree into a File Output node which uses its own settings.\n\nBut right now, doing that just outputs files twice: once as per the scene's output settings and once as per the File Output node.\n\n\n[outputTest.blend](outputTest.blend)\n\nHere is a simple file that quickly renders low resolution workbench images for speed. If Files are written by the scene output settings, they should be in the same folder as the .blend, while the File Output node should put them in the `nodeOutput` subfolder:\n\n\n\nI thought that with just a File Ouput node and no Composite node, only the Folder with frames would be written." ]

[ "Regression: Attempt to move object with Copy Location (with Offset) crashes\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1050/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 536.67\n\nBroken: version: 3.4.0 Alpha\nWorked: version: 3.4.0 Alpha\n\nA consistent crash when trying to move an object with COPY LOC w/ Offset enabled.\n\n\n1. Open the attached blend file\n\n2. Move the pre-selected object in any direction, you may have to do this two or three times\n\n3. program will crash with EXCEPTION_ACCESS_VIOLATION\n\nThis is happening every time.\n\n```\nStack trace:\nblender.exe :0x00007FF73E03C510 blender::bounds::min_max<blender::VecBase<float,3> >\nblender.exe :0x00007FF73E03BD30 blender::FunctionRef<void __cdecl(blender::Bounds<blender::VecBase<float,3> > &)>::callback_fn<<lam\ntbb.dll :0x00007FF9998C4FD0 tbb::interface7::internal::isolate_within_arena\nblender.exe :0x00007FF7425212F0 blender::CacheMutex::ensure\nblender.exe :0x00007FF73E042390 BKE_mesh_minmax\nblender.exe :0x00007FF73E0409F0 BKE_mesh_boundbox_get\nblender.exe :0x00007FF73E02E940 BKE_object_boundbox_get\nblender.exe :0x00007FF73E02F410 BKE_object_dimensions_get\nblender.exe :0x00007FF73E613520 RNA_property_float_get_index\nblender.exe :0x00007FF73E235050 dtar_get_prop_val\nblender.exe :0x00007FF73E235410 evaluate_driver\nblender.exe :0x00007FF73E168590 calculate_fcurve\nblender.exe :0x00007FF73E156270 BKE_animsys_eval_driver\nblender.exe :0x00007FF73E4464F0 blender::deg::`anonymous namespace'::evaluate_node\nblender.exe :0x00007FF73E446320 blender::deg::`anonymous namespace'::deg_task_run_func\nblender.exe :0x00007FF7425175C0 tbb::internal::function_task<Task>::execute\ntbb.dll :0x00007FF9998CF220 tbb::recursive_mutex::scoped_lock::internal_try_acquire\ntbb.dll :0x00007FF9998CF220 tbb::recursive_mutex::scoped_lock::internal_try_acquire\ntbb.dll :0x00007FF9998C4FD0 tbb::interface7::internal::isolate_within_arena\ntbb.dll :0x00007FF9998CA120 tbb::task_scheduler_init::terminate\ntbb.dll :0x00007FF9998CD800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FF9998CD800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FF9A5D192C0 recalloc\nKERNEL32.DLL :0x00007FF9A6F72690 BaseThreadInitThunk\nntdll.dll :0x00007FF9A882AA40 RtlUserThreadStart\n```\n\n", "Crash with certain file. Sculpt mode. Vertex color fill + edge.\nOperating system: Windows-8.1-6.3.9600-SP0 64 Bits\nGraphics card: GeForce GTX 660 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 461.40\n\nBroken: version: 2.93.0 Alpha\n\nWhen I try to apply vertex color Fill (Sculpt mode) on edge in this particular file, I get crash\nAnd all fine for shapes/triangle.\n\n[VC + edge crush.blend](VC___edge_crush.blend)\n\n[debug_logs.7z](debug_logs.7z)\n", "USD: Crash when loading the Intel Moore Lane demo asset (related to Color Attributes)\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 4090/PCIe/SSE2 NVIDIA Corporation 4.6.0 NVIDIA 528.24\n\nBroken: version: 4.0.0 Alpha\nBroken: version: 4.0.0 Alpha\nWorked: version: 4.0.0 Alpha\n\nCrashes during import of the Intel 4004 Moore Lane asset. Seemingly related to Color Attribute processing though the stack traces for the crash are not related.\n\n- Download the \"4004 Moore Lane v 1.1.0\" asset: 4004-moore-lane/\n- Attempt to import the `USD\\4004MooreLane_III_03_ASWF.usd` file with default settings --> observe crash\n- Attempt to import the same file but with the \"Color Attributes\" option unselected --> works\n\nOddly, the crashing stack occurs inside tbbmalloc and while seemingly dealing with mesh normals...\n```\n...\ntbbmalloc.dll!00007ffd7fa29f4e()\tUnknown\nblender.exe!MEM_lockfree_mallocN_aligned(unsigned __int64 len, unsigned __int64 alignment, const unsigned char * str) Line 310\tC\n[Inline Frame] blender.exe!blender::GuardedAllocator::allocate(unsigned __int64) Line 47\tC++\n[Inline Frame] blender.exe!blender::Array<blender::VecBase<float,3>,4,blender::GuardedAllocator>::allocate(__int64) Line 425\tC++\n[Inline Frame] blender.exe!blender::Array<blender::VecBase<float,3>,4,blender::GuardedAllocator>::get_buffer_for_size(__int64) Line 419\tC++\n[Inline Frame] blender.exe!blender::Array<blender::VecBase<float,3>,4,blender::GuardedAllocator>::{ctor}(__int64) Line 122\tC++\nblender.exe!blender::bke::mesh::mesh_normals_loop_custom_set(blender::Span<blender::VecBase<float,3>> positions, blender::Span<blender::VecBase<int,2>> edges, const blender::offset_indices::OffsetIndices<int> faces, blender::Span<int> corner_verts, blender::Span<int> corner_edges, blender::Span<blender::VecBase<float,3>> vert_normals, blender::Span<blender::VecBase<float,3>> face_normals, const bool * sharp_faces, const bool use_vertices, blender::MutableSpan<blender::VecBase<float,3>> r_custom_loop_normals, blender::MutableSpan<bool> sharp_edges, blender::MutableSpan<blender::VecBase<short,2>> r_clnors_data) Line 1368\tC++\nblender.exe!blender::bke::mesh::mesh_set_custom_normals(Mesh * mesh, float[3] * r_custom_nors, const bool use_vertices) Line 1642\tC++\nblender.exe!blender::io::usd::USDMeshReader::process_normals_face_varying(Mesh * mesh) Line 797\tC++\nblender.exe!blender::io::usd::USDMeshReader::read_mesh_sample(blender::io::usd::ImportSettings * settings, Mesh * mesh, double motionSampleTime, bool new_mesh) Line 853\tC++\nblender.exe!blender::io::usd::USDMeshReader::read_mesh(Mesh * existing_mesh, USDMeshReadParams params, const char * * __formal) Line 1094\tC++\nblender.exe!blender::io::usd::USDMeshReader::read_object_data(Main * bmain, double motionSampleTime) Line 249\tC++\nblender.exe!blender::io::usd::import_startjob(void * customdata, bool * stop, bool * do_update, float * progress) Line 288\tC++\n\n```\n\n", "Cycle crash\n[System Information](https://projects.blender.orgattachment)\n\nBlender version: 3.4.1\n\nCrash when switching from Material Preview to Rendered in Viewport Shading and GPU Compute was set for Cycles. In CPU mode it's OK.\n\n\nOpen project.\nSelect CPU in Cycles settings.\nSwitch from Material preview to Rendered in Viewport shading.\nNow it works.\nChange to GPU Compute in Cycles settings.\nCrash...\n\n", "Regression: Crash on deleting shape key from Blender File view.\nOperating system: macOS-13.5-x86_64-i386-64bit 64 Bits\nGraphics card: Metal API AMD Radeon Pro 575 1.2\n\nBroken: version: 4.0.0, 3.2\nWorked: 3.1\n\nCaused by a71a513def202b119328035dbd68e86c2c47f7ac\n\nOutliner on Blender File when delete cause crash guide.\n\n- Open attached file\n- Add shape key\n- Switch outliner to Blender file display mode\n- Expand Shape key\n- select `key` then delete it from context menu\n\nr2jrCca\n\nYoutube:\nOlegeJrBbmE\n\nMy Dropbox file:\nLola%20Bunny.zip?dl=0\n\n- - -\nFirst create new ticket I unable to paste picture this here.\nI did drag there 250MB and (Remove) cause appear <html><head></head><body></body><html> please fix without <html> k5RFuuy\n\n", "Cycle gpu rendering crashes randomly\nOperating system: Win11\nGraphics card: Nvidia RTX 3070 Ti\n\nbroken: all 3.0 versions\n\nRenders keep crashing randomly when I’m using my gpu with cycles, I updated my drivers, my memory use is okay, I tried to not use denoiser, lower my samples, resolution etc but nothing works, sometimes it render 10 frames sometimes 100 but it can't render my whole animation without crashing (also for animations with low poly count and small textures tiles) \nDon't know how to read the crash log file.\n\n", "Blender crash on combining textures (intel)\nOperating system: Windows 10\nGraphics card: Intel HD Graphics 4400\n\nBroken: 2.93.2\nWorked:\n\nWhen I tried to combine 2 textures like this blender is crashing.\n  \n\n[Test.blend](Test.blend)\n[Test.blend1](Test.blend1)\n[system-info.txt](system-info.txt)\n[Test.crash.txt](Test.crash.txt)\n[2021-11-12 09-11-32.mp4](2021-11-12_09-11-32.mp4)\n", "Add error message feedback on constraints, similar to what already exists for modifiers.\nCurrently de-activated constraints only show their status through a small red icon. It would be better for users to also show an error message explaining what is the issue, similar to what modifiers already do.\n\nSee e.g. the invalid settings for the IK constraint of the first bone in this file from #100128: [lib.blend](lib.blend)", "Blender Crashes on Render (Save As), then refuses to start\nOperating system: Windows 10\nGraphics card: NVIDIA GeForce RTX 3060\n\nBroken: 3.6.1\nWorked: N/A\n\nFirst observed with the attached file. I tried to save a png image render from it; when I clicked \"Save As,\" Blender crashed and then... wouldn't open. It seemed to initiate startup, because the little black window that usually notified me that it was reading my userprefs.py file would blink into existence just long enough for me to register what it said, then close itself and act like I hadn't clicked it.\n\nRestarting my computer fixes the problem temporarily, until the next crash. The crashes are inconsistent, but seem to happen only when I hit main menu items like Open or Save. This includes previous versions, apparently, as I still had 3.5 installed and it does the same quick blink.\n\nThe only thing I could think of was that that Read line in the black window referred to finding the userprefs file in C:\\ when I installed everything into M:\\, but I tried copying the required path over to C:\\ with no effect and also it gives me that same message when it's working and starts up fine.\n\nI also tried repairing the 3.6.1 install, and uninstalling/reinstalling, to no effect.\n\nBased on the default startup or an attached .blend file (as simple as possible).\n\n", "blender crash itself while importing a dae (collide) file\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1650 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.94\n\nBroken: version: 3.3.1\nBroken: version: 2.82 and all versions in between then and now\nWorked: n/a\n\nHi, I'm using blender version 3.3 on a windows 10 downloaded on steam.\n\nwhen i try to import dae files, blender crashes itself and shuts down\n(this only happens with some dae files, not all and the ones who crash are not to big ).\n\nI tried to search with powershell to find the problem and I got this:\n\nError: EXCEPTION_ACCESS_VIOLATION\n[bgr_darryls_ship_ground.dae](bgr_darryls_ship_ground.dae)\n\n```\nThe gh->hashfp function pointer is NULL (hasn't been initialized yet)\n\nblender.exe!ghash_keyhash(const GHash * gh, const void * key) Line 133\tC\nblender.exe!ghash_lookup_entry(const GHash * gh, const void * key) Line 411\tC\nblender.exe!BLI_ghash_lookup(const GHash * gh, const void * key) Line 736\tC\nblender.exe!BKE_armature_find_bone_name(bArmature * arm, const unsigned char * name) Line 601\tC\nblender.exe!ArmatureImporter::create_armature_bones(Main * bmain, std::vector<Object *,std::allocator<Object *>> & arm_objs) Line 478\tC++\nblender.exe!ArmatureImporter::make_armatures(bContext * C, std::vector<Object *,std::allocator<Object *>> & objects_to_scale) Line 826\tC++\nblender.exe!DocumentImporter::finish() Line 228\tC++\n[External Code]\t\nblender.exe!DocumentImporter::import() Line 127\tC++\n```\n", "Crash while using several rigid bodies and some force fields.\nOperating system: Darwin-18.7.0-x86_64-i386-64bit 64 Bits\nGraphics card: AMD Radeon R9 M370X OpenGL Engine ATI Technologies Inc. 4.1 ATI-2.11.20\n\n\nBroken: version: 2.80 (sub 75)\nWorked: (optional)\n\n\nThe scene is pretty simple, a house with some elements used as rigid bodies and some force field, vortex and wind force are applied.\n\nWhen I play the animation, it starts and around frame 140-200 it crashes.\n\nAttached blend file. The only thing to do is to play the animation and wait for the crash.\nAttached crash report.[Blender_2019-08-26-215345_MacBook-Pro-de-Kote.crash](Blender_2019-08-26-215345_MacBook-Pro-de-Kote.crash)\n\n[casaFalla.blend](casaFalla.blend)\n\n", "Remove and re-add the rigid body world causes Blender to crash.\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 517.48\n\nBroken: version: 3.3.1\nWorked: I dont know.\n\nWhen the physical world is removed and added again, playback will cause the program to crash.\n\n\n1. Download the 2.blend and open it.\n2. Play 5 frame.\n3. Click the Remove Rigid Body World button.\n4. Click the Add Rigid Body World button.\n5. Click the rigid body world setting Collection. Then select RigidBodyWorld .\n6. Click the rigid body world setting Constraints. Then select RigidBodyConstraints .\n7. Move to 1 frame.\n8. Play 5 frame. Blender will crash immediately.\n\n[2.blend](2.blend)", "Crash with Boolean modifier\nOperating system: Windows-10-10.0.19043-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 496.13\n\nBroken: version: 3.4.0 Alpha\nWorked: 3.1.2 (appears to have been broken since 3.2.0)\n\nBlender crashes when performing an Intersection Boolean between a cube and a bunch of loose faces. (Granted, this is probably not something you should do in the first place, but at least it should be limited to producing a bad result, not a crash.)\n\nI didn't find a crash log, but it's easy to reproduce.\n\n* Load the attached .blend file.\n* Scrub the timeline to frame 209. Blender crashes.\n\n[crash.blend](crash.blend)\n", "Old blend file corrupts to crash blender, I guess\nOperating system: Windows-10-10.0.19042-SP0 64 Bits\nGraphics card: AMD Radeon R9 200 Series ATI Technologies Inc. 4.5.14761 Core Profile Context 20.11.2 27.20.14540.4003\n\nBroken: version: 3.0.0\n\nCrashes here and there. RN I witness crush form select/deselect. help plz ;c\n\nI do, tho, have installed two PROBABLY outdated addons BUT in new file crash doesn't seem to happen.\n\nThe problem is - Idk exact steps, but overall it takes to select/deselect all/smth at all few times\n\nSteps\n- Open the provided .blend below\n- Enter edit mode on the main SN4K3Mesh\n- Select-All, De-Select-All (Repeat)\n\n[#94213-asan.txt](T94213-asan.txt)\n\n```\n# Blender 3.0.0, Commit date: 2021-12-02 18:35, Hash f1cca3055776\nbpy.ops.object.editmode_toggle() # Operator\nbpy.ops.mesh.select_all(action='SELECT') # Operator\nbpy.ops.mesh.select_all(action='DESELECT') # Operator\nbpy.ops.mesh.select_all(action='SELECT') # Operator\n\n# backtrace\nException Record:\n\nExceptionCode : EXCEPTION_ACCESS_VIOLATION\nException Address : 0x00007FF715993808\nException Module : blender.exe\nException Flags : 0x00000000\nException Parameters : 0x2\n\tParameters[0] : 0x0000000000000000\n\tParameters[1] : 0xFFFFFFFFFFFFFFFF\n```\n```\n# Blender 3.0.0, Commit date: 2021-12-02 18:35, Hash f1cca3055776\nbpy.ops.object.editmode_toggle() # Operator\nbpy.ops.mesh.select_all(action='SELECT') # Operator\nbpy.ops.mesh.select_all(action='DESELECT') # Operator\nbpy.ops.mesh.select_all(action='SELECT') # Operator\nbpy.ops.mesh.select_all(action='DESELECT') # Operator\n\n# backtrace\nException Record:\n\nExceptionCode : EXCEPTION_ACCESS_VIOLATION\nException Address : 0x00007FF71599380C\nException Module : blender.exe\nException Flags : 0x00000000\nException Parameters : 0x2\n\tParameters[0] : 0x0000000000000000\n\tParameters[1] : 0xFFFFFFFFFFFFFFFF\n```\n\n[SN4K3.blend](SN4K3.blend)\n\n[SN4K3.crash.txt](SN4K3.crash.txt)", "Access violation when running multiple quick liquid bakes (Python API)\nOperating system: Windows 10\n\nBroken: 3.3 alpha, 3.3, 3.4 alpha, 3.4\nWorked: Not any recent releases\n\nSeems related to T98427 , but the fix for that issue did not fix this.\nAccess violation when running multiple quick liquid bakes\n\n```lines=18\n# backtrace\nException Record:\n\nExceptionCode : EXCEPTION_ACCESS_VIOLATION\nException Address : 0x00007FF9DE2B9224\nException Module : python310.dll\nException Flags : 0x00000000\nException Parameters : 0x2\n\tParameters[0] : 0x0000000000000000\n\tParameters[1] : 0x0000000000000008\n\n\nStack trace:\npython310.dll :0x00007FF9DE2B9220 PyDict_GetItem_KnownHash\npython310.dll :0x00007FF9DE2CF200 PyFrame_GetBack\npython310.dll :0x00007FF9DE430410 PyRun_FileExFlags\npython310.dll :0x00007FF9DE430410 PyRun_FileExFlags\npython310.dll :0x00007FF9DE430270 PyRun_StringFlags\nblender.exe :0x00007FF64D0F4E60 MANTA::runPythonString\nblender.exe :0x00007FF64D0DAF70 MANTA::~MANTA\nblender.exe :0x00007FF64D0DB340 MANTA::`vector deleting destructor'\nblender.exe :0x00007FF6484EB4D0 BKE_fluid_modifier_freeDomain\nblender.exe :0x00007FF6484EB4A0 BKE_fluid_modifier_free\nblender.exe :0x00007FF648274A00 BKE_modifier_free_ex\nblender.exe :0x00007FF648488760 BKE_object_free_modifiers\nblender.exe :0x00007FF6484834A0 object_free_data\nblender.exe :0x00007FF6487EB890 blender::deg::deg_free_copy_on_write_datablock\nblender.exe :0x00007FF6487ECA40 blender::deg::IDNode::destroy\nblender.exe :0x00007FF6487E1F70 blender::deg::Depsgraph::clear_id_nodes\nblender.exe :0x00007FF6487E21E0 DEG_graph_free\nblender.exe :0x00007FF64D64FA80 BLI_ghash_free\nblender.exe :0x00007FF648450510 scene_free_data\nblender.exe :0x00007FF64846F120 BKE_id_free_ex\nblender.exe :0x00007FF64826E980 BKE_main_free\nblender.exe :0x00007FF6482628F0 BKE_blender_free\nblender.exe :0x00007FF6486F1AB0 WM_exit_ex\nblender.exe :0x00007FF6486F1A60 WM_exit\nblender.exe :0x00007FF6477F1260 main\nblender.exe :0x00007FF64D8788C0 __scrt_common_main_seh\nKERNEL32.DLL :0x00007FFA48397020 BaseThreadInitThunk\nntdll.dll :0x00007FFA49C42630 RtlUserThreadStart\n\n\nThreads:\nThread : 00004eb4\nntdll.dll :0x00007FFA49C90780 ZwWaitForWorkViaWorkerFactory\nntdll.dll :0x00007FFA49C42680 TpReleaseCleanupGroupMembers\nKERNEL32.DLL :0x00007FFA48397020 BaseThreadInitThunk\nntdll.dll :0x00007FFA49C42630 RtlUserThreadStart\n\n\nThread : 000051dc\nntdll.dll :0x00007FFA49C90780 ZwWaitForWorkViaWorkerFactory\nntdll.dll :0x00007FFA49C42680 TpReleaseCleanupGroupMembers\nKERNEL32.DLL :0x00007FFA48397020 BaseThreadInitThunk\nntdll.dll :0x00007FFA49C42630 RtlUserThreadStart\n\n\nThread : 00004114\nntdll.dll :0x00007FFA49C90780 ZwWaitForWorkViaWorkerFactory\nntdll.dll :0x00007FFA49C42680 TpReleaseCleanupGroupMembers\nKERNEL32.DLL :0x00007FFA48397020 BaseThreadInitThunk\nntdll.dll :0x00007FFA49C42630 RtlUserThreadStart\n\n\nThread : 00004ab8\nntdll.dll :0x00007FFA49C8CDB0 NtWaitForSingleObject\nKERNELBASE.dll :0x00007FFA47831A40 WaitForSingleObjectEx\nblender.exe :0x00007FF64E410730 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF64E40FF40 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF6479B3A10 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFA473D1B20 configthreadlocale\nKERNEL32.DLL :0x00007FFA48397020 BaseThreadInitThunk\nntdll.dll :0x00007FFA49C42630 RtlUserThreadStart\n\n\nThread : 000033dc\nntdll.dll :0x00007FFA49C8CDB0 NtWaitForSingleObject\nKERNELBASE.dll :0x00007FFA47831A40 WaitForSingleObjectEx\nblender.exe :0x00007FF64E410730 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF64E40FF40 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF6479B3A10 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFA473D1B20 configthreadlocale\nKERNEL32.DLL :0x00007FFA48397020 BaseThreadInitThunk\nntdll.dll :0x00007FFA49C42630 RtlUserThreadStart\n\n\nThread : 00005b3c\nntdll.dll :0x00007FFA49C8CDB0 NtWaitForSingleObject\nKERNELBASE.dll :0x00007FFA47831A40 WaitForSingleObjectEx\nblender.exe :0x00007FF64E410730 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF64E40FF40 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF6479B3A10 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFA473D1B20 configthreadlocale\nKERNEL32.DLL :0x00007FFA48397020 BaseThreadInitThunk\nntdll.dll :0x00007FFA49C42630 RtlUserThreadStart\n\n\nThread : 000044b0\nntdll.dll :0x00007FFA49C8CDB0 NtWaitForSingleObject\nKERNELBASE.dll :0x00007FFA47831A40 WaitForSingleObjectEx\nblender.exe :0x00007FF64E410730 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF64E40FF40 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF6479B3A10 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFA473D1B20 configthreadlocale\nKERNEL32.DLL :0x00007FFA48397020 BaseThreadInitThunk\nntdll.dll :0x00007FFA49C42630 RtlUserThreadStart\n\n\nThread : 000033f0\nntdll.dll :0x00007FFA49C8CDB0 NtWaitForSingleObject\nKERNELBASE.dll :0x00007FFA47831A40 WaitForSingleObjectEx\nblender.exe :0x00007FF64E410730 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF64E40FF40 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF6479B3A10 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFA473D1B20 configthreadlocale\nKERNEL32.DLL :0x00007FFA48397020 BaseThreadInitThunk\nntdll.dll :0x00007FFA49C42630 RtlUserThreadStart\n\n\nThread : 000031f8\nntdll.dll :0x00007FFA49C8CDB0 NtWaitForSingleObject\nKERNELBASE.dll :0x00007FFA47831A40 WaitForSingleObjectEx\nblender.exe :0x00007FF64E410730 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF64E40FF40 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF6479B3A10 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFA473D1B20 configthreadlocale\nKERNEL32.DLL :0x00007FFA48397020 BaseThreadInitThunk\nntdll.dll :0x00007FFA49C42630 RtlUserThreadStart\n\n\nThread : 00004aa4\nntdll.dll :0x00007FFA49C8CDB0 NtWaitForSingleObject\nKERNELBASE.dll :0x00007FFA47831A40 WaitForSingleObjectEx\nblender.exe :0x00007FF64E410730 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF64E40FF40 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF6479B3A10 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFA473D1B20 configthreadlocale\nKERNEL32.DLL :0x00007FFA48397020 BaseThreadInitThunk\nntdll.dll :0x00007FFA49C42630 RtlUserThreadStart\n\n\nThread : 000049b8\nntdll.dll :0x00007FFA49C8CDB0 NtWaitForSingleObject\nKERNELBASE.dll :0x00007FFA47831A40 WaitForSingleObjectEx\nblender.exe :0x00007FF64E410730 IlmThread_3_1::Semaphore::wait\nblender.exe :0x00007FF64E40FF40 IlmThread_3_1::ThreadPool::numThreads\nblender.exe :0x00007FF6479B3A10 std::thread::_Invoke<std::tuple<void (__cdecl usdBlender__pxrReserved__::HdRenderThread::*)(void) _\nucrtbase.dll :0x00007FFA473D1B20 configthreadlocale\nKERNEL32.DLL :0x00007FFA48397020 BaseThreadInitThunk\nntdll.dll :0x00007FFA49C42630 RtlUserThreadStart\n\n\nThread : 00005340\nntdll.dll :0x00007FFA49C90780 ZwWaitForWorkViaWorkerFactory\nntdll.dll :0x00007FFA49C42680 TpReleaseCleanupGroupMembers\nKERNEL32.DLL :0x00007FFA48397020 BaseThreadInitThunk\nntdll.dll :0x00007FFA49C42630 RtlUserThreadStart\n\n\nThread : 00002654\nntdll.dll :0x00007FFA49C90780 ZwWaitForWorkViaWorkerFactory\nntdll.dll :0x00007FFA49C42680 TpReleaseCleanupGroupMembers\nKERNEL32.DLL :0x00007FFA48397020 BaseThreadInitThunk\nntdll.dll :0x00007FFA49C42630 RtlUserThreadStart\n\n\nThread : 00004c64\nntdll.dll :0x00007FFA49C8CDB0 NtWaitForSingleObject\nKERNELBASE.dll :0x00007FFA47831A40 WaitForSingleObjectEx\ntbb.dll :0x00007FFA373BD800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFA373BD800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFA473D1B20 configthreadlocale\nKERNEL32.DLL :0x00007FFA48397020 BaseThreadInitThunk\nntdll.dll :0x00007FFA49C42630 RtlUserThreadStart\n\n\nThread : 000024a4\nntdll.dll :0x00007FFA49C8CDB0 NtWaitForSingleObject\nKERNELBASE.dll :0x00007FFA47831A40 WaitForSingleObjectEx\ntbb.dll :0x00007FFA373BD800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFA373BD800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFA473D1B20 configthreadlocale\nKERNEL32.DLL :0x00007FFA48397020 BaseThreadInitThunk\nntdll.dll :0x00007FFA49C42630 RtlUserThreadStart\n\n\nThread : 00005664\nntdll.dll :0x00007FFA49C8CDB0 NtWaitForSingleObject\nKERNELBASE.dll :0x00007FFA47831A40 WaitForSingleObjectEx\ntbb.dll :0x00007FFA373BD800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFA373BD800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFA473D1B20 configthreadlocale\nKERNEL32.DLL :0x00007FFA48397020 BaseThreadInitThunk\nntdll.dll :0x00007FFA49C42630 RtlUserThreadStart\n\n\nThread : 000048c4\nntdll.dll :0x00007FFA49C8CDB0 NtWaitForSingleObject\nKERNELBASE.dll :0x00007FFA47831A40 WaitForSingleObjectEx\ntbb.dll :0x00007FFA373BD800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFA373BD800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFA473D1B20 configthreadlocale\nKERNEL32.DLL :0x00007FFA48397020 BaseThreadInitThunk\nntdll.dll :0x00007FFA49C42630 RtlUserThreadStart\n\n\nThread : 00005b4c\nntdll.dll :0x00007FFA49C8CDB0 NtWaitForSingleObject\nKERNELBASE.dll :0x00007FFA47831A40 WaitForSingleObjectEx\ntbb.dll :0x00007FFA373BD800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFA373BD800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFA473D1B20 configthreadlocale\nKERNEL32.DLL :0x00007FFA48397020 BaseThreadInitThunk\nntdll.dll :0x00007FFA49C42630 RtlUserThreadStart\n\n\nThread : 00004eb0\nntdll.dll :0x00007FFA49C8CDB0 NtWaitForSingleObject\nKERNELBASE.dll :0x00007FFA47831A40 WaitForSingleObjectEx\ntbb.dll :0x00007FFA373BD800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFA373BD800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFA473D1B20 configthreadlocale\nKERNEL32.DLL :0x00007FFA48397020 BaseThreadInitThunk\nntdll.dll :0x00007FFA49C42630 RtlUserThreadStart\n\n\nThread : 0000446c\nntdll.dll :0x00007FFA49C8CDB0 NtWaitForSingleObject\nKERNELBASE.dll :0x00007FFA47831A40 WaitForSingleObjectEx\ntbb.dll :0x00007FFA373BD800 tbb::thread_bound_filter::try_process_item\ntbb.dll :0x00007FFA373BD800 tbb::thread_bound_filter::try_process_item\nucrtbase.dll :0x00007FFA473D1B20 configthreadlocale\nKERNEL32.DLL :0x00007FFA48397020 BaseThreadInitThunk\nntdll.dll :0x00007FFA49C42630 RtlUserThreadStart\n\n\nLoaded Modules :\n0x00007FF647680000 3.3.0.0 blender.exe C:\\Users\\ecl25031\\blender-3.3.0-alpha+master.4c3b984b3daa-windows.amd64-release\\blender.pdb \n0x00007FFA49BF0000 10.0.19041.1741 ntdll.dll \n0x00007FFA48380000 10.0.19041.1741 KERNEL32.DLL \n0x00007FFA47810000 10.0.19041.1741 KERNELBASE.dll \n0x00007FFA48E00000 10.0.19041.546 WS2_32.dll \n0x00007FFA499D0000 10.0.19041.1706 RPCRT4.dll \n0x00007FFA47C50000 10.0.19041.1503 USER32.dll \n0x00007FFA47B10000 10.0.19041.1766 win32u.dll \n0x00007FFA373B0000 tbb.dll \n0x00007FFA48790000 10.0.19041.1620 GDI32.dll \n0x00007FFA2BF10000 tbbmalloc.dll \n0x00007FFA47B40000 10.0.19041.1706 gdi32full.dll \n0x00007FFA47310000 10.0.19041.789 msvcp_win.dll \n0x00007FFA473B0000 10.0.19041.789 ucrtbase.dll \n0x00007FFA48EA0000 10.0.19041.1682 ADVAPI32.dll \n0x00007FFA48440000 7.0.19041.546 msvcrt.dll \n0x00007FFA486F0000 10.0.19041.1586 sechost.dll \n0x00007FFA49040000 10.0.19041.1741 SHELL32.dll \n0x00007FFA48F50000 10.0.19041.1706 SHLWAPI.dll \n0x00007FFA2FB80000 14.29.30139.0 VCRUNTIME140.dll \n0x00007FFA477C0000 10.0.19041.1620 CFGMGR32.dll \n0x00007FFA15BF0000 14.29.30139.0 MSVCP140.dll \n0x00007FFA43A40000 14.29.30139.0 VCRUNTIME140_1.dll \n0x00007FFA47AE0000 10.0.19041.1023 bcrypt.dll \n0x00007FFA485C0000 10.0.19041.1202 ole32.dll \n0x00007FFA35C50000 10.0.19041.867 dbghelp.dll \n0x00007FFA48A40000 10.0.19041.1741 combase.dll \n0x00007FFA49870000 10.0.19041.546 PSAPI.DLL \n0x00007FFA49B00000 10.0.19041.1741 shcore.dll \n0x00007FFA484E0000 10.0.19041.546 IMM32.dll \n0x00007FFA03B30000 10.0.19041.1741 OPENGL32.dll \n0x00007FF9D3660000 9.0.0.0 openvdb.dll \n0x00007FF9D19D0000 59.18.100.0 avcodec-59.dll \n0x00007FF9F78D0000 59.16.100.0 avformat-59.dll \n0x00007FFA2D720000 59.4.100.0 avdevice-59.dll \n0x00007FF9F8750000 57.17.100.0 avutil-57.dll \n0x00007FFA48970000 10.0.19041.985 OLEAUT32.dll \n0x00007FF9FBCC0000 1.21.1.0 OpenAL32.dll \n0x0000000070440000 1.0.28.0 libsndfile-1.dll \n0x00007FFA0FA30000 6.4.100.0 swscale-6.dll \n0x00007FF9F7770000 2.0.20.0 SDL2.dll \n0x000000006ACC0000 libgmp-10.dll \n0x00007FFA41050000 libgmpxx.dll \n0x00007FFA47F10000 10.0.19041.1741 SETUPAPI.dll \n0x00007FFA40FF0000 tbbmalloc_proxy.dll \n0x00007FF9DE1A0000 3.10.2150.1013 python310.dll \n0x00007FFA30660000 6.10.19041.1110 COMCTL32.dll \n0x00007FFA2AB80000 10.0.19041.1 AVIFIL32.dll \n0x00007FFA415D0000 10.0.19041.546 VERSION.dll \n0x00007FFA44B10000 10.0.19041.746 dwmapi.dll \n0x00007FFA2A530000 10.0.19041.1741 GLU32.dll \n0x0000000070680000 libfftw3-3.dll \n0x00007FFA29450000 4.3.100.0 swresample-4.dll \n0x00007FFA3D230000 10.0.19041.546 Secur32.dll \n0x00007FFA2EAF0000 10.0.19041.1 AVICAP32.dll \n0x00007FFA39CA0000 10.0.19041.546 WINMM.dll \n0x00007FFA28820000 10.0.19041.1 MSACM32.dll \n0x00007FFA291F0000 10.0.19041.1 MSVFW32.dll \n0x00007FFA37020000 10.0.19041.1 winmmbase.dll \n0x00007FFA47200000 10.0.19041.1586 SSPICLI.DLL \n0x00007FFA45AD0000 10.0.19041.546 kernel.appcore.dll \n0x00007FFA452E0000 10.0.19041.1741 windows.storage.dll \n0x00007FFA46D70000 10.0.19041.1682 Wldp.dll \n0x00007FFA47250000 10.0.19041.844 profapi.dll \n0x00007FFA47610000 10.0.19041.1415 bcryptPrimitives.dll \n0x00007FFA48510000 2001.12.10941.16384 clbcatq.dll \n0x00007FFA3FB50000 10.0.19041.1503 MMDevApi.dll \n0x00007FFA470C0000 10.0.19041.1620 DEVOBJ.dll \n0x00007FFA3FC00000 10.0.19041.1741 AUDIOSES.DLL \n0x00007FFA468E0000 10.0.19041.546 powrprof.dll \n0x00007FFA46750000 UMPDC.dll \n0x00007FFA44880000 10.0.19041.1620 uxtheme.dll \n0x00007FFA2D140000 10.0.19041.1566 MPR.dll \n0x00007FFA415E0000 10.0.19041.546 drprov.dll \n0x00007FFA45F70000 10.0.19041.546 WINSTA.dll \n0x00007FFA2ED10000 10.0.19041.1566 ntlanman.dll \n0x00007FFA2EB50000 10.0.19041.546 davclnt.dll \n0x00007FFA22F70000 10.0.19041.546 DAVHLPR.dll \n0x00007FFA46500000 10.0.19041.1645 wkscli.dll \n0x00007FFA27030000 10.0.19041.546 cscapi.dll \n0x00007FFA467B0000 10.0.19041.546 netutils.dll \n0x00007FFA46CC0000 10.0.19041.546 CRYPTSP.dll \n0x00007FFA462D0000 10.0.19041.1052 rsaenh.dll \n0x00007FFA46CE0000 10.0.19041.546 CRYPTBASE.dll \n0x000002627AAF0000 3.10.2150.1013 python3.DLL \n0x00007FFA2CA10000 3.10.2150.1013 _bz2.pyd \n0x00007FFA28ED0000 3.10.2150.1013 _lzma.pyd \n0x00007FFA2C980000 3.10.2150.1013 _socket.pyd \n0x00007FFA46770000 10.0.19041.546 IPHLPAPI.DLL \n0x00007FFA40E90000 3.10.2150.1013 select.pyd \n\n# Python backtrace\n\n```\n\nPut this ins a file called test.py\n```\n\nimport bpy\n\nCUBE = bpy.data.objects[\"Cube\"]\nCUBE.select_set(True)\nbpy.ops.object.quick_liquid()\nbpy.ops.fluid.bake_all()\n\nbpy.ops.object.select_all(action='DESELECT') \nmesh = bpy.data.meshes.new('Basic_Cube')\nCUBE2 = bpy.data.objects.new(\"Cube2\", mesh)\nCUBE2 = bpy.data.objects[\"Cube2\"]\nbpy.context.scene.collection.objects.link(CUBE2)\nCUBE2.select_set(True) \nbpy.ops.object.quick_liquid()\nbpy.ops.fluid.bake_all()\n```\n\nthen, in terminal, run:\n\n```\nblender -b --python test.py \n```" ]

[ "blender crashes when add a inverse kineatics constraint to bone\nwhen added a inverse kinematics to a bone it causes a crash.\n\nmy OS:\nwindows 7 ultimate x86\nprocessor: amd phenon 9500 quadcore 2.21GH\nnvidia geforce 6050SE nforce 430\n\nblender version:\nblender 2.78a hash e8299c8\n\nthe bug not is happening in previous versions.\n\nsteps\nopen blender, delete defaut cube and add a single bone, go to edit mode, extrude the bone tree times and duplicate the last bone, unparent it, go to pose mode and add a inferse kinematic constraint setup, uncheck use tail opition, and it will crash blender.\n\nor simply enable auto IK and try editing the pose.\n\nthis file crashed blender all times:\n\nconsole screenshot on exact moment of crash:\n\n[crash.blend](crash.blend)\nwatch?v=yce51lIKgvU&feature=youtu.be\n\nsory my bad english.\n" ]

[ "Support Preview Image Storage for More Data-block Types\nBasically introduce the `Preview *preview` member for each data-block/ID type that should be \"previewable\" in the Asset Browser. This doesn't mean adding support for automatic previews, but at least to support letting the user set custom previews.\n\n* - [x] Node Groups f315a46982\n* - [ ] Cameras\n* - [ ] Armatures\n* - [ ] Masks\n* - [x] Grease Pencil Strokes (`bGPdata`) (already done in `asset-greasepencil` branch.\n...", "Animation: create a full separate copy of BoneCollections while in armature edit mode\nThis is to avoid touching the main data structures in the Armature datablock while in edit mode, which interacts badly with undo.\n\nRelevant:\n- #111965\n\n", "Remove redundant linked & library override icons\nFor some reason we have a whole bunch of redundant icons for linked data and library overrides. \n\nThese link icons are redundant (only one of each type should be kept):\n- `LINKED`\n- `DECORATE_LINKED` (unused)\n- `LINK_BLEND`\nSame with these library override icons:\n- `LIBRARY_DATA_OVERRIDE`\n- `DECORATE_LIBRARY_OVERRIDE` (only used in custom properties UI, could be replaced easily)\n\nAlso, the `RESTRICT_INSTANCED_OFF` icon looks like thicker version of the linked icon. It and its `RESTRICT_INSTANCED_ON` counterpart are not used.\n\n----\n\nRelated:\n- This looks like the link icons, but indents to indicate locked/unlocked.  Should just use a lock instead.\n- Here the decorator is supposed to indicate a library override (it's `DECORATE_OVERRIDE`), but doesn't look like that at all. Can use the regular override icon instead ", "Missing DNA Defaults (\"Reset to Default Value\" sets to 0)\nSince f5e0dfe59c, we have a simple & practical way of setting up defaults for DNA data. These defaults are used for the \"Reset to Default Value\" operator, available in the context menu for buttons.\nHere's a helper script to extract the defaults: 432509\n\nThere are still a number of properties that don't have a default assigned. These can be reported here.\n\n\nNote that these missing defaults typically appear in \"groups\" because the entire data-structure misses the default value definition. In that case there's no need to list each individual value, just mention where they are (e.g. Armature object, relation settings).", "Add support for 'editable but not saved' properties for liboverrides\n**Status:** WIP\n\n---\n\n**Team**\n**Commissioner:** @SimonThommes \n**Project leader:** @mont29 \n\n**Description**\n**Big picture:** A few properties of linked data are still editable by the user (though not saved on disk), using the `PROP_LIB_EXCEPTION` RNA flag. A similar behavior could be useful for overrides too.\n\n**Use cases**:\n* Deal with temporary not-really-working material overrides cases (it is currently possible to override the active material, however if assigned to obdata and said obdata is not overridden, this will break materials assignments.).\n* On a larger perspective, allow finer control over some properties when using override.\n\n**Design:**\n* `?` *Mental model, end user usability, paradigms, mockups.*\n\n**Engineer plan:**\n* `?` *Abstract code samples - linked here, UML, API guidelines, library changes.*\n\n**Work plan**\n\n**{icon check-circle color=green} Milestone 1 - optional name**\n*Optional description of milestone.*\nTime estimate: `?`\n\n*All existent tasks edited out (removed) since the milestone is fully done.*\n\n**Milestone 2 - optional name**\n*Optional description of milestone.*\nTime estimate: `?`\n\n* - [x]*Task or feature.*\n* - [ ]*Task or feature.*\n* . . .\n\n**Branch**: `-` *Existing patches or branches.*\n\n---\n\n**Relevant links**:\n* `-`", "Normal map support in Cycles standalone\nWindows 8.1 64-bit\n\n**Cycles Version**\nd553b8de7180\n\n\nIn the latest version available in the Cycles standalone repository, commit d553b8de7180, using a normal map node in a shader graph will cause Cycles to crash. This behavior was not present in release 1.7.0 but has been added since then.\n\nThis looks to be caused by an improperly initialized member variable in the ccl::NormalMapNode type. Upon creating an object of type NormalMapNode, the ustring named 'attribute' is initialized such that attribute.c_str() will return a nullptr.\n\nCycles crashes when the nullptr used to construct a std::string in NormalMapNode::attributes in this bit:\n\n```\t\n\t\telse {\n\t\t\tattributes->add(ustring((string(attribute.c_str()) + \".tangent\").c_str()));\n\t\t\tattributes->add(ustring((string(attribute.c_str()) + \".tangent_sign\").c_str()));\n\t\t}\n```\n\n\nI have attached a simple cycles standalone scene that triggers this error here: [sample_scene.xml](sample_scene.xml)", "Usability issues with old group operators\nBroken: 2.79, 2.8x\n\n\nWhen using linked objects, all the Group commands [Remove From Group / Remove From All Groups / Add Selected To Active Group / Remove Selected From Active Group] except [Create New Group (Ctrl G)], do nothing\n\n\nIn my particular usage, I link a scene from one file into my main scene, afterwards I create a [New scene with Link Objects] from the Linked scene. \n\n [GroupsFail.blend](GroupsFail.blend)\n [GroupsFail-MainScene.blend](GroupsFail-MainScene.blend)\n\nI can Create Groups by selected objects, and I can remove the group from the object in the Object Properties tab but the only way to remove multiple groups is through python script or deleting the group itself.\n\nSince this report is used to more generally handle issues with the old group operators and other reports have already been merged here, will also list these:\n- Ctrl G is placing new collections inside \"Orphan Data\"", "Recursive Purge Orphans datablocks on file save\nFully deleting datablocks from a blend file doesn't happen as easily as most users might expect, since \"delete\" is often used in the UI when \"un-link\" is what is actually meant. This can cause a lot of confusion and uncertainty during production. Proper deletion of datablocks requires that they are first un-linked, then they can be purged via File->Clean Up->Recursive Unused Datablocks. This works beautifully, and I would like to propose adding an option to have this operation executed after existing deletion operators. For example, after deleting a collection of objects, it would then also delete everything that was used only by those objects, such as their mesh and shape key datablocks, materials, actions, textures, etc.\n\nMy previous proposal was to have an option to execute Recursive Purge only on file save, which seemed to make sense to everyone I talked to at the time. (@dr.sybren @mont29 @JulianEisel @SimonThommes ) \nI still think that solution would be a step forward, but the new proposal above would achieve the same goals in an even more foolproof way. \n\n### Use case\nCurrently, this can very easily happen:\n- Animator duplicates an overridden character. All its objects get a .001 suffix.\n- Animator deletes that copy, saves their file, and later re-opens it. Only one \"layer\" of unused objects has been cleaned, so there can still be orphan object datablocks lying around unseen.\n- Animator duplicates the character again. Now the number suffix of their objects are all over the place. Some of them have .001 and others have .002.\n\nAs they continue working and duplicating characters, the old, unused objects slowly get deleted, and the name suffixes really get all over the place. It's not nice to get your objects names all mixed up like this, as things get really confusing when trying to troubleshoot problems in a production file, since everywhere where one object references another, the mismatching suffixes give the troubleshooter anxiety and uncertainty. The only way to avoid this currently is to tell the entire production team to always run a recursive Purge after deleting anything, which is very error-prone. \n\n\n### Potential issues\n\n- ~~The current recursive purge purges Text datablocks, which is bad because they don't have the possibility of a \"Fake User\", so they should be excluded (or better yet, be included in the \"fake user\" system, and get a fake user by default). ~~ (Fixed by [D10983](D10983))\n- Actions don't have fake user enabled by default. The long standing issue of Action datablocks \"disappearing\" on users and making them lose hours of work if they forget to click Fake User would be exacerbated by this change. This could be trivially fixed by giving Actions a fake user on creation, and forcing users to clean up their unused Actions themselves.", "Blend File Format: Fix using `0` as `SDNAnr` for non-SDNA data ('raw' arrays of bytes)\nCurrently when writing an allocated arrays of chars, ints etc., functions like `BLO_write_string` or `BLO_write_int32_array`. Those end up calling `writedata`, which sets the `SDNAnr` of the written `BHead` to `0`.\n\nThis value happens to be the DNA index of an actual struct (currently `DrawDataList`)... Among other potential problems, it confuses our `blendfile` python module when accessing such pointer, since it tries to recreate a `DrawDataList` from some totally unrelated data.\n\nProposal is to either use `INT_MAX` as `SDNAnr` for raw bytes blocks, or maybe even have a set of values covering the basic types (int, char, pointer, etc.).\n\nAnother idea would have to create 'dummy' DNA struct for those basic types, although it would probably be problematic to deal with alignment constraints then?", "Segfault in CustomData_blend_read\nOperating system: Debian Linux, Stable\nGraphics card: GTX970\n\nBroken: 3.6.0 Alpha, branch: main (modified), commit date: 2023-04-05 18:55, hash: 92c9c1d4003e, type: Release\nWorked: Unsure. Old blend I loaded and worked on for a while. Errors on open now.\n\nSegfault on load.\n\n1. Load blend, or open blender and load blend\n\n(No idea what the gvfs error is.)\n```\nFailed to open dir (No such file or directory): /run/user/6001/gvfs/\n[New Thread 0x7fffe3ae9700 (LWP 2310937)]\n...\n[New Thread 0x7fffdb9fe700 (LWP 2310945)]\nRead blend: \"/tmp/segfault.blend\"\n\nThread 1 \"blender\" received signal SIGSEGV, Segmentation fault.\n0x0000000001009af9 in CustomData_blend_read (reader=0x7fffffffd498, data=0x7fffd9c471f0, \n count=5002)\n at /opt/src/blender-git/blender/source/blender/blenkernel/intern/customdata.cc:5194\n5194 if (layer->flag & CD_FLAG_EXTERNAL) {\n(gdb) l\n5189\n5190 int i = 0;\n5191 while (i < data->totlayer) {\n5192 CustomDataLayer *layer = &data->layers[i];\n5193\n5194 if (layer->flag & CD_FLAG_EXTERNAL) {\n5195 layer->flag &= ~CD_FLAG_IN_MEMORY;\n5196 }\n5197\n5198 layer->flag &= ~CD_FLAG_NOFREE;\n(gdb)\n```\n\n", "Assets: the same catalog UUID appearing in different asset libraries causes failed assertions in the 'All' library\nOperating system: Linux-5.15.0-73-generic-x86_64-with-glibc2.35 64 Bits, X11 UI\nGraphics card: NVIDIA GeForce GTX 1080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 470.182.03\n\nBroken: version: 4.0.0 Alpha, branch: main, commit date: `Mon Jun 5 14:21:39 2023 +0200`, hash: `c18ee2e84dcebaf5831640c987e5fd434219e46c`\nWorked: probably broken since the introduction of the 'All' library\n\nHaving two asset libraries that both have a catalog with the same UUID cause an error:\n\n```\nBLI_assert failed: blender/source/blender/blenlib/BLI_map.hh:1093, add_new__impl(), at '!this->contains_as(key)'\n```\n\n\n- Extract the attached ZIP file. It contains a blend file and two asset libraries.\n- Run Blender in debug mode.\n- Open the blend file.\n- Run the script in the text editor (which adds the two asset libraries to your preferences; don't worry, they won't be saved)\n- Refresh the 'All' asset catalog.\n- See the failed assertion\n\n**Workaround**\n\nApplying this patch will work around this issue:\n\n```diff\ndiff --git a/source/blender/asset_system/intern/asset_catalog.cc b/source/blender/asset_system/intern/asset_catalog.cc\nindex d27763c08d7..8512824b6d4 100644\n--- a/source/blender/asset_system/intern/asset_catalog.cc\n+++ b/source/blender/asset_system/intern/asset_catalog.cc\n@@ -686,7 +686,7 @@ static void copy_catalog_map_into_existing(const OwningAssetCatalogMap &source,\n {\n for (const auto &orig_catalog_uptr : source.values()) {\n auto copy_catalog_uptr = std::make_unique<AssetCatalog>(*orig_catalog_uptr);\n- dest.add_new(copy_catalog_uptr->catalog_id, std::move(copy_catalog_uptr));\n+ dest.add(copy_catalog_uptr->catalog_id, std::move(copy_catalog_uptr));\n }\n }\n \n```\n\nA better solution would be to implement some comparison function like `bool AssetCatalog::is_consistent_with(const AC &other);` that tests both the UUID and the catalog path are identical, and issue a warning if they are not, similar to what `AssetCatalogService::parse_catalog_file()` is doing.\n\n", "Possible 4.0 Top Menu Changes\nI'm hoping that with Blender 4.0 we can take the chance to move some menu items slightly.\n\nOne concern I'd love to address is the discoverability of the \"app\" (Blender logo) menu. I'm sure most users find it eventually, but that won't happen be immediately for new users.\n\nI would address this simply by moving out of that menu anything that a new user might need, and leave it for things that more advanced users use. This moves \"Splash\" and \"About\" items to the Help Menu. The \"System\" submenu expands out. \"Install Application Template\" moves to File Menu. This leaves it with only items that most new users can safely ignore at first: \n\n\n\nMoving \"Install Application Template\" to the \"New\" submenu makes it more clear that you can add to the list:\n\n\n\nWith the \"File\" menu it would be nice to group the \"Loading\" and \"Saving\" items together. This separates \"Import\" and \"Export\". These two items have contents that look almost identical so it is easy to select an import instead of an export and cause damage. This also makes the list shorter so that it can better accommodate #104678.\n\n\n\nFor the Edit menu we can group the items slightly better by making \"Undo History\" a part of the Undo/Redo group. And also make it clear that \"Adjust Last Operation\" is the same that is done when \"Repeat Last\" and is part of repeat history:\n\n\n\nI'd like for the Help menu to look a bit simpler by following our existing recommendations of not showing repeating item icons. And also move the Splash and About here:\n\n\n\nI have a sample PR that can be tested here: 106654\n\n", "Hidden links\n* - [ ] Allow hiding and showing links.\n* - [ ] Show some icon when a link is hidden (see image below).\n* - [ ] Show hidden links that are connected to a selected node or a node within a selected frame.\n* - [ ] When the link is coming from a named reroute socket, display the name on the receiving side (still needs to be explored more). \n\nThis is also related to #93291, which will make it much easier to create long-distance links.\n\nOriginal exploration: 21306\n\nProposal from @Erindale:\n", "Experiment: Compositor meta data to support different workflows.\nTo test a new workflow at the blender institute we want to use the same meta data as stored in the input file.\nThe compositor already has the ability to keep the meta data when using cryptomatte which we want to use.\nThis is just an experiment and not intended to land in blender master ATM.\n\n\nMetadata for single layered textures are stored as `ImBuf.metadata` (Properties)\nI expect that in `void OutputSingleLayerOperation::deinit_execution` we could fill them with information that we get from COM_MetaData.\nWe should then also fill the meta data in `ImageOperation`.", "AssetBrowser: Add cleanup operator for asset index.\nUsers can rename/relocate/remove asset libraries. When they do created indexes would lead to files that will not be used anymore. Having a cleanup operation would help the user to free space on their storage device.\n\n## Approach\n\nUsers can have multiple blender versions installed with different asset libraries configured. The challenge here is that you when cleaning up automatically you can clean up indexes that are still used.\nThe approach to help the user would be to have an operation with an option to filter what indexes should actually be cleaned.\n\n### Requirements\n\n- User should be able to remove all indexes [OP1].\n- User should be able to only remove indexes that aren't used by the current open blender version [OP2].\n- User should be able to remove indexes of a specific asset library [OP3].\n\n## Design questions\n\nThere are 4 locations where such an operation makes sense:\n- **Blender Icon->System menu**: Current cleanup operation works on the opened blend file, very hidden.\n- **File -> Cleanup menu**: Currently only shows cleanup operations on the opened blend file. [OP1, OP2]\n- **Blender Preferences -> File Paths -> Asset Libraries**: The cleanup operations we want to introduce works on multiple asset libraries so it might be a good location. Concern is that it is hidden far away and is very specific. [OP1, OP2, OP3]\n- **Asset Browser Space** -> Only shows/works on a single asset library. Adding a cleanup operation here might be confusing, or only when it targets the current index. What is already done automatically, so might not make much sense. [OP3] might fit here.\n\n| **System Menu** | **Operation Fit OP1** | **Operation Fit OP2**| **Operation Fit OP3** |\n| -- | -- | -- | -- |\n| Blender Icon | Not very discoverable| Not very discoverable| Not very discoverable |\n| File menu | Ok fit | Ok fit | Nok fit as the current library could not be visible in the current screen |\n| Preferences | Ok fit| Ok fit| Ok/Nok fit, when user actually want to use it the operator is far away |\n| Asset Browser | Nok | Nok | Ok |\n\nLooking at the table above I would propose to add OP1, OP2 to the preferences and OP3 to the asset browser space inside the edit menu.\n" ]

[ "Improve handling of embedded data-blocks (root nodetrees and master collections)\nEmbedded data-blocks often require specific handling in ID management (freeing, duplicating, referencing, RNA paths, etc.).\n\nCharacteristics of embedded IDs:\n* They have one and only one owner ID.\n**They are flagged as such with `LIB_EMBEDDED_DATA`.** This owner ID takes care of their management (creation, duplication, deletion...).\n** IDTypeInfo provides a way to get that unique owner ID from an embedded one.\n* They are not stored in Main database.\n**However, they can be referenced by other IDs *(e.g. master collection of scene can be a parent of many other collections)*.** New embedded IDs are created with `LIB_TAG_NO_MAIN`, however existing old ones may not have it, and there is no `do_version` to set it on those...\n* They do have their own animation data.\n* Their name is not unique.\n* They do have a valid `session_uuid`.\n* They are stored in .blend files as 'regular' data, they do not have the `bhead->code` set to their ID type.\n\nProposals for Improvements\n****\n\nHere are ideas to help improve handling of the embedded IDs. \n\n1 - Storing a reference to owner ID in embedded ones\n----------------------------------------------------------------\n\nWhile 'loop-back' ID pointers can be problematic, their support in general ID management code is now fairly safe and sound. This would solve the current issue where we have to loop over all potential owner IDs to find the proper one.\n* - [x] Store that `id_owner` pointer in `ID` struct itself, or in data of each specific ID type? [D15838](D15838), cd49fee741\n** *~~Would rather go for generic pointer in `ID` struct.~~*\n\n2 - Tags and Library of embedded IDs should match their owner's\n-----------------------------------------------------------------------------\n\nEmbedded IDs should share the same 'ID management' tags as their owner ID (if owner ID is `LIB_TAG_NO_MAIN`, its embedded ones are too, etc.). Embedded IDs are not stored in `G_MAIN`, but are often handled as such, can be referenced by other regular IDs (like master collection being parent of other collections), etc. Similarly, they should also always share the same Library pointer as their owners.\n* - [ ] Needs to ensure tags are in sync with their owners (need to check exactly which ones need it, document it, and probably add an API to set tags that ensures this behavior). *This should wait for #88555/T90610 to be addressed.*\n* - [x] Needs to ensure lib pointers are always in sync. //Seems to already be the case, 788952705c just adds some extra asserts to double-check that on file read.\n\n3 - Generic 'foreach' looper over embedded pointers only\n-------------------------------------------------------------------\n\nThis could help making some of IDManagement related tasks more generic when dealing with embedded pointers, needs to be investigated further.\n\nThere can be two approaches:\n - Add an extra flag option to the existing `foreach_id` API (aka `BKE_lib_query`). This imply `IDTypeInfo.foreach_id` callbacks of types that may own embedded IDs to be aware of this option, to optimize their code by only processing those embedded pointers. *Least effort solution, but requires exposing some info from generic foreach_id code into types callbacks. not necessarily a bad thing though.*\n - Add a new dedicated callback to `IDTypeInfo`, that would be somewhat the complement to the existing `owner_pointer_get` one.\n\n-----------------------\n*Original task (note that since 2019, we clearly moved towards the second solution described below):*\n\n`bNodeTree` is currently both a datablock, and embedded in materials. Same for collection and master collection of a scene.\n\nThis leads to messy datablock management and problems creating RNA paths (e.g. #68971).\n\nA solution to this could be to make `bNodeTree` node longer a datablock. Instead it would be embedded in `Material`, `World`, `Light`, and a new `NodeGroup` datablock.\nThis will require Python API breakage and some relatively complicated versioning, but the resulting design should be cleaner.\n\nAnother solution is to design a proper handling of those private data-blocks all around our code base (ID/Lib management in BKE_library area, RNA handling of IDs, etc.). We already do that to some point, but so far it has mostly been hacks around issues on a case-by-case basis.\nSolution to #68971 already added a new ID flag to mark those private datablocks, e.g. we could extend RNA handling of IDs to detect that one and automatically consider those structs as the other private sub-data ones (like e.g. render settings from scene, etc.)." ]

[ "Can't resize node in visible bounds\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 536.40\n\nBroken: version: 4.0.0 Alpha\n\nWhen trying to resize a node, the node cannot be resized if the mouse is precisely on the edge of the node\n\nAdd any node in shader o geometry node editor and try to resize the node with the cursor in the border line of the node\n<video src=\"attachment\" title=\"2023-07-11 14-28-46.mp4\" controls></video>\n\n", "Bug with inconsistency combining workspaces and main windows (it happens in ALL BLENDER VERSIONS)\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 770/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.71\n\nBroken: version: 3.0.0\n\nWhen working with more than 1 main window of the same Blender project, if new workspaces (Layout, Sculpting, Scripting) are added in one window, they are not properly displayed on the other. (See image for clarification)\n\nImportant! I have tried to reproduce the steps with the latest versions: 2.83.2, 2.93.9, 3.1.0 beta, 3.2.0 alpha (and older 2.93 and 2.83 installed versions) and the bug is the same in all of them. It seems to be a bug that has always existed.\n\n\n1. Open a Blender project (no matter if it's new or already-existing one).\n2. Add a new main window (Window > New Main Window)\n3. Once both windows are open, add any new workspace on one of them - let's say it's WINDOW1 (+ Add Workspace > General > Scripting)\n4. Go to WINDOW2 and click on the new workspace 'Scripting' Tab that has appeared. See that the workspace is corrupted as it's not displaying the \"Scripting\" workspace.\n\nI have replicated the error myself on the attached Blender file:\n\n[Blender workspaces bug.blend](Blender_workspaces_bug.blend)\n", "Allow node inputs named A and B to be swapped\nPossible fix to allow node inputs to be swapped.\n\nThis is defined in the `node_link_socket_match` function.\n\nReported in chat by @SimonThommes \n\nIt may be a better solution to add a swappable flag to sockets.", "Node Editor: Improve working with frame nodes\n### Problem\nWorking with frame nodes works well for most simple cases, but there are a few things that are confusing or cumbersome:\n* A node sitting on top of a frame is not necessarily part of the frame but there is no visual indication ([example video](attachment))\n* Which frame a node is being added to during transform is based on the position of the mouse cursor and can lead to unexpected results ([example video](attachment), also see issue #80218)\n* Nodes not being able to be removed from frames, slows down the workflow, since the user has to stop and explicitly detach the node\n\n### Proposal\nAfter testing a few different variations for how this could work ([together with users on the forums](28241)), this is what seemed to work best overall:\n\n1. Nodes should always be attached to the frames that include them\n * This reduces ambiguous situations, where nodes are on top of a frame without being part of it\n * This behavior is more predictable than the cursor based attachment logic we have now\n2. Allow detaching nodes during transform with a hotkey\n * Pressing `Shift` during transform to freely drag nodes out of a frame - even if \"Shrink\" is enabled (`Shift` is consistent with the re-parenting shortcut in the outliner)\n3. Allow adding/removing nodes from a frame by resizing it\n4. Additionally tweak node resizing, since it will be used more often:\n * Allow moving the frame during resizing by pressing `Space` (same as box selection, [demovideo](attachment))\n * Improve hitzone for resizing to make it more accessible\n\n### Alternative: Remove \"parenting\"\n\nThe most promising alternative I tried was to remove parenting altogether. The premise was that frames simply always affect the nodes that are inside them.\n\n**What worked well:**\n* No ambiguity about whether a node is inside a frame or not\n* Users seemed to like the ability to add nodes by resizing the frames itself (which has been included in the current proposal)\n\n**What didn't work:**\n* Users didn't seem to like how multiple frames were able to affect one node at the same time:\n * In some situation it would lead to partially overlapping frames \"stealing\" nodes from one another\n * Overall users feared that it would be easy to create situations that are hard to untangle\n\nSince \"node stealing\" is inherent to this approach, I settled on the proposed compromise that keeps the \"one node, one frame\" concept but reduces some of the quirqs.\n\n---\n\n### Prototype Patch\nA pull request implementing the proposed changes for testing can be found here: 105457\n\n---\n### Steps for implementation\nThe prototype patch I've used for testing can be broken up into a few smaller changes:\n\n- [ ] Always add nodes to the frames they overlap\n- [ ] Improve resizing of nodes:\n - [x] Hitzone for node resizing\n - [ ] Allow moving nodes during resizing\n - [ ] Respect snapping during resizing (I haven't actually implemented this in the prototype)\n\nThere are a few more things that are not required for this but might be nice:\n- Improve how the parenting of nodes to frames is propagated when deleting or grouping nodes.\n- Remove the \"parent space\" positioning of nodes so the node location is always in view space. I think this could also help removing the need for `offsetx` and `offsety`.\n\n", "Rendering with F12 does not consider the different resolutions in each scene\nOperating system: macOS-10.13.6-x86_64-i386-64bit 64 Bits\nGraphics card: ATI Radeon HD 5750 OpenGL Engine ATI Technologies Inc. 4.1 ATI-1.68.25\n\nBroken: version: 3.4.0 Alpha\nWorked: I don't know, but even it never worked, it doesn't mean it is correct :)\n\nRender (F12) does not respect **Render Layer** node dimension\n\nI have two scenes with different dimension\n- Scene 1080 px - for rendering scene\n- Scene 2160 px - for compositing scene 1080 px\n\nScene 1080 rendered by pressing render icon at **Render Layer** node result correct with expected mix in compositor of Scene 2160 ...\n\n\n\n... but rendering by pressing **F12** in Scene 2160 renders Scene 1080 also in 2160 px (so compositor fails in result).\n\n\n\nFor more info or sample file see 2214\n\nI hope it is would be something small to fix. \nThank you for your time.\n\n- Open attached file\n- Compare rendering with F12 (which should render the two scenes at different sizes) with rendering using the render button on node\n[bse275864_OneScene.blend](bse275864_OneScene.blend)", "Wrong result mixing 2 nodes of an image texture\nOperating system: Windows 10\nGraphics card: AMD\n\nBroken: 3.3, 2.93\nWorked: unsure\n\nWhen using 2 nodes of the same texture and connecting the Color output from one and the Alpha from another, the result looks different in comparison when using both outputs from one node. Broken in EEVEE\n\n1) Open the file, switch to material preview/rendered\n2) Reconnect the color (or the alpha) link to another texture\n[Bug1.blend](Bug1.blend)", "Cycles: Bump node doesn't generate anything in certain conditions\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3090/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 522.25\n\nBroken: version: 3.3.0\nWorked: all versions from 2.79 to 3.4.0 alpha are affected, there is no working build\n\nI'm trying to generate a normal map using Bump node and color output of the env map as height map, it doesn't work on Cycles while it works on Eevee. Seems like the \"reflection\" output of Texture coordinate node is broken, but it works fine when I'm using another outputs though\n\nOpen the project, then click on viewport Rendered mode button - it'll start Cycles which for some reason can't generate bump map, then switch to Material preview mode - it'll start eevee which works as expected\n[bug_report.blend](bug_report.blend)\n\nEdit: typos and bad english grammar fixed, I believe", "Add option for Render Layers node to render in native scene resolution\nOperating system: 64bit Windows 10\nGraphics card: RTX 3090\n\nBroken: 2.93.6; 3.6.0\nWorked: - \n\nWhen using a file output node to save different scenes in different formats while rendering an animation, the file will always be in the format dimension of the currently active scene.\n\n**Possible overlapping with:**\n87447\n\nYou use the provided sample files and press render. \nYou will need to navigate to the set output paths in the Compositing node. \nC:\\OutputTest(3.6) and C:\\OutputTest_293(2.93)\n\n**Replicate yourself:**\n\n1. open default scene\n2. duplicate scene using \"Linked Copy\"\n3. in freshly duplicated Scene, change the \"Format\" > \"Resolution\" to 9:16 1080x1920\n4. open Compositing > \"Use Nodes\"\n5. add a \"File Output\" node > select it\n6. expand \"Properties\" and create a second output using \"Add Input\"\n7. duplicate \"Render Layers\" node and select the other scene in the duplicated \"Render Layers\" node\n8. plug each \"Render Layers ⇾ Image\" into one of the inputs of the \"File Output\" node\n9. now press Render\n\n**Expected Result:**\nEvery image is saved by the File Output node in its own resolution, respective of the parent scenes \"Format\" settings.\n\n**Actual Result:**\nAll images have the resolution of the currently selected Scene. \n\n**2.93.6 File**\n[OutputNode_Test_293.blend](attachment)\n**3.6.0 File**\n[OutputNode_Test_36.blend](attachment)\n\n", "Node sockets disappearing but still dragging works\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1650/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.93\n\nBroken: version: 3.4.0 Alpha\n\nWhen a node is added and zoomed out, the node sockets disappear. But still, the drag function works.\n\n[Node Socket Bug Report .mp4](Node_Socket_Bug_Report_.mp4)", "Ghost / False Node inputs\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.40\n\nBroken: version: 3.1, 3.2, 3.3\n\nWhen the output type of a node changes:\n\nVector Math multiply (vector output) ----> Vector Math distance (float output)\n\nIt causes the node that has the changing type as it's input to be in a bugged state.\n\n* Open .blend file\n[ghost_inputs.blend](ghost_inputs.blend)\n* Change the multiply to distance (easiest one I could think of)\n* The other math node will be in a messed up state\n\n* You can even drag from the ghost input and you will be able to see that Blender thinks it is still connected.\nThanks!", "Pin Scene to Workspace doesn't work (when multiple blender windows are open)\nOperating system: Linux-5.15.0-53-generic-x86_64-with-glibc2.35 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 470.141.03\n\nBroken: version: 3.4.0 Alpha\nBroken: version: 3.3.0\n\n\nPin Scene to Workspace doesn't work\n\nOpen file, observe that scene.002 is active and pined.\n\n[switch scenes.blend](switch_scenes.blend)\n\nSwitch to Modeling tab then back to Layout.\n\nObserve that instead of scene.002, scene.001 is active. \n\nNOTE: this is only failing with multiple blender windows open (if the second window is closed, it works as expected)", "Texture Node Editor Using nodes causes weird issues in sculpt mode.\nOperating system: Linux-5.15.0-48-generic-x86_64-with-glibc2.31 64 Bits\nGraphics card: NVIDIA GeForce RTX 2060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 515.76\n\nBroken: version: 3.3.1\nWorked: perhaps never? (can redo in 2.83)\n\nAdding an image from the nodes of the texture editor for example, a sculpting brush, causes really weird issues. Disabling nodes and using the texture properties does not have this issue.\n\n- Check the blend file. Try sculpting, you will see weird noise being displaced with the texture.\n- Now disable the nodes and set the texture type from the texture properties to the same image. Sculpting again you see the issue is not happening.\n[bug.blend](bug.blend)\n\n\n\n", "Changing UI resolution scale messes up nodes snapping\nOperating system: Windows-10-10.0.25393-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2080 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 536.23\n\nBroken: version: 4.0.0 Alpha\nWorked: Never AFAIK\n\nWhen you have a node tree all snapped to grid and is perfectly organized, you move it and it all moves in an organized manner. But say you change your UI resoltuion scale, or line width; the perfectly organized node tree is not organized anymore 🥲. Now it may seem that this bug is really rare, but it's really infuriating when you open someone's file and you mess up the node tree in seconds by just moving.\n\nI created a small video that demonstrates the bug, but basically the steps are.\n1. Create a well-organized node tree that is snapped to grid.\n2. Change the UI resolution.\n3. Observe your node tree not snapped to grid anymore.", "Thread race condition with lock interface enabled\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 461.92\n\nBroken: version: 2.93.2 Release Candidate\n\n**Description**\nUI redraw is still taking place in the render_init and render_complete handlers when use_lock_interface = True. This lock was only implemented for 3D viewport, but this is not mentioned in any of manuals.\n\nAdding lock to node editor seems to resolve the issue:\n\n```\ndiff --git a/source/blender/editors/space_node/space_node.c b/source/blender/editors/space_node/space_node.c\nindex ff848a7bb95..9813b8aaa86 100644\n--- a/source/blender/editors/space_node/space_node.c\n+++ b/source/blender/editors/space_node/space_node.c\n@@ -1073,6 +1073,7 @@ void ED_spacetype_node(void)\n art->cursor = node_cursor;\n art->event_cursor = true;\n art->clip_gizmo_events_by_ui = true;\n+ art->lock = 1;\n\n BLI_addhead(&st->regiontypes, art);\n```\n\nTrace: \n```\n>\tblender.exe!UI_block_end_ex(const bContext * C, uiBlock * block, const int * xy, int * r_xy) Line 1873\tC\n\n \tblender.exe!node_draw_basis(const bContext * C, const View2D * v2d, const SpaceNode * snode, bNodeTree * ntree, bNode * node, bNodeInstanceKey key) Line 1733\tC++\n \tblender.exe!node_draw_default(const bContext * C, ARegion * region, SpaceNode * snode, bNodeTree * ntree, bNode * node, bNodeInstanceKey key) Line 1943\tC++\n \t[Inline Frame] blender.exe!node_draw(const bContext *) Line 2002\tC++\n \t[Inline Frame] blender.exe!node_draw_nodetree(const bContext *) Line 2059\tC++\n \tblender.exe!draw_nodetree(const bContext * C, ARegion * region, bNodeTree * ntree, bNodeInstanceKey parent_key) Line 2097\tC++\n \tblender.exe!node_draw_space(const bContext * C, ARegion * region) Line 2235\tC++\n \tblender.exe!ED_region_do_draw(bContext * C, ARegion * region) Line 569\tC\n \tblender.exe!wm_draw_window_offscreen(bContext * C, wmWindow * win, bool stereo) Line 726\tC\n \tblender.exe!wm_draw_window(bContext * C, wmWindow * win) Line 868\tC\n \tblender.exe!wm_draw_update(bContext * C) Line 1067\tC\n \tblender.exe!WM_main(bContext * C) Line 654\tC\n \tblender.exe!main(int argc, const unsigned char * * UNUSED_argv_c) Line 563\tC\n\n```\n\nOpen below file, run the script and then press f12.\n\n[ui lock ignored on some handlers which are not on the main thread.blend](ui_lock_ignored_on_some_handlers_which_are_not_on_the_main_thread.blend)\n\n", "Geometry Nodes: Color mix nodes don't effect the alpha\nOperating system:\nGraphics card:\n\nBroken: 3.4.1\n\nwhenever I try doing any math with colors, the alpha isn't affected when I use the color mix node in any blend mode. \n\n1. Open blender geometry nodes.\n2. Add 2 combine color, color or whatever node with different different values for the alpha's.\n3. Use a mix color node with any operation like add, subtract etc.\n4. See that the alpha didn't change (use the spreadsheet or the hover over)." ]

[ "Selecting (or connecting) nodes in Compositor is malfunctioning when when monitors have different dpi\nOperating system: Windows-10-10.0.16299-SP0 64 Bits\nGraphics card: GeForce GTX 1080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 431.86\n\nBroken: version: 2.83 (sub 5)\n\nSelecting nodes in Compositor is glitching when when monitors have different dpi. This happens when Compositor is opened on both monitors.\n\n - In Windows 10 set custom scale on the second monitor in Display Settings > Scale and Layout, for example to 175%, so that each monitor has different scale.\n - Put one Blender window on each monitor (same Blender session).\n - Open Compositor in both windows. Enable nodes.\n - In the window with scale equal to 100% try clicking on the Composite node couple times. Selection will switch to the other node.\n[custom_dpi_compositor_bug.blend](custom_dpi_compositor_bug.blend)\n[custom_dpi_compositor_bug_02.03.2020.mp4](custom_dpi_compositor_bug_02.03.2020.mp4)" ]

[ "2.83lts Regression: Viewport render image ignores colormanagement, and appears to be hardcoded to use sRGB.\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 445.87\n\nBisect points to 190fd795a9\nBroken: version: 2.83.0\nWorked: 2.82a\n\n`Viewport Render Image` ignores color management. The image produced is clamped to 1.0 according to the color sampler. It appears to be hard coded to sRGB.\n\nCreate a Scene that has scene linear color values above 1.0. Render with Filmic. Then do `viewport render image` and see how it produces an image with blown out highlights.\n\n2.82a:\n\n\n2.83:\n\n\n\n[test.blend](test.blend)\n\n", "Cycles does not generate the exact same images when a scene is rendered twice\nOperating system: Ubuntu 20.04\nGraphics card: Nvidia 2070 Super\n\nBroken: master\nWorked:\n\n\nThe results generated by Cycles are not 100% deterministic.\nAs a consequence path guiding ( #92571 ) can not be implemented deterministically.\n\n\n1. Start Blender and open a scene like `monster`\n2. Render the scene with 64spp and store the result as exr image (e.g., monster-run-0.exr).\n3. Repeat 1. and 2. and save the result as exr again (e.g., monster-run-1.exr).\n\n4. Use an image comparison tool such as tev (tev) and compute the difference. \n\nYou will see that, even if both renderings were performed on the same machine, the resulting images have minor differences.\nNote: It might be necessary to scale the diff-images to see the errors.\n\nRun 0:\n\nRun 1:\n\nDiff:\n\n\n\n\n\n\n \n\n\n \n\n \n\nBased on the default startup or an attached .blend file (as simple as possible).", "Cycles: Attribute node renders black when some reserved standard attribute names are used\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: Radeon (TM) RX 480 Graphics ATI Technologies Inc. 4.5.14760 Core Profile Context 20.45.37.01 27.20.14537.1001\n\nBroken: 2.79 - 2.93.0 Alpha\nWorked: Never?\n\nIn Cycles, the Attribute node returns black when an attribute name is reserved (such as \"color\" or \"temperature\").\nThe attribute can be created from the attribute panel or through Geometry Nodes.\n\nSeems like any these attribute names could cause the issue:\nattribute.cpp#L320-L392\n\n- Open attached file\n- Display render preview (note that the object is black)\n- Change the name on the node attribute and Vertex Color to \"color2\" (or switch to node having \"temp\" attribute)\nNow the node attribute works\n\n[vcol_cycles_name_bug.blend](vcol_cycles_name_bug.blend)\n[temperature_attribute_shader_handling.blend](https://projects.blender.orgattachment)\n", "\"Regression\" : Using normal or bump maps in cycles produces flat washed-out shading (caused by terminator softening)\nOperating system: Windows 10 x64\nGraphics card: GeForce GTX 1660 Dual OC\n\nBroken: 2.81, 2.82, 2.83 from 2020-03-06 22:46\nWorked: 2.80\n\nCaused by c3be14e151\n\nUsing normal or bump maps in cycles produces flat washed-out results comparing to EEVEE.\nThe bug also appears in the latest 2.83 build, it affects both Diffuse and Principled Shader but not Glossy.\nNormal Map and Bump nodes seem to produce identical results with the 2.80 version, so the problem lies with the Shader node itself.\n\n\n[bug_report.blend](bug_report.blend)\n1. Open file\n2. Switch to Rendered viewport shading", "EEVEE-Next: Refactor renderpass storage\nCurrently render-passes are allocated in multiple texture buffers.\nThey take a lot of binding points and currently prevent the deferred pipeline to output diffuse color correctly.\n\nThe plan is to combine all render-passes into the same array texture and use an associated UBO containing offset to the layer of each pass type.\n\nThis would very similar to the AOV support.\n\nAnother direct benefit of this would be that the GBuffer encoding could be simplified.\n\n", "Cycles Image Texture Alpha conversion inconsistency with EEVEE and Image Editor\nOperating system: Windows 10 Pro 64-bit (10.0, Build 18363)\nGraphics card: NVIDIA GeForce 920M (Laptop)\n\nBroken: v2.83.13 Beta, 2020-04-21\n\nIf an image is packed, it will get displayed differently in Cycles and EEVEE. It changes in EEVEE when\nAlpha Mode is changed from Straight to Premultiplied to match Cycles, but it does not change in Cycles\nby changing that option. If the image is unpacked Cycles will look like EEVEE with Straight Alpha Mode\nand again it will not be changed by the Alpha Mode option like EEVEE is.\n\n1. Open the provided file.\n2. Select the Image Texture Node\n3. Go into the right panel in the Shader Editor (N-panel) into the Item tab and Properties subsection\n4. Render Image\n5. try changing Alpha Mode and rerendering.\n6. try unpacking the image and rerendering.\n7. try changing the Alpha Mode on unpacked image and rerender.\n[packed_image_cycles_display_inconsistency.blend](packed_image_cycles_display_inconsistency.blend)\n", "Data Transfer modifier works only for Byte Color & Face Corner color attributes\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3060 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 472.12\n\nBroken: version: 3.2.0\n\nIf a target color attribute Domain is *Vertex* or it's Data Type is *Color*, the data are not transferred.\n\n[data-transfer-color-attr-bug.blend](data-transfer-color-attr-bug.blend)\nYou can see in the attached .blend that the Data Transfer modifier is set correctly, but the Col attribute isn't updated.\nAfter doing Convert Attribute -> Domain: *Face Corner*, Data Type: *Byte Color*, it starts to work correctly.\n\n\n", "Cycles vertex color bump map banding \nOperating system: Win10\nGraphics card: GTX 1070\n\n2.79b release\n\nUsing vertex colors in cycles material as a mask for bump/normal texture causes banding/hard edges in render.\n\nBanding of vertex color is also visible when using it as a color for shader. \n\nIt might have something to do with the fact that vertex colors are represented in 8-bit sRGB values but on the other hand it renders as expected when using it as a mask for example for mix shader node. But the bump doesn't look correct in that case.\n\n\n[vertex_color_mask_problem.blend](vertex_color_mask_problem.blend)", "VSE: Alpha Over + Convert to float changes the color of the top layer\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: Intel(R) UHD Graphics 600 Intel 4.5.0 - Build 27.20.100.8853\n\nBroken: version: 2.92.0 Alpha\n\nAdd two color strips - not black, and change compositing of the top strip to Alpha Over. Toggle 'Convert to float' under color while pressing arrow left/right to update the preview. The top layver will now change color, it's a regression and a bug. \n\n", "Theme: Inconsistent solid color for the background in the 3D viewport\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1070 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 445.75\n\nBroken: version: 2.90 (sub 0)\nWorked: 2.79b\n\nThe color of the 3D viewport background set to theme seems to not match with the color set in the theme setting.\n\nCheck the examples below with the color #36393F\n\n2.79b: color is exactly like in the theme setting.\n\n\n2.80 to 2.82: color looks good but seems grainy.\n\n\n2.83 to 2.90: color is slightly changed, when sampled not the expected theme color.\n\n\n- Open blender\n- In the initial layout change the Outliner editor to Preferences\n- Go to background color option\n- Use the colorpicker to get a sample of the background color\nThe color shouldn't change, but it does", "Vertex paint + x-ray + polygon selection mode = weird shading\nOperating system: Windows-7-6.1.7601-SP1 64 Bits\nGraphics card: GeForce GTX 660 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 441.20\n\nBroken: version: 2.82 (sub 6)\n\n\nFor me it looks like this\n\n\n\n[vertex paint x-ray strange shading.blend](vertex_paint_x-ray_strange_shading.blend)", "Color management look contrasts apply the wrong gamma when the display device is set to None\nOperating system: window 11\nGraphics card: RTX 2060s\n\nBroken: 3.1 stable, 3.0 stable\n\nWhen the display device is set to None, applying a look contrast results in what looks like a double gamma correction. This doesn't happen when the Display device is set to XYZ.\n \n1. set the display device to None\n2. set the look contrast to any other setting than None or Medium", "Grease pencil eyedropper doesn't sample correct vertex color.\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce RTX 2070 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 457.30\n\nBroken:\n```\nversion: 2.90.0\nversion: 2.92.0 Alpha\nversion: 2.83.9\n```\n\nWorked: (unknown maybe never)\n\nWhen using the grease pencil eye dropper with filmic color management selected; the sampled color is different than the original vertex stroke color. \nWorks as expected with standard color management selected\n(Note this is confusing to new users as they expect to color sample and return the previous color of their previous stroke.)\n\n\n - Start with a new file\n - Set color management to filmic.\n - Add grease pencil object\n - Enter Draw Mode\n - Paint in vertex color mode using a lighter color.\n - With a blank palette use the color eyedropper tool on the stroke you just made\n - now paint with the sampled color in the palette\n - result should be slightly darker than the first stroke.\n\nvideo reference:\n**with filmic enabled:**\n[Desktop 2020.11.25 - 02.35.06.03.mp4](Desktop_2020.11.25_-_02.35.06.03.mp4)\n\n**with filmic disabled:**\n[Desktop 2020.11.25 - 02.36.30.06.mp4](Desktop_2020.11.25_-_02.36.30.06.mp4)\n\n\n", "Paint Mode: Color Fill Tools\nNOTE: This design is still under discussion and is incomplete.\n\n# Issue\n\nCurrently there are different implementations of filling a solid color while in a mode:\n- Fill Tool (Texture Painting)\n- Fill operator (Vertex Paint Mode)\n- Color Filter (Sculpt Mode)\n\nThis needs to be reworked for consistency and to ensure more convenient and efficient workflows.\n\n# Requirements\n\nThere are certain broader features that will be required, no matter how the design of color filling looks like:\n- Fill everything that is not masked\n- Gradual Fill (Not full strength)\n- Fill via gestures (Box, Lasso, Line, Polyline)\n- Fill entire UV-islands, mesh-islands, face sets\n- Fill faces instead of pixels\n\nThere is also the common use of gradients and expand. This should be further notes down in #99770.\n\n# Goal\n\n### Key Problem\n\nThere are many different ways to give all of these tools to the user.\nWhat we should avoid is to have a large set of tools or tool settings scattered throughout the UI.\nThis is already an observable problem with the existing gesture and filter tools in sculpt mode:\n- None of the tools have shortcuts\n- Picking the tool, tool-type and adjusting the settings is time consuming\n- Each tool has their own disconnected settings\n\n### Conclusion\nWe should aim for a more systematic approach.\nAll of the requirements need to be easy to understand and access.\nThat is easier to accomplish if there are fewer tools that are not specifically just for a single purpose.\n\n# Proposal\n\nInstead of adding a lot of new tools/brushes and settings for filling in colors, we can rely on the current and planned masking features.\n\n### Mask -> Fill\n\nFilling should be as instant as possible. But first you need to define what to fill.\nBlender often has a strong design of \"Selecting -> Operation\". By strengthening the masking features via #97955 we could rely on this for color filling operations too.\n\n### Fill Operator\n\nThe most important part as a baseline is a simple operator that immediately fills in everything that is unmasked.\nCurrently this is implemented in Vertex Paint Mode with the shortcut `Shift + K`.\nThis could get a more convenient shortcut on the left of the keyboard.\n\nIt needs to be always clear what color will be used for the operator.\nThe easiest answer would be the primary color of the currently used tool/brush.\nIf the current brush/tool doesn’t have a primary color, then the primary color of the scene color should be used.\nIn that case it becomes very important that this color is somehow exposed. Otherwise there will be too much unnecessary switching between masking and painting tools.\n\n### Auto-Masking\n\nThis way of targeting specific areas is useful for any brush or tool. It already proved itself essential in Sculpt Mode.\nThe aim should be to use auto-masking for the fill operator as well.\n\nIt will be vital to add auto-masking support to the image editor for painting. Topology auto-masking in this case could mask UV islands instead.\n\nOpen question:\n- Is a “UV-island” auto-masking in the 3D Viewport needed or is it more convenient to initialise face sets from UV seams?\nThis could save us space in the auto-masking pie menu.\n\n### Masking Modes\n\nWe also need to allow filling by faces instead of pixels.\n\nWe are already aiming to implement three masking modes in Paint Mode (#97955) and an inverted masking approach (#97903) that could also cover this use case.\nWith this feature it would be as easy as pressing `2` on the keyboard and using any masking tool/operator to mask faces.\nThe fill operator can then fill all unmasked areas, or they can be painted in more detail with any painting tool.\n\n### Gradual Filling\n\nFor this the Color Filter, fill type should still be used. \nBut the fill operator could also get a strength input in the adjust last operation panel.\n", "\"Reset to Default Value\" sets alpha incorrectly in [Material Properties > Viewport Display > Color]\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 511.09\n\nBroken: version: 3.5.0\n\nRight-clicking \"Reset to Default Value\" on [Material Properties > Viewport Display > Color] sets both the color's \"value\" and \"alpha\" bars to 0.8, when the alpha one specifically should reset to 1. Alpha resets incorrectly.\n\nRight-click \"Reset to Default Value\" on [Material Properties > Viewport Display > Color] on any material.\n\n" ]

[ "Sculpt vertex colors conversion washed out\nOperating system: Win 10\n\nBroken: blender-2.90.0-613d314251f3-windows64\nWorked: Dunno..\n\nThis is something I found when loading vertex colors made in an external application (zbrush) into blender's sculpt vertex colors. \nWhen I convert the regular vertex colors into sculpt vertex colors, the colors shows up washed out...\nSimple example below:\n\nLoad the following sphere fbx file. **File/Import/FBX** (it contains vertex colors created in zbrush)... The model might show up very tiny, so you may need to frame the model to see it. **View/Frame Selected**\n[PolySphere.fbx](PolySphere.fbx) \nSelect the model and go to Vertex Paint mode. This is how it the colors look like (in flat shading), no issues...\n\nNow go to Sculpt mode\nSwitch the color type to Vertex, in the shading popover\n\nIn the Object Data properties tab, add a Sculpt Vertex Color layer, and then click on the \"Load Sculpt Vertex Color\" button, to load/convert the regular vertex colors into sculpt vertex colors\n\nAnd this is how it loads the colors (washed out)\n\n\nI don't now if it's a bug/limitation or if there's something else that needs to be done to make it work as expected... So yeah..", "Regular Vertex and Sculpt Vertex paint brightness difference\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 472.12\n\nBroken: version: 3.0.0 Alpha\nWorked: don't know\nI used sculpt vertex paint to paint a sphere and converted (stored) the sculpt vertex to regular vertex paint. That results in different brightness values.\n\nOpen the file and switch the sculpt vertex color (Color) and the regular vertex color (Col). There is a massive difference when it comes to the brightness.\n[regular vertex and sculpt vertex paint brightness difference.blend](regular_vertex_and_sculpt_vertex_paint_brightness_difference.blend)\n\n[2021-10-03 19-24-56.mp4](2021-10-03_19-24-56.mp4)\n\nYou can replicate the bug by using sculpt vertex paint on an object, store the sculpt vertex color to a regular vertex color layer and compare both in material or render preview via shader nodes and or via viewport shading and the vertex option.\n\nRegular vertex color should have the correct brightness i believe." ]

[ "bpy.ops.object.transform_apply no working after physics simulation\nOperating system: Windows 10 Pro\nGraphics card: 2x Nvidia 1080 GTX\n\nBroken: 2.8, b46245470f79, 2018-04-18\nWorked: 2.7x\n\nWhen running a rigid body simulation, it is not currently possible to apply transformation to freeze a specific frame. If you start scrubbing in the timeline, the object either goes back to original position or jumps to random position that was generated during the simulation.\n\n - add a plane and add a passive rigid body\n - add a suzanne and add an active rigid body\n - run the simulation\n - choose a frame\n - select the suzanne object\n - apply transformation/apply visual transformation\n # scrub in the timeline\n\n\n\n\nBased on the default startup or an attached .blend file (as simple as possible).", "Cycles: Persistent Data option make Point Cloud position is dirty for transformed objects\nOperating system: Linux-6.2.6-76060206-generic-x86_64-with-glibc2.35 64 Bits\nGraphics card: NVIDIA GeForce RTX 3080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 525.105.17\n\nBroken: Curent main.\nWorked: Never (3.0+).\n\nIf you turn on `Persistent Data` in cycles render settings, and change some object position properties, point cloud objects gonna be incorrect positions have on render.\n\n| Shifted point cloud. Expected. | Second result with persistent data option. |\n| -- | -- |\n|  |  |\n\n1. Create Point Cloud object.\n2. Turn ON `Persistent Data`.\n3. Try to render.\nEverything is okay.\n4. Open N-panel. Just click on point cloud position properties. No need to really change something.\n5. Try render again.\n\n", "Regression: Visual bug when using the object gizmos\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 471.68\n\nBroken: version: 3.0.0 Alpha\nWorked: 2.93\nCaused by 917a972b56\n\nDifferently from using the transform tools in the toolbar, when activating the object gizmos that stay visible also with the select tools, a strange bug occurs that the e.g. moving the object on an axis keeps a grey axis where it was before, the gizmo is always present (differently again from using the transform tools), and if you switch from this to the transform tools to the selection with gizmo again, it messes the origin of the object.\n[2021-09-20 14-59-52.mp4](2021-09-20_14-59-52.mp4)\n\n\nI'm pretty sure that this is caused by the recent change of keeping the viewport gizmo visible when doing these operations.\n\nUse the select tool in the toolbar, activate on or more object gizmos that stay always visible, do a transform operation.\n[Based on the default startup]\n\n", "XYZ symmetry options: unclear release notes\nThe 2.91 release notes for 5502517c3c are lacking, both in clarity of the description and the discoverability of that change for the people it impacts.\n\n- The change is described in the Sculpting section, under \"[Sculpting Workflow with multiple objects](Sculpt#Sculpting_Workflow_with_multiple_objects)\". However, it is much broader than just that. Since this change also impacts weight painting, it is relevant to riggers as well. It should be described in a place such that riggers can find it.\n- The only description is \"XYZ symmetry is now a per mesh settings which is shared between all painting and editing modes\". This doesn't convey what this change means practically, it doesn't show the changes in the UI, and doesn't tell riggers how they have to change their workflow to account for this change.\n\nAs a side-note, the #animation_rigging module should have been involved in design task #79785 and the review of [D8587](D8587), since this change impacts workflow of riggers. Instead, it went completely under the radar.\n", "GPU Subdivision: Wrong objects statistics information\nWindows 10\ni7-4790\n16 GB RAM\nRTX 2070\n\nBlender v3.4.1 (Steam)\n\nWrong objects statistics information in object mode when GPU Subdivision is enabled\n\n- Open `Rat.blend` file\n- Enable GPU subdivision\n- Rotate/move the bone\n- Switch to object mode (notice the viewport statistics)\n\n\n\n", "Fly and walk navigation in AZERTY issues\nOperating system: windows 10\nGraphics card: rtx 3060\n\nBroken: Blender 3.6.1 (Steam)\n\n\nHello,\nIm on a AZERTY keyboard and the problem is that when i change my movements keybinds for fly (or walk) navigation it keeps repeating the action (even if i unpress the key) for the binds that i changed. For exemple i changed \"forward\" key by Z and when i press Z the camera will keep going forward forever.\n\n- Go in the View 3d fly (or walk) modal keymap settings\n- In \"Forward\" replace W by Z, \"Left\" Q by A, \"Down\" A by Q, \"Z axis correction\" Z by W\n- Go in fly (or walk) navigation mode\n- Try to move with ZQSD, A and E\n- Then when you press S,D or E it will work normaly but if you press Z, Q or A and release the key it will keep going in that direction exept if you press the opposit direction key (S for Z, D for Q, E for A)\n\nThanks for paying attention, lov u <3\n\n", "Skin modifier root on every vert\nOperating system: win10\n\nBroken: (2.79b, 2.82)\n\nSingle vert is screwed into spiral via Screw modifier. Skin modifier treats every vert as root, ending up with bad polygon flow.\nLeft is object with modifiers, right is applied version to see issue clearly:\n\n\n1. Use single vert with offset from origin to create spiral via Screw modifier\n2. Put Curve modifier on it\n3. Add Skin modifier\n4. Apply every modifier before Skin\n5. Enter edit mode and see many roots\n6. Click \"Mark Root\" in Skin settings -> polygons calculated correctly\n\nIn attached file:\n6. Click \"Mark Root\" in Skin settings -> polygons calculated correctly\n[screw_skin_root.blend](screw_skin_root.blend)\n", "No Vector Displacement On Instances\nWindows 10 x64\n\n2.79.4\n\nVector displacement doesnt work with instances, It mirrors the displacement on the first object, Im sure this is working as intended but there are many cases where you need to displace actual instances as well.\n\nHaving a checkbox somewhere to enable instance displacement would be amazing\n\n\n\n\n\n[instances.blend](instances.blend)\n", "Bone rotation results in different roll (editmode vs. posemode + apply as restpose vs. setting tail position). \nOperating system: Windows-10-10.0.19043-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 511.65\n\nBroken: version: 3.0.1\nWorked: unknown, possibly never.\n\nBone rotation in editmode using the transform system introduces roll.\n\n[#95560.blend](T95560.blend)\n- Open .blend\n- a single bone in editmode (with its tail at 0/1/1\n- set its tail to 1/0/1 in the sidebar (bone roll remains at zero)\n- undo\n- rotate on global Z axis -90 degrees (head/tail are now exactly like above example)\n- bone roll is now at 26 degrees\n- undo\n- enter posemode\n- rotate on global Z axis -90 degrees\n- `Pose` > `Apply` > `Apply Pose As Rest Pose`\n- check in editmode again (head/tail are now exactly like above example): bone roll is now at -45 degrees\n\nSo this might be expected behavior (not sure it is very predictable, there is some bone roll correction going on?), but\n- it is undocumented\n- if this is the desired behavior, shouldnt above methods result in the same bone roll?\n\n\n**Original Report**\n\nRotating a bone around the 3d cursor in edit mode, using Global Z axis, introduces roll into the bone (IFF the head and tail of the bone are not perfectly in line). This behaviour is unintuitive and does not occur if the head and tail of the bone are perfectly in line with the Z axis. See Armature 1 in attached image. Performing the same action in pose mode has different behaviour, in line with expectations.\n\nOpen the attached blend file. Select one of the two example bones (Armature 4 will produce the error, Armature 5 will not. Enter edit mode, select the bone body in edit mode. Hit R -> Z -> 90 (rotate it around the Z axis 90 degrees). Observe the roll of the bone. \nFor comparison, perform the same steps in pose mode.\n\n\n\n\n[Bone rotation confusion example.blend](Bone_rotation_confusion_example.blend)\n\n", "Applying Skin Modifier to square loop object and in edit mode braking the loop makes prob\nOperating system: Windows-10-10.0.17134 64 Bits\nGraphics card: AMD Radeon HD 8400 / R3 Series ATI Technologies Inc. 4.5.13399 Core Profile Context 15.200.1065.0\n\n\nBroken: version: 2.80 (sub 74)\nWorked: (optional)\n\n\n\nSkin modifier disturbs the geometry form one corner on braking the loop form one corner of closed loop (of square shape). Although in this case the angles are 90% but it's will make prob...\n\n\n\n\n", "Preset icon overlaps panel header text\nOperating system: Mac 10.9.5\nGraphics card: Macbookpro NVIDIA GeForce GT 750M 2048 MB\n\nBlender 2.8 downloaded today\n\nIn movie editor\nthe text tracking settings is somehow mixed up with a button.\nI can sent a pic if you want.\n\nI am a amateur and English is not my first language, so i'm very sorry if i've wasted your time.\n\nGreetz \nLes", "Small Custom Bone Shapes have some Wireframes Artifact.\nWindows 10 64 Bits.\nGraphics card GeForce GTX 1080 Ti \n\nBroken: version: 2.82.6, hash: a854840e76ae\nWorked: (optional)\n\nSmall custom bone shape has wireframe glitching, the Scale is based on real world but if you resize it in/out u can see it happen.\n\n{[F8257584](Shape.gif) size = full}\n\n\n\n[Bug.blend](Bug.blend)\n", "VSE: Animation operators missing cache invalidation\nOperating system: Linux Mint 19.1\nGraphics card: GTX 980\n\nBroken: blender-3.1.0-alpha+master.c34ea3323a3a-linux.x86_64\n\nRemove created action from NLA editor, this will still animate VSE elements\nSome weird VSE action cache?\n\n- Add any image to VSE strip\n- Set Rotation=0 at first frame and Rotation=220 for some another frame. It's rotating now, right?\n- Go to NLA tab -> push down Scene Action and remove it. Thus, we don't have any actions in our project but the image is still rotating. Why?\nAfter rebooting blender app the magick will gone. Is this a bug?\n\nThis has quite broad reach as all of these (if not more) dont invalidate the cache:\n- clearing keyframes\n- deleting actions\n- removing drivers\n- removing dopsheet channels\n- removing graph editor channels\n- removing nla tracks\n- removing nla strips\n- removing fcurve modifiers\n\n\n", "Principled BSDF Anisotropy black faces\nWin10, 1080Ti\n\nBroken: 2.79\n\nIt seems that using anisotropy produces black odd artifacts under certain mesh circumstances. I don't think this happens often as it seems the actual mesh caused it, possibly by being a little bit too symmetrical? \n\nI suppose often it goes unnoticed so I thought it might be good for someone to investigate upon if there isn't some inconsistency.\n\nRotating the object by the tiniest bit helps (probably helps by something being non 0 ^__^).\n\nMoving random vertices randomly produces different results that are often completely correct or at least 50% correct. Good to try that in Rendered mode.\n\nScreenshot: \n\nBlend: [Anisotropy_Bug.blend](Anisotropy_Bug.blend)", "Motion Blur glitches, the further an object is placed from the world origin (even with no motion)\nMac Os 10.14.5\nRadeon Rx580 \n\nBlender 2.8 & 2.81 Beta (5/11/19 build)\n\nObjects are seriously corrupted when the motion blur checkbox is ticked, even when there is no camera or object motion.\n\nI have attached a simple project with only a Cube, Camera & light... & placed it 500,000m from the world origin on the Y-axis.\n\nThe image is distorted in rendered camera view...moving the cube backwards towards the world origin fixes the issue at around 50,000m\n\nMoving the object further away (try adding a zero to push the cube 1,000,000 from the world origin) distort the object more.\n\nSwitching off motion blur on the (static!) camera restores the image quality.\n\nI'm doing a space animation with some large distances involved, & this bug has just bitten me. I'd appreciated if others could confirm.\n[Motion Blur Draw Distance Test.blend](Motion_Blur_Draw_Distance_Test.blend)\n\n\n\n" ]

[ "Second character motion actuator overwirte forward motion [patch]\n%%%This bug report is related to [#36329] Character physics type doesn't react to actuator?.\n\nI know that I can enable the \"Add\" button to get it work. But if add is enabled and you move and rotate the rotation speed does not match with the character rotation.\nI made a fix for that. I also fixed the double jump problem. Now it is possible to jump permanently and make a double jump. I also removed the now unnecessary \"Add\" button.\n\n\n\n\n\n%%%" ]

[ "Show Emitter toggle still hides object after Particle System has been deleted.\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 1050/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 451.67\n\nBroken: version: \n- 2.92.0 Alpha\n- 2.92.0 Alpha\nWorked: version: \n- 2.92.0 Alpha\n- 2.91.0 Release\n\nWhen you delete a particle system with show emitter unchecked under Viewport Display the object stays hidden. You have to re-add the particle system and check Show Emitter.\n\n- Add Particle System to default cube.\n- Uncheck Show Emitter under Viewport Display\n- Delete Particle System. \n\n", "Action editor key frames not displaying sometimes\nSometimes when working with multiply animations and switching through them in the action editor. The Keyframes stop displaying for certain action. \nWhen the Panel with the action editor View is removed and added again, the displaying is working again.\nThis bug was seen in Version 2.76 and daily Build from 10/30/2015.\n\nError:\n\nFixed after closing and opening action editor view\n\n\nMy Project without Mesh. Bug is seen in walk_back_cycle action.\n[animation_editor_error_project.blend](animation_editor_error_project.blend)\n\nThis bug is very annoying in my workflow.\n\nGreetings and thank you for your work!", "weird viewport visual glitch\nOperating system: Windows 11 64 bits\nGraphics card: NVIDIA GeForce RTX 3060 Ti Gaming X Trio\n\nBroken: version: 3.3.1\nWorked: cant bother sorry\n\nits like the position of the viewport has an offset relative to the orientation of the camera, and the viewport has a set size and cant change.\n\nfirst go to the layout workspace and have an animation playing (must have an animation playing for the bug to happen), then create a new viewport by sliding it like im doing in the video. then change the viewport into rendering mode like im doing in the video and when you resize the viewport, it has a set size and cant be changed and the position of all parts have an offset according to the position and orientation of the camera.\n\n", "Animation: create a full separate copy of BoneCollections while in armature edit mode\nThis is to avoid touching the main data structures in the Armature datablock while in edit mode, which interacts badly with undo.\n\nRelevant:\n- #111965\n\n", "modifier display/render driver not responding.\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1050/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 527.56\n\nBroken: version: 3.4.0\nWorked: unknown\n\nIf I have my Mask display/render driver from another objects display/render setting, changing controlling obj disp/render doesn't take effect without coaxing.\n\n1. Open the attached file. \n2. In outliner, disable the viewport display of the Cube object (drives Mask visibility of part of Sphere object)\n3. For the Mask modifier, slide the Threshold a little. Display effect now takes place.\n\nThis all works without a problem if I animate the Cubes visibility. This appears to be a problem for any/all modifier display drivers (e.g., Subdiviison). Sometimes I just need to hover over a value like Threshold, or just over time it responds to the change.\n\n", "outliner visibility not binded to bpy.context.object.hide_viewport\nOperating system: Linux-5.7.2-zen1-1-zen-x86_64-with-glibc2.29 64 Bits\nGraphics card: GeForce GTX 1080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 440.82\n\nBroken: version: 2.83.0\n\noutliner visibility not binded to bpy.context.object.hide_viewport\n\nclicking the \"eye\" icon of an object in the Outliner shows \"bpy.context.object.hide_viewport = True\" in the info panel.\nBut in PythonTerminal the output is:\n>>> bpy.context.object.hide_viewport\n\nFalse\n\nAlso not possible to make it visible again with bpy.context.object.hide_viewport = False.\nSeems like bpy.context.object.hide_viewport is not connected to the Outliner icon at all.\n\n\n", "Object cryptomatte is not respecting the camera Dof\nOperating system: Windows 11\n \n\nBroken: 3.0, 3.6\nWorked: never?\n\n\n\nCycles object cryptomatte is not respecting the camera depth of field while the material one respects it properly. See the image for comparison\n\n\n\n\nload the attached .blend (it already has the object and the material cryptomattes enabled)\n\nrender\n\ncompare the current cryptomatte types in the compositor\n\n", "Z-fight with invisible face in edit mode\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 1660 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 457.51\n\nBroken: version: 3.0.0 Alpha\nWorked: \n\nI have 2 cubes. Bigger one is set viewport display as wire. Both share the same XZ plane. In Object mode shading is fine, but if I enter in Edit mode with smaller cube artifacts are appear.\n\n[ladder_01.blend](ladder_01.blend) \n\n", "Generated texture coordinates missing when switching from solid to rendered mode\nOperating system: Windows 10 64 bit\nGraphics card: Nvidia GTX 750 Ti\n\nBroken: 2.80.0-git.14884cda1ff5-windows64\nWorked: 2.79b f4dc9f9d68b\n\nThe \"Generated\" texture coordinates make textures \"stick\" to the surface, even if the object is being deformed.\nHowever, this does not work as expected in 2.80 with the Armature Modifier, neither in Eevee nor Cycles.\nIn 2.79 in works correctly.\n\nPossibly related: #59275.\nHowever, this bug affects both Cycles and Eevee, while that report only concerns Eevee (and Cycles works correctly).\nThe difference is that here an Armature is used, while that report uses Shape Keys.\n\n[generated.blend](generated.blend)\nThe scene contains a cube with an armature modifier.\n- Open in both 2.79 and 2.80.\n- Press F12 to render.\n- See how the texture mapping differs.", "Applying screw modifier doesn't respect hide status of edges from edit mode.\n\n\nUPDATE: If the Unhide hotkey is used, despite no parts actaully being \"unhidden\", the uneditable parts become editable.\n\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 531.61\n\nBroken: version: 3.5.0\n\n\nApplying a screw modifier with parts of mesh hidden in edit mode will result in those generated mesh from hidden parts to:\n\n1. only show as edges,\n2. be still selectable in edge select mode,\n3. and becoming uneditable.\n\nwhen going back to edit mode.\n\nOriginal description:\n\nIf certain parts of a mesh with a screw modifier are hidden in edit mode (though still visible with the screw modifier, just not selectable), and then the modifier is applied to the mesh in object mode, the previously hidden parts will become uneditable, though still selectable in edit mode.\n\n\n- Add a screw modifier to an object\n- Hide parts of the object in edit mode\n- Apply the screw modifier in object mode\n- Go to edit mode, the parts that's hidden supposedly has edges on display.\n - If use edge select mode, you can actually select those ghost edges, but not editable.\n - If use alt-h, then everything is shown and behaving normal again.\n", "Radius in place of Size at Empty viewport menu\nI{[F6596506](RIGHT.JPG)}\n\n am new to blender and not know where to post this but plz check it.It seem wrong to me\nthanks", "Local view mode is not exited correctly when closing New Windows\nOperating system: Mac OS*\nGraphics card: Macbook pro M1 MAX, 64G ram\n\n**Blender Version** 3.41 (same issues found in 3.2)\nBroken: Local view doesn't work(/)\nWorked: \n\nit's hard to say when it will happened, I used MacBook with two monitors, (one is Mac Studio, an other one is iPad Pro), when I duplicate a new windows to my second screen, like to iPad for sculpting, some times, local view function(hot key\"/\") doesn't work and show 'no more than 16 local views\" error. I met this error 3times from version 3.2 to 3.41\n\n\n\n\n- Repeat the following 16 times\n - Open a new `Window`->`New Window`\n - Go into Local view in the new 3d viewport that appears\n - Close the window while still inside local view\n- Attempt to go into local view 1 more time\n- Observe failure\n\nThe issue is that local view mode is not exited when those temporary New windows are closed. Blender continues to think that those local views are still in use and runs out of available space for any new ones.\n\n[test.blend](test.blend)\n\n", "Mask invisible on object with modifiers when entering sculpt mode\nOperating system: Linux-5.4.0-81-generic-x86_64-with-glibc2.27 64 Bits\nGraphics card: NVIDIA GeForce RTX 2070 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 470.57.02\n\nBroken: version: 3.0.0 Alpha\nWorked: unkown\n\nWhen selecting a previously unselected object with a modifier and a sculpt mask, the mask is invisible at first when entering sculpt mode.\nPainting the mask reveals it.\n\n- Create an object with a modifier (e.g. mirror or array) and paint a sculpt mask on it. Or use the file attached.\n- Outside the sculpt mode first make sure the object is unselected, then select it and enter sculpt mode.\n- The object appears with no mask visible, still you can not sculpt the masked parts.\n- Painting the mask reveals it.\n\n[mask.blend](mask.blend)\n\n[mask.mp4](mask.mp4)", "2.8 Cycles viewport Ray Visibility is not updated in viewport when changed by Python\nOperating system: Mac OSX, 10.13.6 (a MacBook Air)\nGraphics card: Intel HD Graphics 5000 1536 MB\n\nAlso tried on a linux virtual machine (Ubuntu), same issue\n\nBroken:\nMac build: 2.80.43, build 2019-02-01 22:37\nLinux build: 2.80., 2019-02-15 7:29\nWorked: 2.79, unsure if there was a working version of 2.8\n\n\nWhen updating an object's CyclesRay visibility (e.g. camera visibility) via python from a script or console, there is no impact in the viewport; while the UI as well as render will reflect the changed state of the setting (checking or unchecking the box), the visibility will not have changed in the viewport. Note: this is only an issue for the viewport, when rendering it will reflect the setting update properly. The issue persists even between leaving and re-entering rendered mode in the viewport, until the user manually clicks to toggle the UI setting directly which resolves the problem.\n\nBased on the default startup or an attached .blend file (as simple as possible).\n\n1) Open default scene\n2) Open console window/scripting workflow,\n3) Enable cycles and enter a viewport into rendered mode\n4) With Default cube selected, type: C.object.cycles_visibility.camera = false\n5) Notice how viewport scene does not update; but if you move the mouse over the property panel:Objects/Cycles settings, you'll see the draw refresh will show the camera visibility turns off\n6) Exit and re-enter rendered mode in the viewport, to find the issue persists (cube is visible to camera)\n7) Press render to notice that the setting however is reflect in render (cube will not be visible to camera)\n8) Manually toggle the camera visibility setting in by clicking the tick box twice, to verify the setting works when done by hand; and see that now the cube does not appear in the viewport (as expected).\n9) Repeat for reverse; ie turning setting to True when initial camera visibility is false.\n\nNote I also tested this and found the same issue with the Glossy visibility using a similar process (adding a mirror-like object next to the cube), so I expect this is true for all of the cycles object visibility settings. \n\nVideo demonstration, happens in any scene including the default\nview?usp=sharing\n\nWhile I agree this is less of an issue since ultimately the correct results appear in render, this is an issue for addons that control this setting and confuse the user as prior to render, there then appears to be a discrepancy between the viewport visibility and the panel settings. ", "mask visibility in sculpt mode using shape keys or generative modifiers\nOperating system: Windows 10\nGraphics card: nvidia 1060 Ti\n\n2.8 RC2\n\nIf you have a mesh with a mask and create a shape key, the mask will be no longer visible, still active but not visible.\n\n\n- Add UV Sphere\n- Sculpt mode\n- Paint a mask\n- Create shape keys\n\n\nOr from the default startup:\n- in sculptmode, `Mask` > `Invert Mask` (default cube should turn dark)\n- strokes dont have an effect now (this is expected)\n- add **any** generative modifier to the default cube (e.g. triangulate)\n- back in objectmode, save the file\n- reload the file\n- enter sculptmode\n- no mask is visible\n- strokes dont have an effect now (this is confusing since you are not aware masks prevent an effect since you dont even see the masks)\n\n(note we have the option `Use Deform Only` in sculptmode which might be exactly for that purpose, still not being able to use masks with generative modifiers should be more clear)" ]

[ "Can't switch mode when bone hidden\n**System Information** \nOperating system: Linux\n\n\n\nversion: 2.80 (sub 39), branch: blender2.8, commit date: 2019-01-01 23:46, hash: 93a8556bf4b1, type: Release\n\n\n \n* Select a bone \n* Go into pose mode\n* Hide the selected armature in the outliner using the eye icon (while in pose mode)\n* Try to select the other bone\n* It doesn't work and you are unable to change the mode you are in (pose mode)\n* Expected behavior, you should be able to select other armatures and change the mode even if the armature is hidden\n\n[BoneSelection.blend](BoneSelection.blend)", "Hiding a collection with an active rig in pose mode will lock the 3D Viewport from selecting new objects\nOperating system: Windows 10\nGraphics card: RTX 2080\n\nBroken: 2.80, dbfbd6b9043d, 2019-03-16 02:58\n\nThe UI (correctly) displays it switches to Object Mode if the active rig is in Pose Mode and the collection it's in is hidden. Unfortunately this locks the viewport from being able to make any new selection unless you specifically select an object in outliner. After this is done you can select objects in 3D Viewport again.\n\n" ]

[ "Mirror Uv\nThere already exists an operator called Copy Mirrored UV Coords. This could be extended to support the features described below\n\n\n- From +X to -X for example (source can be positive or negative X, Y or Z) \n- Options:\n - By object space\n - By world space\n - By topology \n\n\n\n**Starting point**\n\n\n**Desired result**\n\n", "Remesh + symmetry model bug? : Blender 3.0.0\nOperating system: Windows 10 Home N\nGraphics card: GTX 1660 ti\n\nBroken: Blender 3.0.0 - Stable December 21, 02:21:47 - f1cca3055776 - zip - 234.03MB x64\n- also 3.0.0 from front page have same bug\n\nafter remesh, symmetry broken symmetry_bug_fix.jpg\n(contains ways i fixing the error currently like merge by distance)\n\n1.Remesh model, 2.symmetrise it (weird line appears) \n\n[blender_3.0_remesh.symmetry_wierd_line.blend](blender_3.0_remesh.symmetry_wierd_line.blend)", "Bevel Bug\nblender-2.80-07765499abee-win64 \n\n**cylinder :**\non the cylinder with a lot of segments, bevel doesn't merge in one point. Create nice see n-gon(normal we must see the only verts in one point)\n\n\n**Cube :**\nThe middle loop has deformed when we bevel sharp corner for small mesh. On the large mesh, we cant see this deform .\n\n\n[Cylinder_Bevel_Bug.blend](Cylinder_Bevel_Bug.blend)\n\n[Cube_Bevel_Symmetry_Bug.blend](Cube_Bevel_Symmetry_Bug.blend)\n ", "Skin modifier can produce undesirable geometry\nBroken: blender-2.74-65328fa-win32\nWorked: never as the skin modifier always had problems like that\n\nThe skin modifier sometime fail and generate very bad geometry\n\n[skinbug.blend](skinbug.blend)\nload the attached blend, and see the problem as shown in the screenshot {[F177512](screenshotbug01.jpg)}(the underlying vertices are all symmetrical, as that base was done with the mirror modifier) \nTest with the various x/y/z options in the skin modifier and see how that buggy geometry can move in other place of the generated mesh\n\nOnly way to fix is to try to either move around the vertices or to rescale them (CTRL+A with the skin modifier) , not real solutions if you need all the branches to be +/- the same diameter and at exact locations.", "Bug - Refraction BSDF in Eevee interferes with Render properties > Film > Transparency - it blocks the transparency so to speak.\nBest demonstrated with a video (please note that the refraction bsdf is transparent, so objects can se seen through, so can the world background - when the world is not transparent that is) :\n\n[DE3VFV8aIO.mp4](DE3VFV8aIO.mp4)\n\n", "GP: Time offset and Armature modifier causes glitches\nOperating system:WINDOWS10\nGraphics card:GTX1060\n\nBroken: 2.93LTS\n\n\n[#89433.blend](T89433.blend)\n\n- Open file\n- Move selected bone to left until it hits constraint, and then a bit randomly\n\nGP object will keep rotating in one direction and can be reset only by changing frame.\nSame happens when scrubbing in negative frame range.\n", "Adding new hook while in Edit mode breaks when trying to undo/redo the operator\nWin7 (32bit)\n\nBroken: 2.78a\n\n\n- Add Path from Curve menu, add new Armature\n- Assign Bone to control point as Hook\n- Blank Hook modifier appears along with real one\n\nScreencast: [screencast0001-0250.mp4](screencast0001-0250.mp4)\n", "percent bevel by weight doesn't work properly\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 528.49\n\nBroken: version: 3.4.1\n\npercent bevel by weight doesn't work properly \n\n\n\n", "Modifiers added in edit mode are undone out of order if you go back to object mode\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1070 with Max-Q Design/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 441.66\n\nBroken: version: 2.83 (sub 13)\nWorked: Never (2.8+)\n\nModifiers added in edit mode are undone out of order if you go back to object mode and then start undoing.\n\nAffects both old and new undo system. There's often mentions of edit<->object transitions being problematic but just in passing; I think this is a long standing issue. Just wanted to note it down as I test out the undo systems.\n\nHoping the new undo system can account for this now.\n\n- Default scene\n- Go to edit mode on the default cube\n- Add any modifier (bevel, subdivision, etc.)\n- Add any loop cut 1\n- Add another loop cut 2\n\n- Go to object mode\n- Undo --> Modifier is removed first (out of order)\n- Undo --> Loop cut 2 is removed\n- Undo --> Loop cut 1 is removed\n\n\n", "Blender solidify modifier bevel convex bug ?\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.94\n\nBroken: version: 3.3.0\n\nOn the solidify modifier, under Edge Data > Bevel Convex, the bevel weight seems to be added randomly instead on the inner and outer edges of the rim. Is it a bug or there's something I'm missing ? \n\nI recorded it : WB83jpVG3qo\nThe video is on the version 3.2.2 but it does the same in the latest 3.3 build.\n\n- Open attached file or\n - Add a solidify modifier onto a mesh,\n - then add bevel weight by putting the value of Bevel Convex to 1.\n - Then, add a bevel modifier with Limit Method set to Weight.\n[Solidify+Bevel_Bug.blend](Solidify_Bevel_Bug.blend)\n\n", "Error de progama\nno me aparece el modo vèrtice\n\n", "Mesh Mirror Node\n{[F9914997](mirror.png) size=500}\n\n**Changes and simplifications in comparison to the modifier:**\n\n1. It reflects on an arbitrary plane which is defined by position and rotation.\n\n2. It shows a gizmo that visualizes the reflection plane.\n\n3. Merge by distance operation can be done by a separate node.\n\n**Open Questions:**\n\n1. Will we handle UV data in this node or a separate one?\n\n2. How to handle the Bisect option. Having Bisect as separate node makes sense, but to use it in combination with the mirror node would require to some effort to get them lined up correctly e.g. using at least the same plane for both nodes.", "Cycles shading glitch\nOperating system: Windows 10\nGraphics card: AMD\n\nBroken: 3.4.1\nWorked: ?\n\nCycles shows incorrect shading, also different from Eevee. (The file is very simplified just to showcase the problem)\nI'm thinking it is a variation of T99933 bug but not sure so tell if that's the case\n\n[bug.blend](bug.blend)\n1) Open the file, switch to rendered view. Normals should look like this\n\n2) In the material disconnect the alpha socket, now it should look different.\n\nIs alpha supposed to affect normals at all?\n3) Repeat the same in Eevee, results look consistent", "Texture painting with mirroring and viewport clipping does not mirror\nOperating system: Linux-5.8.4-200.fc32.x86_64-x86_64-with-fedora-32-Thirty_Two 64 Bits\nGraphics card: GeForce GTX 970M/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 450.66\n\nBroken: version: 2.91.0 Alpha\nWorked: never, 2.79 also did not do this\n\nTexture painting with mirroring and viewport clipping does not mirror\n\n[texturepaint_mirror_clipping.blend](texturepaint_mirror_clipping.blend)\n- Open .blend\n- Do a stroke on the right side --> updates on the left as well because x-mirroring is turned ON\n- (Undo)\n- {key Alt B} a clipping region in the viewport\n- Do a stroke again on the right side --> updates on the right side\n- {key Alt B} to clear the clipping region in the viewport, see no updates have taken place on the clipped left side\n\nExpected: stroke should update the clipped area as well [as we do for vertexpaint, weightpaint and sculpting], everything else would be inconsistent\n\n", " Add a 'relative mirror' option for mesh editing (similar to bones)\n**Team**\n**Commissioner:** `?` *Core developer or artist, sign off the project, works as client.*\n**Project leader:** @mano-wii *Main developer responsible for the implementation, only in one project.*\n**Project members:** `-` *Optional other developers, can be in multiple projects at once.*\n\n**Description**\n**Big picture:**\nJust as it works on bones in `Pose Mode`, meshes editing could benefit from the \"`Relative Mirror`\" option. \nThe idea is to consider the initial offset of the mirrored vertices when the \"`Topology Mirror`\" option is enabled and move the vertices respecting that offset.\nThis would solve problems like #74340 (Uncontrolled vertex movement in symmetric editing).\n\n**Use cases**:\n* Model asymmetric objects but with a symmetrical topology avoiding unexpected jumps.\n\n**Design:**\n* As seen with bones, the \"`Relative Mirror`\" option will also be available for meshes." ]

[ "Bevel Modifier misinterprets normals of mesh generated by other modifiers.\nLinux kenanb 3.12.9-2-ARCH #1 SMP PREEMPT Fri Jan 31 10:22:54 CET 2014 x86_64 GNU/Linux\n\nBroken: b53f310be8c1faf53e51b875eb3ca790c5efeee2\nWorked: 2.69 release\n\nBevel modifier misinterprets normals of mesh generater by other modifiers, therefore creating completely inside out bevels that looks like huge artifacts.\n\nReproduction:\n- Open default scene.\n- Enter edit mode\n- Move the cube on x-axis (preferably more than 1 unit to see the effect clearly)\n- Enter object mode\n- Add mirror modifier on x-axis\n- Add bevel modifier", "Mirror+Bevel modifiers produce incorrect result.\nWindows 7 64 bit, 2x gtx 570\n\nBroken: Blender 2.69, built 2013-12-18 18:35, hash 4b206af\nWorked: Blender 2.69 r61228\n\nWhen using a mirror modifer above a bevel modifier, the bevels on the mirrored side of the mesh are incorrect: lUdZMfj.jpg\nWhen viewing vertex normals, you can see the mirrored verts have strange normals which is probably the cause of it: http:*i.imgur.com/3RGGzIx.png (though in older builds that work, the vert normals show strangely as well but the bevel works fine: http:*i.imgur.com/zCiwvXd.png)\n\n1. Add an edge loop vertically on the default cube\n2. Delete the four verts on one side\n3. Add a mirror modifier and set the axis appropriately\n4. Add a bevel modifier.\n\nThis is probably due to recent changes to the bevel modifier ([rev1 ]], [[ http:*developer.blender.org/rB1158c1f7c0ec3f5d31989cb85207a08888d2c0d3 | rev2 ]], [[ http:*developer.blender.org/rB7a899ce9fc7fbcea219c12afe49c08114505b3aa | rev3 ]], [[ http:*developer.blender.org/rB9a78cda32111be9e1c9e2b9c158b27334d3c8804 | rev4 ]], [[ http:*developer.blender.org/rB39673718864f018307e49fac62696341b3230628 | rev5 ]], [[ http:*developer.blender.org/rBbfb9cefccbf8ff9733e9739e9c4dbc93a762bebe | rev6 ]]), but could also be because of changes to the mirror modifier ([[ rB7214001cdb142cb2556135b1f6e35290c62b9a69 | rev ](http:*developer.blender.org/rBda91575206cda63966fbd526378de98a866e8e5c)). Or both." ]

[ "Bmesh module does not support smooth shading / split normals\nBroken: 2.75 RC2\n\nBMLoop.calc_normal() returns normal for flat-shading, regardless of applied shading.\n\nBased on a (as simple as possible) attached .blend file with minimum amount of steps\n\nRun the following script, both APIs should give the same normals, but Bmesh module returns the same for all 3 objects.\n\n```\nimport bpy\nimport bmesh\n\nscene = bpy.context.scene\nfor ob in scene.objects:\n ob.select = True\nbpy.ops.object.delete()\n\nbpy.ops.mesh.primitive_cube_add(location=(0,3,0))\ncube1 = bpy.context.object\n\nbpy.ops.mesh.primitive_cube_add(location=(0,0,0))\nbpy.ops.object.shade_smooth()\ncube2 = bpy.context.object\n\nbpy.ops.mesh.primitive_cube_add(location=(0,-3,0))\nbpy.ops.object.shade_smooth()\nbpy.ops.mesh.customdata_custom_splitnormals_add()\ncube3 = bpy.context.object\n\n# no difference\ncube1.data.use_auto_smooth = True\ncube2.data.use_auto_smooth = True\ncube3.data.use_auto_smooth = True\n\ncube1.data.calc_normals_split()\ncube2.data.calc_normals_split()\ncube3.data.calc_normals_split()\n\nbm1 = bmesh.new()\nbm1.from_mesh(cube1.data)\nbm1.faces.ensure_lookup_table()\n\nbm2 = bmesh.new()\nbm2.from_mesh(cube2.data)\nbm2.faces.ensure_lookup_table()\n\nbm3 = bmesh.new()\nbm3.from_mesh(cube3.data)\nbm3.faces.ensure_lookup_table()\nbm3.normal_update()\n\nprint(\"Standard API:\")\nprint(cube1.data.loops[0].normal, \"Flat\")\nprint(cube2.data.loops[0].normal, \"Smooth\")\nprint(cube3.data.loops[0].normal, \"Smooth\")\nprint(\"---\")\nprint(\"Bmesh module:\")\nprint(bm1.faces[0].loops[0].calc_normal(), \"Flat\")\nprint(bm2.faces[0].loops[0].calc_normal(), \"Smooth\")\nprint(bm3.faces[0].loops[0].calc_normal(), \"Smooth\")\nprint(\"------\")\n```\n\nNote: Standard API could be used instead, but there's no RNA method for triangulation (only modifier and operator). Mixing Bmesh module and standard API seems unsafe (and there's no direct access to loops in bm). Object.to_mesh() could use support for triangulation, hence quad_mathod / ngon_method to improve the situation.", "Z depth not correct when using a volume shader\nOperating system: Linux-5.13.0-48-generic-x86_64-with-glibc2.31 64 Bits\nGraphics card: NVIDIA GeForce GTX 980/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 470.129.06\n\nBroken: version: 3.2.0\n\nZ depth gives strange \"far away values\" when using volume shader and thus cant be used for compositing\n\n\nJust render this blend file\n[Zmist.blend](Zmist.blend)\n\n\n", "Inaccuracy in 'Info' editor\nThis bug is difficult to describe specifically, since I have no idea of *all* of the affected properties. All I noticed is that *some* are reported inaccurately (i.e. the reported rna path does not exist or does not point to the changed property).\n\nA few I found are on the View Layer tab in the Properties editor:\n`bpy.context.scene.view_layers- [x].cycles.denoising_store_passes` is reported as `bpy.context.scene.denoising_store_passes`\n`bpy.context.scene.view_layers- [x].cycles.pass_debug_render_time` is reported as `bpy.context.scene.pass_debug_render_time`\n`bpy.context.scene.view_layers- [x].cycles.use_pass_volume_direct` is reported as `bpy.context.scene.use_pass_volume_direct`\n`bpy.context.scene.view_layers- [x].cycles.use_pass_volume_indirect` is reported as `bpy.context.scene.use_pass_volume_indirect`\nPlus all of the Cryptomatte and Denoising properties.\n\nI have no idea how the Info editor receives this information, but it looks as if the message is partially hard-coded and not up to date?", "Dynamic Indirect light cache\nThe idea would be to have a multiple level cache that stores a voxelized version of the scene to render the irradiance points.\n\nEach voxel would be linked to a set of surfel (surface elements) that stores color and normal of the surface.\nEach time the lights in the scene changes, the voxelized version is updated and propagates the change to the irradiance points.\n\nThis means having a list of surfel per voxel (say 6, 8) and a list of voxel per irradiance points.\nThis would be a mix of The Division dynamic indirect lighting [link]] and Mc Guire raytraced irradiance probes [[http:*jcgt.org/published/0008/02/01/|link](https:*www.gdcvault.com/play/1023273/Global-Illumination-in-Tom-Clancy).\nThis feature might be limited to hardware that supports compute shaders.\nThis is not a trivial feature and needs a lot of R&D.\n\nEstimate : 4 weeks\n", "Able to paint/select \"through\" disabled UI items.\nBroken: 2.80, 2.81...\n\nWith region overlap, if a UI element is diabled, clicks are passed through it, which can result in paint strokes.\n\n\n1. Load the default scene and enter Texture Paint mode on the cube.\n2. Add a paint slot and make a stroke.\n3. In the UV/Image editor, select the painted image, open the sidebar and switch to Image tab. The UI items should be disabled there now.\n4. Make sure that sidebar overlaps the image.\n5. Click on sidebar background.\n6. Click on any of the disabled items.\n\nIn (5), nothing happens. In (6), the click passes through and leaves a paint stroke on the image.\n\n", "Cycles, inconsistent power units between mesh light and rectangle area light\nOperating system: Windows 10 x64\n\nBroken: version: 2.82 (sub 6), branch: master, commit date: 2019-12-17 23:52, hash: e0cd5b980fc5, type: Release\n\nAs seen in #53517 comment :\n\n> Lamps are in Watts, [...]. Mesh lights are in Watts/m^2 [...]\n\n\nRendering the following two emissive items should give the same lighting :\n\n* rectangle area light, 1m x 1m, power 1W\n* plane mesh, 1m x 1m, Emission shader, strength = 1.0\n\nThis is not the case, area light power must be multiplied by 4 to get the same result.\n\n\nWith the attached blend file, [test-emissive-vs-arealight-1m².blend](test-emissive-vs-arealight-1m².blend)\n\n* enable 'Rendered' viewport shading\n* the 'Emissive plane' brings more light to the 'Ground' that the 'Area light'\n* change the 'Area light' power to 4W\n", "Exaggerated Ambient Occlusion Effect with Subsurface Scattering Materials in Cycles\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 471.68\n\nBroken: version: 2.93.2\n\nWhen rendering with Cycles, whenever meshes with materials that have subsurface scattering intersect, a very obvious ambient occlusion effect is generated around the intersections. It functions as expected with Eevee, but in Cycles, there's always a very unattractive dark seam line.\n\nApply any subsurface scattering materials to any two intersecting meshes, and render out with Cycles.\n\nHere are my results, I used a procedural snow shader, but I get similar results with any subsurface scattering materials.\n[subsurface scattering.blend](subsurface_scattering.blend)\nResults with Subsurface Scattering in Cycles\n\nSame shader in Cycles with Subsurface scattering turned off\n\nSame shader in Eevee with baked indirect lighting and subsurface scattering turned on\n\n\n", "Cycles Image Texture Alpha conversion inconsistency with EEVEE and Image Editor\nOperating system: Windows 10 Pro 64-bit (10.0, Build 18363)\nGraphics card: NVIDIA GeForce 920M (Laptop)\n\nBroken: v2.83.13 Beta, 2020-04-21\n\nIf an image is packed, it will get displayed differently in Cycles and EEVEE. It changes in EEVEE when\nAlpha Mode is changed from Straight to Premultiplied to match Cycles, but it does not change in Cycles\nby changing that option. If the image is unpacked Cycles will look like EEVEE with Straight Alpha Mode\nand again it will not be changed by the Alpha Mode option like EEVEE is.\n\n1. Open the provided file.\n2. Select the Image Texture Node\n3. Go into the right panel in the Shader Editor (N-panel) into the Item tab and Properties subsection\n4. Render Image\n5. try changing Alpha Mode and rerendering.\n6. try unpacking the image and rerendering.\n7. try changing the Alpha Mode on unpacked image and rerender.\n[packed_image_cycles_display_inconsistency.blend](packed_image_cycles_display_inconsistency.blend)\n", "Custom Normals Render Artifact in Cycles \nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3060 Laptop GPU/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 471.99\n\nBroken: version: 3.0.1\nWorked: Unknown\n\nWhen rendering a cube with custom normals in cycles the render result in cycles has artifacts. This is an issue for us because we also use cycles to bake Ambient Occlusion data to the vertex colors. Those artifacts are than also visible in the vertex colors. The same artifact occurs in previous versions of blender aswell (LTS Blender 2.93.7)\n\n\nBased on startup file:\n\n- Shade the default cube smooth\n- Apply Normals Auto Smooth\n- Add a single bevel using a bevel modifyer and apply it\n- Add a Weighted normal modifyer with weight 100 to the cube and appy it aswell\n# Render in cycles and behold render artifacts near the beveled edges\n\n[custom-normals-cycles.blend](custom-normals-cycles.blend)\n\n\n", "'Custom Split Normals' generated by the Boolean modifier does not match the normals of the object\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 1080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 431.07\n\nBroken: version: 3.0.0\n\nIf the object used as operand in the Boolean modifier has `Custom Split Normals` (either through the `WeightedNormal` modifier or manually added) the resulting object will have a \"broken\" `Custom Split Normal`.\nThis kind of works with objects which do not have any `Custom Split Normals` (so only shade flat and shade smooth, no weighted normals etc.) (although it also has it's problems. See #95913 (Strange shading behaviour when using boolean modifier))\nWhen `Custom Split Normals` are involved, the modifier uses the `Custom Split Normals` of the operand object on the target object to determine the shading on the affected areas.\n\n- Open attached file. (The left example has only shade smooth on the boolean object and on the right example the boolean object has a weighted normal modifier on the modifier stack).\nThe issue still persists even if you delete the weighted normals modifier on the target object in the right example.\n[Test_file2.blend](Test_file2.blend)\n\n---\nAnother example:\n[Test_file.blend](Test_file.blend)\n", "Baking vertex colors does not respect alpha with \"Diffuse\" selected but respects alpha with \"Direct + Indirect\" also selected\nOperating system: Windows-10-10.0.22000-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3060 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.59\n\nBroken: version: 3.2.1\n\nI believe this is a bug: when baking vertex colors, the alpha channel is included ONLY if \"Direct + Indirect + Color\" are all enabled, but then lighting gets baked in as well which is often not intended.\nThis leads to the situation that It doesn't look like there's a way right now to bake just color + alpha to a Color Attribute with alpha channel.\n\n- open the attached blend file: [BugDiffuseAlphaVertexBake.blend](BugDiffuseAlphaVertexBake.blend)\n- select the Cube\n- switch viewport to Eevee\n- make sure Render Engine is set to Cycles so things can be baked\n- in the Bake tab, click \"Bake\" with Bake Type=Diffuse and Contributions=[Color]\n- note the result doesn't have an alpha channel in vertex colors, it's just black\n\n- now, enable \"Direct\" + \"Indirect\" and Bake again\n- note the result has aloha channel BUT now looks incorrect as only diffuse was intended to be baked...\n\n\n- also note that neither \"Color + Direct\" nor \"Color + Indirect\" result in an alpha channel\n- both \"Direct\" only and \"Indirect\" only result in alpha channels but then the color can't be baked.", "Denoise artifacts in Albdeo and Normal pass caused by material roughness\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 970/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.71\n\nBroken: version: 2.93.0 Alpha\n\nThe `Denoise Data Albedo` shows gray flat colored areas if the `metallic/specular` material have the `roughness` value below `~0.274`.\nThis also affects the `Denoise Normal` pass because it can appear black in these areas.\nTo show these artifacts for `specular` and `metallic` materials, the depending values are: `Base Color`, `Metallic/Specular`, `Roughness` and the angle (between camera and face).\nAll of this leads to visible seams in the denoised image (NonLocalMeans and OpenImageDenoise). The more noise the more it is noticeable.\n\nRender passes of material with roughness on 0.0 (See Albedo and Normal pass):\n\nThis shows both the roughness dependent gray area overwriting the the camera dependent gray area.\n\nThis is just to show that irregular roughness inputs (like a noise) also create these. It is the same with Textures.\n\n\n**Edit (10.02.2021):**\nJust noticed and wanted to clarify. The Denoise Normal pass shows black in the top corner, because it reflects the background.\nHere is an example of an object being reflected in that area. \n\n\n**Steps for others to reproduce the error**\n- open attached file or:\n - create a plane\n - set Material to a Principled BSDF Node having: Base Color: 0.047; Specular: 1.0; Roughness: Linear Gradient.\n - set the camera to a viewing angle of 60° to the plane.\n - set the Worlds background shader color to a checker board to have something to reflect on the plane.\n- Render in Cycles with Denoise Data in the View Layer Properties activated.\n[Denoise_Layer_bug.blend](Denoise_Layer_bug.blend)", "Fill material act weird when two layers masks each other and one of them has Fill Holdout material \nOperating system: Windows-10-10.0.19043-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1050/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 472.39\n\nBroken: version: 3.1.0\nfill material act weird when two layers masks each other and one of them use holdout material and inverting mask.\nalso that happen when we use stroke material as holdout or both stroke and material holdout\n see video \n[0001-1466.mp4](0001-1466.mp4)\n", "Drawing size of particles not matching DPI\nOperating system: MacOS\nGraphics card: AMD Radeon Pro 450\n\nBroken: 2.83\nWorked: N/A\n\nWhen moving the blender window to the external monitor, the halo size of the particles changes, which can cause them to disappear.\nBest way to see what I mean is to watch this short clip of me moving the window, and also zooming in and out to show that not all colors are lost: Vu6MSHVxEUwYLvob8\n> In #82238#1283352, @jenkm wrote:\n> The problem here is with the `U.pixelsize` or something like that, as usual.\n> \n> When you move a window between Low and High DPI screens, the size of the spheres does not change correctly.\n> It seems that the size changes twice, like the UI scale plus the size of the spheres additionally.\n> \n> | Low DPI Screen | Moved to High DPI | Low DPI with Thick \"Line Width\" |\n> |  |  |  |\n> \n> And further, depending on the size, these spheres can be hidden by a mesh.\n> \n> | Thick | Default | Thin |\n> |  |  |  |\n> \n> The size of the spheres should not depend on `U.pixelsize`, first of all.\n> The visibility of these spheres also depends on the viewing angle, what is also not clear whether this is intentional or not.\n\n\n- Open attached file\n- Update the particles if needed\n- Move the window to another monitor with different DPI\n||\n| -- | -- |\n[particles.blend](particles.blend)\n\n---\n**Complex file:**\n[rain.blend](rain.blend)", "Diffuse Direct, Indirect Bake to Vertex Color Artefacts\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.94\n\nBroken: version: 3.5.0\nWorked: This has never worked properly (At least in my experience)\n\nWhen baking Diffuse light to vertex color, black triangles and random dark spots appear in vertex color. This effect is more apparent with:\n- Area lights.\n- Emissive surfaces.\n- Lights with very high power.\n\n1. Create geometry and give each a byte_color vertex color channel.\n2. Place area lights in scene with high power.\n3. Bake diffuse direct and indirect to color attribute.\n\nSee .blend of my project attached. In the file i've already baked the emissive light to vertex color where you can see the artefacts. Switch to cycles live render view to see the correct output. I think the tiled texture may have something to do with it. I've added a tiled texture and packed it in the example.\n\n" ]

[ "Bump Mapping via Displacement input with OSL Doesn't Work\n%%%--- Operating System, Graphics card ---\nMac OS 10.8.4 - AMD Radeon HD 6750M\n\n\n - Blender version with error, and version that worked ---\nBlender 2.68a. I don' think any other version works. I only tried it out with 2.68a and 2.67b.\n\n - Short description of error ---\nBump mapping via the Displacement input on the Material Output node does not work with an OSL shader. Maybe it is supposed to be this way, but it just seems kind of odd.\n\n\n - Steps for others to reproduce the error (preferably based on attached .blend file) ---\nTurn on Open Shading Language in the Render settings. Create a simple OSL shader in Blender (I just created a diffuse(N) shader) as you can see in the attached file. Connect the BSDF output of that to the input on the Material Output node. Now connect a texture to the Displacement input on the Material Output node. No bump mapping shows up (the only reason that I can think that this wouldn't work is that the Normals are being affected by the OSL shader...but unless I'm doing something completely wrong, I can't get that to look anywhere near as nice as when I just connect something to the Displacement input). To show that Bump mapping works otherwise, disconnect the OSL shader and just connect a Diffuse Shader to the Material Output. There should be bump mapping now. (One thing to note: True Displacement works with OSL when the Experimental feature set is turned on).\n%%%" ]

[ "3DConnexion Mouse not working in new Blender instance\nOperating system: Darwin-18.7.0-x86_64-i386-64bit 64 Bits\nGraphics card: AMD Radeon VII OpenGL Engine ATI Technologies Inc. 4.1 ATI-2.11.20\n\nBroken: version: 2.81 (sub 5)\nWorked: (optional)\n\n3dconnexion mouse works only in one instance of Blender\n\nIf I open 2nd instance of Blender 3d mouse does not work in it, only in the original file. This behaviour is same in official 2.8. I have tested this in Rhino for Mac. There Mouse works in both instances, which ever is active of course.\n", "Blender Crashing When Painting Stroke into a \"Moving Animated Character\" (heap-use-after-free)\nBroken: 4.0 alpha\nBroken: version: 2.79 - 2.82 (sub 7)\nWorked: Never\n\nI have this experiment of making LIVE talking head, via OSC. And I noticed that if I tried painting a stroke on a talking moving animated head, it will crash. Even with timeline not running.\n\nI got a video (Might need iPhone X with MOM addon to test this....) :\njhHdG6ZXFDU\n\n- Open attached file\n- Run animation\n- Paint over the animated object\n[paint_animated_shapekey_bug.blend](paint_animated_shapekey_bug.blend)\n\n\n", "Segfault in CustomData_blend_read\nOperating system: Debian Linux, Stable\nGraphics card: GTX970\n\nBroken: 3.6.0 Alpha, branch: main (modified), commit date: 2023-04-05 18:55, hash: 92c9c1d4003e, type: Release\nWorked: Unsure. Old blend I loaded and worked on for a while. Errors on open now.\n\nSegfault on load.\n\n1. Load blend, or open blender and load blend\n\n(No idea what the gvfs error is.)\n```\nFailed to open dir (No such file or directory): /run/user/6001/gvfs/\n[New Thread 0x7fffe3ae9700 (LWP 2310937)]\n...\n[New Thread 0x7fffdb9fe700 (LWP 2310945)]\nRead blend: \"/tmp/segfault.blend\"\n\nThread 1 \"blender\" received signal SIGSEGV, Segmentation fault.\n0x0000000001009af9 in CustomData_blend_read (reader=0x7fffffffd498, data=0x7fffd9c471f0, \n count=5002)\n at /opt/src/blender-git/blender/source/blender/blenkernel/intern/customdata.cc:5194\n5194 if (layer->flag & CD_FLAG_EXTERNAL) {\n(gdb) l\n5189\n5190 int i = 0;\n5191 while (i < data->totlayer) {\n5192 CustomDataLayer *layer = &data->layers[i];\n5193\n5194 if (layer->flag & CD_FLAG_EXTERNAL) {\n5195 layer->flag &= ~CD_FLAG_IN_MEMORY;\n5196 }\n5197\n5198 layer->flag &= ~CD_FLAG_NOFREE;\n(gdb)\n```\n\n", "lost 90% of my progress\nOperating system:\nGraphics card:\n\nBroken: (example: 2.80, edbf15d3c044, master, 2018-11-28, as found on the splash screen)\n\n\nBased on the default startup or an attached .blend file (as simple as possible).\n\ni was 90% through my build and had forgetten to save although i havent made saving every five minutes my priority anyways i was almost done when it freezes then crashes and all of mt work is now gone this has already happened once but i didnt lose anything so i therefore didnt report it\n\n", "Not recognizing numpad page up /down keys .\nOperating system: Windows-10-10.0.17134 64 Bits\nGraphics card: GeForce GTX 1050 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 431.60\n\n\nBroken: version: 2.80 (sub 75)\nWorked: (optional)\n\n\nBlender is failing to recognize the numpad keys !!!\n\nTrying to use the change workspace shortcut ctrl+PageUp /down and blender is not recognizing the keys in numpad and keep assuming the key 9 and 3 . Other software is fine and i can use Page up and Down, and the home end keys. \n\n", "Grease Pencil mask error\nOperating system:windows10\nGraphics card:GTX1060ti\n\nBroken: 2.83&2.9\nWorked:greasepencil-Mask error\n1.When 2.83 or later, creates two objects, the masks collide, causing a strange phenomenon, and in 2.82 there are no errors.\n2.And creating two masks under the same object can cause problems too。\niJlqkf556sA\n[GP剪贴测试.blend](GP剪贴测试.blend)", "adding drivers in the compositor always results in a dependency cycle which often causes blender to hang when reopening the saved .blend file.\nOperating system: Windows-10-10.0.17763-SP0 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 442.92\n\nBroken: version: 2.90.0 Alpha\n\nadding drivers in the compositor always results in a dependency cycle which causes blender to hang when reopening the saved .blend file.\n\n\nJust to clarify, the append issue I mention in the video only happens if you copy and paste 3d objects from a file with the dependancy cycle problem into a new file, save the new file, and then try and append that new file into another new file, in this instance it also results in a missing data block if trying to re-open the file you copied into.\n\n[compositor dependancy cycle on all drivers.blend](compositor_dependancy_cycle_on_all_drivers.blend)\n\n[2020-06-02 17-50-21.mp4](2020-06-02_17-50-21.mp4)\n\n", "Add sticky keys support for accessibility\nOperating system: Windows 10\nGraphics card: Geforce GTX-970m\n\nblender-2.80.0-git.1fa527bfa3a-windows64\n\n\nSticky keys not working properly.\nSfZ7Ua-ReuQ", "Closing System Console can lead to loss of work\nOperating system: Windows\n\nAny\n\n* Start Blender\n* Toggle the System Console (2.83 and newer `Windows > Toggle System Console`\n* Press the - [x] button on the system console", "Use draw manager for clip editor drawing\nCould be stalled to 2.93 as there isn't an real issue here as far as we are aware of. The compositor issues are more tangable.", "Hair particles crashing when baking or animating\nOperating system: windows 10 64 , windows 8 64\nGraphics card: intel, nvidia\n\nBroken: blender-2.81-9bd0d8b5505a-windows64\nWorked: (optional)\n\nwhen trying to bake a hair particles the system hangs giving this message\n\nError : EXCEPTION_ACCESS_VIOLATION\nAddress : 0x00007FF7CEBD24A4\nModule : C:\\Users\\dribeiro\\Desktop\\blender-2.81-9bd0d8b5505a-windows64\\blender.exe\n\n\nopen the file\nclick 'play' on timeline or bake on hair particles\n\nNEW INFO\nremoving the hair particles the animation runs normally\n[giselle.blend](giselle.blend)", "Blender 2.92.0 crashes when baking steam simulation\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce RTX 3090/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 461.92\nCPU: Intel Core i7-8700K\nMemory: 32 GB\n\nBroken: version: 2.92.0\nWorked: I have not tried any other version than 2.92.0 (sorry)\n\nI tried to make a simple steam simulation following the steps described here: watch?v=_K0XsHht6pg. It crashed while baking, and tried to simplify the setup with a minimal set of objects, and it still crashes after baking about 30-50% of the 300 frames.\n\nFollow the steps as described in the video watch?v=_K0XsHht6pg\n\nAttached is .blend file that I tried to make as simple as possible: [steam.blend](steam.blend)\n\n - Open the attached file\n - Press \"Bake Data\" for Physics Properties for the cube\n - Blender crashes after 30%-50% of baking\n\nI have not tried a newer experimental release or an older release.\n\nI was able to reproduce the issue with the default 32 Resolution Divisions (which is way faster to reproduce than the 160 in the included file). I then tried experimenting which exact setting caused the crash. I got it to crash with Adaptive Domain disabled, and Gas Heat 1.5 and Dissolve Time 30, but now I'm not able to reproduce anymore with any setting (at least with 32 Resolution Divisions).\n\nAfter some more experimenting, it seems it randomly crashes and sometimes works. It is definitely reproducible with 32 Resolution Divisions. I kept all settings as I had before, which had worked successfully, but then changed the Time Scale to 0.8, and then it crashed again when baking. However, it had previously crashed at least once with the exact same settings with which I had also successfully baked (and the Time Scale at the default 1.0), so it is not the Time Scale that is causing the issue.", "Crash (memory use after free) on freeing RigidBody world\nReproducible with current master, probably there since forever.\n\n# Steps\nWith an ASAN build:\n\n- Open `rigid.anim.blend`\n- Close Blender\n\n[rigid.anim.blend](rigid.anim.blend) \n\n(Case found while working on #86379 (When using \"Append\" rigid body objects + empty constraint, none of its constraints work.))\n", "Rendering animation freezes with particle system\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce RTX 3090/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 461.40\n\nBroken: version: 2.91.2\nWorked: 2.83LTS (not sure, this is said in most of the comments)\n\nI am trying to render out an animation based on the CGBoost Blender Launchpad Tutorial. Every time I try to render, it will render around 50-400 frames and then it will just stop rendering new frames. My CPU usage goes straight up to 100% when that happens and my GPU usage drops down to 0%. The Blender UI is still responsive, it is just that the render seems to get stuck on a frame. If I cancel the render, then Blender becomes totally unresponsive and I have to force quit via the Task Manager.\n\nIssue happens in both EEVEE and Cycles\n\n**Workarounds (picked up from comments)**\n- Render animation from command line\n- Check lock interface: {nav Render > Lock interface}\n\n1. Open attached Blend file \n2. Press Ctrl-F12 to render out animation.\n3. Wait for render to freeze on a frame. This can happen within the first few frames or after 400-500 frames. It seems random.\n4. Examine CPU and GPU usage.\n\n[Blender Launchpad.blend](Blender_Launchpad.blend)\n[ING BG.blend](ING_BG.blend) (File provided by @Megaseppl: #85267#1122053)", "Select Loop Inner-Region issue\nWin8 64bit | GTX680\n\nBroken: 2.77 509270e\n\nWhen working with face loop, this tool doesn't always work as expected.\n\n\n\n\nNot sure if this can be a bug. Please check." ]

[ "2.80 32-bit appears broken (on 64-bit Windows at least)\nWindows 10 64-bit, Nvidia 1060\n\nBroken: Current 32-bit 2.80 daily\nblender-2.80-4dd0a90f4213-win32 \n\nBlender intermittently fails to start, fails to read date/hash, silently quits opening .blend, throws exception on exit, etc.\n\nAfter reading\n\n4757\n\nI downloaded the current 2.80 32-bit beta and it definitely seems to have issues, at least running on top of current Windows 64-bit (I don't have a 32-bit system to test on currently).\n\nIt will fail to start (exits silently) most (all?) of the time when started normally.\nIt will start when run with --debug, but throws an EXCEPTION_ACCESS_VIOLATION if you immediately quit.\nOpening any .blend file has about a 50% chance of causing an immediate exit with the same exception.\nThe --debug output shows \"Version 280 sub 39 date unknown hash unknown\" about half the time, but the rest of the time it finds the date and hash successfully." ]

[ "Gizmo position is wrong for meshes with modifiers\nWin10, Win7, Mac OS Sierra\nGTX970, GTX750\n\nBroken: 2.78a e8299c8 , 2.66.0 r54697\n\n\nWhen a joint (for example, an arm) is bent, paint weight is not enough to express deformation around the joint. So it's a common practice to add a shape key for the bent shape and edit it.\nBut when you enter edit mode and select polygons to edit it, the manipulator is on wrong position. \n\nJust select faces. Manipulator is supposed to sit on the selected faces but it's not.\n\n[manipulator bug.blend](manipulator_bug.blend)", "Able to paint/select \"through\" disabled UI items.\nBroken: 2.80, 2.81...\n\nWith region overlap, if a UI element is diabled, clicks are passed through it, which can result in paint strokes.\n\n\n1. Load the default scene and enter Texture Paint mode on the cube.\n2. Add a paint slot and make a stroke.\n3. In the UV/Image editor, select the painted image, open the sidebar and switch to Image tab. The UI items should be disabled there now.\n4. Make sure that sidebar overlaps the image.\n5. Click on sidebar background.\n6. Click on any of the disabled items.\n\nIn (5), nothing happens. In (6), the click passes through and leaves a paint stroke on the image.\n\n", "Constrain axis always draws on top [covers mesh in unfavorable ways]\nThe mesh outline is not visible when extruding or moving along the axes. The axes overlap the faces of the mesh themselves, and because of this, it is not convenient to precisely position the extruded faces.\nIt exists on all versions of the 2.8x blender.\n\nTo repeat the problem, you can select an edge in the editing mode and simply drag along the y axis. In this case, the contour will not be visible by how much it is moved from its actual position.\n\n\n\nWith the same action in 2.79, I see black the contours of the mesh as far as it changes when moving.\nBring it back as it was in 2.79", "When updating render percentage, the Box Mask node in compositor is also affected\nI ran into a confusing issue the other day that made me log bug #50113, thinking there was an issue related to masks and render size, but it turns out the issue is in the viewer node not updating correctly when adjusting render size.\n\nIn the attached fine, when I change the render percentage and update a value in the compositor, the result in the viewer changes so the mask is not consistent.\n\nIf I re-render the scene, the mask would appear relative.\n\nUnfortunately, since the viewer updates and displays the incorrect information, this can be confusing to the user and cause them (me) to freak out and try and adjust things that don't need to be adjusted. For example, I ended up changing my mask values over and over again, never achieving the correct result. \n\n - Open: [maskIssue.blend](maskIssue.blend)\n - Render an image\n - adjust the render resolution percentage to 100% (so far it's ok)\n - click on the `width` property of the `Box Mask` node and drag to a lower value\n\nNotice that the mask changes in the preview. Render again, and the mask looks correct.\nRendering resolution should not influence the result of the composite.\n\n", "Geometry Nodes: Only re-evaluated log do update available attributes in drop-down menu\nBroken: version: 4.0.0 Alpha\n\nThis is a fairly insignificant UI bug. If the GeoNodes modifier is disabled, the list of available attributes to select in one of the inputs is not updated. For example, when a vertex group is renamed, that won't be reflected in the list, until the modifier is enabled again. Can result in a bit of confusion.\n\n1. Open attached file.\n2. Notice the GeoNode modifier is disabled.\n3. Rename the vertex group.\n4. Try assigning the renamed vertex group as the modifier's Factor input; You can't, because it won't appear in the list.\n5. Enabling the modifier updates the list again, and then all is well.\n\n", "Displacement support for USD export & import\nMesh displacement is critical to realistic materials particularly for environment surfaces. The current USD exporter (and I assume importer) seems to ignore the displacement shader nodes.\n\nDemo file attached, or can be downloaded from [this page](concrete_rock_path) (CC0).\n\nExpected result:\n\n\n\nImported USD without displacement:\n\n\n\n", "Undo in Sculpt Mode\n\nOperating system: Windows-10-10.0.17763-SP0 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 457.30\n\nBroken: version: 2.90.1\n\nWhen I apply undo in sculpt mode it applies to my previous settings as well (symmetry activation, some of brush settings and any settings in sculpt mode). \nThis is so annoying moment please help with it?\n\n", "Dynamic paint sub-steps don't work when the brush position is set in geometry nodes\n\nI believe this **may** be the same issue as T40597\n\nIn order to have a brush follow a potentially very large collection of curves in sequence, I'm using geometry nodes to set the position, which avoids the pain of having to join the curve paths or painstakingly tweak the offsets and influence keyframes for each and every one.\n\nThis works, but with sub-steps enabled it seems that the nodes are still only being updated once/frame. Or, if they are being updated, then the `seconds` input from the `Scene Time` node is being derived from the `frames` value which is clamped to an integer.\n\n\n[text_path_test.blend](text_path_test.blend)\n", "Generated texture coordinates missing when switching from solid to rendered mode\nOperating system: Windows 10 64 bit\nGraphics card: Nvidia GTX 750 Ti\n\nBroken: 2.80.0-git.14884cda1ff5-windows64\nWorked: 2.79b f4dc9f9d68b\n\nThe \"Generated\" texture coordinates make textures \"stick\" to the surface, even if the object is being deformed.\nHowever, this does not work as expected in 2.80 with the Armature Modifier, neither in Eevee nor Cycles.\nIn 2.79 in works correctly.\n\nPossibly related: #59275.\nHowever, this bug affects both Cycles and Eevee, while that report only concerns Eevee (and Cycles works correctly).\nThe difference is that here an Armature is used, while that report uses Shape Keys.\n\n[generated.blend](generated.blend)\nThe scene contains a cube with an armature modifier.\n- Open in both 2.79 and 2.80.\n- Press F12 to render.\n- See how the texture mapping differs.", "Undo doesn't work with image type empty\nOperating system: Windows-10-10.0.17763-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 497.09\n\nBroken: version: 3.1.0 Alpha\n\nUndo doesn't work with image type empty (reference image) when you use the bracket or the centre cross to scale or move the empty image, that is very annoying when you accidentally modify your reference image image. A working undo or an option to disable the direct moving/scaling possibility (yellow gizmo) for such empty will be welcome. The undo is active when we use the normal scaling or moving gizmos or panels. \n \n\n- Create an image reference (drag and drop an image in the viewport) \n- Change size and position using the yellow gizmo on the object itself \n- Undo. \n\n", "Self-referencing driver works in viewport, but not in render.\nOperating system: Windows-10-10.0.17763-SP0 64 Bits\nGraphics card: GeForce GT 730/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 391.35\n\nBroken: version: 2.82 (sub 6)\nWorked: (optional) 2.79b\n\nRotation property of object is updated in viewport, but not in render.\nIf run same script in 2.79 - all works fine both in viewport and render.\n\n```\ndef rotator():\n var1=bpy.data.objects[\"Suzanne\"].rotation_euler.z\n var3=bpy.data.scenes[\"Scene\"].render.fps\n\n new_value=var1+math.radians(var3)\n\n return new_value\n```\n\n\n[Script_bug_2.8.2.blend](Script_bug_2.8.2.blend)\n\n- run the file Script_bug_2.8.2 ('Run script', 'Update Dependecies')\n- play in the viewport - the animation works correctly\n- render (Ctrl + F12) - the animation does not work;\n\nViewport:\n[bug_video_render0001-0050.mkv](bug_video_render0001-0050.mkv)\n\nRender:\n[bug_video_render_20001-0050.mkv](bug_video_render_20001-0050.mkv)\n\n2.79 file:\n[Script_work_2.79.blend](Script_work_2.79.blend)\n\n", "Sculpt with Perspective mode. Changing brush size is impossible with closups\nOperating system: Windows-8.1-6.3.9600-SP0 64 Bits\nGraphics card: GeForce GTX 660 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 461.40\n\nBroken: version: 2.93.0 Alpha\n\n\nAs title says. This behavior prevents to use {key F} as size modifier on closeups.\n\n[2021-03-20_10-32-46.mp4](2021-03-20_10-32-46.mp4)\n\n[brushsize.blend](brushsize.blend)", "texture in solid mode not updated\nOperating system: Ubuntu 20.04\nGraphics card: Geforce MX230\n\nBroken: 3.2, 2.93\nWorked: probably never\n\nif the shader editor is located in another workspace, in which there is no viewport in solid mode, then the texture on the object will not be updated\n\n- Open .blend file\n- In both layout an shading workspace: make sure skin texture is selected\n- switch to shading workspace and select blue texture\n- Now switch back to layout workspace (notice `blue` texture is selected in node space but viewport is not updated)\n\n[#99598.blend](T99598.blend)\n\n---\noriginal file and video provided by the reporter\n[jupiter.blend](jupiter.blend)\nVideo [F13283900](2022-07-11_00-19-08.mp4)", "Regression: GPU Subdivision: EEVEE doesn't refresh viewport correctly with Armature modifier\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 2060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.94\n\nBroken: version: 3.3LTS, master\nWorked: 3.2.2\n\nGPU Subdivision doestn't refresh viewport correctly an object when Armature and Subdivision Surface Modifiers are used.\n\n1) Start Blender.\n2) Add Armature object\n3) Switch to \"Shading\" workspace.\n4) Select the Cube and add `armature modifier` then `Subdivision Surface Modifier`.\n5) Switch to Properties Editor to \"Modifier Properties\"\n6) Add armature in armature modifier's object property. \n7) Switch between object-edit mode\nIt works fine when \"GPU Subdivion option\" is turned OFF (Prefrences > Viewport > Subdivision > GPU Subdivision)\n\nI added \"BeforeGPUSubDBug.blend\". You can reproduce this issue by loading the blend file and doing step 7)\n", "Undo of operations on modifiers is inconsistent when object is in edit mode\nBroken: version: 2.79, 2.82 (sub 1), branch: master, commit date: 2019-11-22 18:14, hash: 373e936e3e\n\n\nwhen a modifier that has been **added in edit mode** to a object and is **deleted in edit mode,** it is not restored when using undo. \nThe operation shows up in the undo history but does nothing.\n\n\n1. go into edit mode\n2. add a modifier\n3. delete the modifier\n4. use undo, notice modifier is not restored.\n\ncreating the modifier in object mode or deleting it in object mode will restore it when using undo. this is a very confusion inconsistency that i noticed with the new undo method.\n" ]

[ "Painting displacement image doesn't update the 3D display in realtime\n%%%I am using blender 2.55 beta 32bit for linux on a 64bit ubuntu 10.10.\n\nWhat I did was create a quick and dirty image in the image editor and start painting it. Next I applied the image as a displacement image and added a displacement modifier. I noticed that the 3D display doesn't update in realtime, and only updates after I tab into and out of edit mode once.\n\nPerhaps its a good thing that the mesh isn't altered in real time when painting, especially if everything was high poly, but perhaps it would be nice to apply the changes on a mouse up event after painting at least?\n\nI made a quick screencast of the problem, but it has been in a \"still processing\" state on youtube for a few hours now. If you need the video, I will try to attach it to this bug report later (~50 mb, but I could always shrink it) or wait for youtube to finish it and paste the url in another comment.\n\nHope this helps! Cya.%%%" ]

[ "Using viewport visibility as a driver doesn't work without constantly updating it in the driver property\nBroken: version 2.93\nWorked: Never (2.8+)\n\n\nUsing viewport visibility as a driver from another object doesn't works unless you constantly hit \"update dependencies\" it in the driver property.\n \n \nFor this, i used the cube's visibility as a driver on the shape key of the sphere.\n[The not working file.blend](The_not_working_file.blend)\nIn the outliner, enable the viewport visibility for the cube, then disable it again. The shape key doesnt update unless you go into the driver property and hit the update dependency button.\n\n\n\n[The working one.blend](The_working_one.blend)\nThis is what i was hoping it would do. The only way I could use viewport visibility as a driver is to use a custom property on that object then use the custom property as a proxy.\n\nIs there a reason why it dont update if you use visibility as a driver from another object or is that a bug?\n\n**Steps to reproduce**\n[The not working file.blend](The_not_working_file.blend)\n\n- Open file\n- Click on Disable in Viewports icon on Cube object in outliner\nSphere shape should be change, but it doesn't", "stack_overflow exception while rendering with viewport render on\nOperating system: win10\nGraphics card: 2070super\n\n[blender_system_info.txt](blender_system_info.txt)\n\nBroken: 2.83.4, 2.91 6fe609d4c3\n\n\nwhen pressing f12 in render mode, blender crashes,\nwhen pressing render in solid mode, blender doesnt crash\n\n[blender_debug_output.txt](blender_debug_output.txt)\n```\nError : EXCEPTION_STACK_OVERFLOW\nAddress : 0x00007FFF42963FDB\nModule : C:\\Windows\\system32\\DriverStore\\FileRepository\\nv_dispsi.inf_amd64_6af46268c19e4980\\nvcuda64.dll\n```\n[2.blend](2.blend) \n1. switch to render mode\n2. press f12 & watch task manager.\n3. blender will close", "Intersect Selection in edit mode\nOperating system: Linux-5.17.5-76051705-generic-x86_64-with-glibc2.35 64 Bits\nGraphics card: NVIDIA GeForce GTX 1080 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 510.73.05\n\nBroken: version: 3.3.0 Alpha\n\nIntersect existing Selection apply just to what in the viewport frame \nI can t reproduce it with an exact steps\nopen the linked file and select some vertexes with Select Box and with Intersect Mode\nthe intersection apply just to the vertexes shown in the viewport frame \nsame problem in 2.83 - 3.1 - 3.2\n[bug.blend](bug.blend)", "Cycles rendered 3DViewport massive performance drop, using lights with node groups. --Intel MacOS--\nOperating system: macOS-13.4-x86_64-i386-64bit 64 Bits\nGraphics card: Metal API AMD Radeon Pro 5700 XT 1.2\n\nBroken: version: 4.0.0 Alpha\n\nSee screenrecording video! \nWhen switching to a light with nodegroups in Cycles rendered 3DView, performance drops a big time. (~50%).\n\n1. Open the performance.blend file with Intel MacOS blender 4.0, switch to cycles rendered 3DView, select the cube and rotate the view.\n2. Then switch to the caustic light and try to rotate the view again.", "Problem with stop animation when View Viewport Render Animation\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1650 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 497.17\n\nBroken: version: 3.5.0\n\nThere is big gap and Problem with **stop** animation when View -> Viewport Render Animation is started in blender, eg. when it's selected accidentally and as alternative to View -> Viewport Render Image.\n\nTry to stop this animation. By me when I made this mistake not found solution other by kill blender process, cos stop button not was available or \"x\" btn like render (in case render view updating each image on Evee rendering process).\nNo imagine some big project and situation that user must waiting to finish, and he is not sure about saved file and blender is hanged due this process.\n\n", "No space mouse support in --debug mode\nOperating system: macOS 12.6.7\nGraphics card: Apple M1 Max GPU\n\nBroken: 3.6.1\n\nConnexion Space Mouse does not work if Blender is launched with --debug.\nNormal launch works.\nIs this related to #93170 ?\n\nLaunch Blender with --debug\nTry to use Space Mouse\n\n", "Brushes Assets Viewport User Interface\n{[F13331829](image.png), size=full}\n\nAdditional Notes:\n\n{[F13331831](image.png), size=full}", "Dual monitor issue\nOperating system: macOS-13.2.1-arm64-arm-64bit 64 Bits\nGraphics card: Metal API Apple M2 Pro 1.2\n\nBroken: version: 3.5.0\n\nUsing a new Mac Mini Pro with single monitor (Apple Studio Display), Blender 3.5 is very fast.\nAdd in a second Studio display and the interface becomes 'clunky', contextual menus take sometimes 2 or 3 seconds to appear, playback is jerky. \n\nTried various settings in preferences but no change. Unplugging the second monitor returns good performance...\n\nUodate:\nSetting 'GPU Backend' to 'OpenGL' improves things a little...\n\nUpdate-2:\nVersion 3.4.1 seems to have similar issues\n\nUpdate-3\nI seem to have solved some of the issue..\nThe Apple Studio Displays have 4 input ports, one Thunderbolt port and 3 USB-C ports.\nThe external display was plugged into one of the USB-C ports.\nSwapping input to the Thunbderbolt port seems to speed things up.\nStill a lot of lag on contextual menu pop-up,and selection box drag out.\n\nUpdate-4\nSo it tried reducing the Blender window to 1/4 screen, Blender becomes smooth again.\n\nI then methodically made it bigger and bigger until the latency suddenly reappears.\n\nProvided I leave around 15mm of gap around the edge of the Blender window with the desktop peeking around the edge, Blender stays very smooth.\n\nPull it a bit further out and the latency returns.\n\nIt can’t cope with absolute full screen, knock it back 5% smaller and it’s fine.\n\nWeird!!\n\n\n\n", "Metal backend: View Port drawing issue with face orientation\nOperating system: macOS-13.4-x86_64-i386-64bit 64 Bits\nGraphics card: Metal API Intel(R) Iris(TM) Plus Graphics 1.2\n\nBroken: version: 3.5.1, 3.6-beta, 4.0\nWorked: with OpenGL backend\n\nWhen enabling \"face orientation\" in the view port the display does not update until I rotate the model. When rotating the model the face orientation colour overlay display is broken.\n\nSee attached screen movie showing the issue.\n\nTo reproduce:\n1. start with a new default scene\n2. enable face orientation overlay from view port overlays menu\n3. rotate the cube\n\nThis was observed using the default startup cube.\n\n", "Keyed Particles don't render as expected\nOperating system: Windows-10-10.0.17763 64 Bits\nGraphics card: Intel(R) HD Graphics 4000 Intel 4.0.0 - Build 10.18.10.5069\n\nBroken: version: 2.81 (sub 16)\nWorked: (optional)\n\n[KeyedParticlesJL.blend](KeyedParticlesJL.blend)\n\n- Open file\n- Go to frame 105\n- Render\n- Compare render to viewport\n\nRender is different to viewport\n\nThat's what I expected:\n[ViewportRender.mp4](ViewportRender.mp4)\n\nThat's what I get when rendering with Eevee:\n[EeveeRender.mp4](EeveeRender.mp4)\n\nThat's what I get when renderng with Cycles:\n[CyclesRender.mp4](CyclesRender.mp4)\n", "Viewport: Auto-Depth rotation ignore transparent image reference empties\nOperating system: Windows-10-10.0.17763-SP0 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 442.92\n\nBroken: version: 2.90 (sub 0)\n\ncant rotate around partially occluded image objects.\n\n\nopen file, make sure rotate around mouse is enabled in preferences. hover mouse over either side images and hold in the mouse wheel, move the mouse, the view will rotate around the mouse. Now do the same for the middle image, the view no longer rotates around the mouse.\n\n[rotate around mouse bug with image objects that are occluded.blend](rotate_around_mouse_bug_with_image_objects_that_are_occluded.blend) \n\n", "Blender 3.1a sticks and doesn't render when rendering with Metal GPU on Mac\nOperating system: macOS-12.1-arm64-arm-64bit 64 Bits\nGraphics card: Apple M1 Max Apple 4.1 Metal - 76.3\n\nBroken: version: 3.1.0 Alpha\nWorked: This project rendered on the same 3.1a build on the CPU, just not on the GPU.\n\nWhen I render this scene on the GPU, blender just sticks and doesn't render. (Blender 3.1a has always rendered other scenes perfectly on the GPU.)\n\n- Open .blend file\n- Render as Image (`F12`)\n The issue may be at the fault of either the fact that I have displacement on the plane, or that I have 2 render layers\n[try1.blend](try1.blend)\n\n", "Blender renders wrong in viewport\nOperating system: macOS-11.2.3-x86_64-i386-64bit 64 Bits\nGraphics card: AMD Radeon Pro 5500M OpenGL Engine ATI Technologies Inc. 4.1 ATI-4.2.15\n\nBroken: version: 2.93.0 Alpha\n\n\n\n1) steps to reproduce:\n\nopen blend file [circular_circles_wave_emission.blend](circular_circles_wave_emission.blend)\n\n2) in viewport change to rendered view\n\n3) you will see:\n\n\n\n4) select the circles\n\n5) press tab and tab again\n\n6) you will see:\n\n\n\nwhich is correct\n\n", "Viewport Render (and Render Animation) color looks completely off from the Viewport\nOperating system: Darwin-19.0.0-x86_64-i386-64bit 64 Bits\nGraphics card: AMD Radeon Pro 560X OpenGL Engine ATI Technologies Inc. 4.1 ATI-3.2.24\n\nBroken: version: 2.82 (sub 6)\n\nRendering the viewport animation gives a very different result from the actual viewport.\n(Also: Viewport Anti-Aliasing isn't taken into consideration on the Viewport Render).\n\n||\n| -- | -- |\n\nThe problem is more noticeable when the Viewport Shading is set to \"Flat\", even though it also shows in \"Studio\". It delivers very dull colors with less contrast and a blue-ish tone.\nChanging between Cycles and Eevee doesn't affect the Viewport Render (from my tests).\n\n||\n| -- | -- |\nflat:\n||\n| -- | -- |\n\n- Open attached file.\n- Go to View -> Viewport Render Image\n[Viewport Colors.blend](Viewport_Colors.blend)", "Eevee viewport soft shadows are broken on higher sample count\nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce RTX 3080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 461.92\n\nBroken: version: 2.93.0\nWorked: ---\n\nEevee soft shadows in viewport have color spills. But it works fine on final render (F12)\n\n\n\n- Open .blend file\n\n[bug_soft_shadows.blend](bug_soft_shadows.blend)\n\n- Switch to Rendered Mode of Viewport Shading\n- Wait half a min and see the problem(" ]

[ "3D Viewport Bug - Ghosting of Geometry - Blender 2.8 d28d202fd1 on macOS High Sierra 10.13.5\nMacbook Pro 15inch, 2017\nmacOS High Sierra 10.13.5\nIntel Core i7 2.9GHz\nRadeon Pro 560 4096 MB\nIntel HD Graphics 630 1536 MB\n\nBroken: 2.8 d28d202fd1 x86_64\nWorked: 2.79 ;)\n\nWhen I scroll in on an object, i. e. a cube, when I zoom back out i still see some slight silhouette of the cube blocking parts of the grid and other objects. As soon as i resize the viewport it disappears but comes back as soon as i zoom or rotate the 3D View. I attached a screencapture to elaborate.\n[Blender-d28d202fd1-x86_64-viewport-bug.mp4](Blender-d28d202fd1-x86_64-viewport-bug.mp4)\n\nVanilla Blender 2.8 without modifications. Regular Startup File with Eevee." ]

[ "The program crashes when texturing \nOperating system: Windows-10-10.0.19041-SP0 64 Bits\nGraphics card: GeForce GTX 1660/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 461.72\n\nBroken: version: 2.93.0\n\nWhen try texture object in texture paint mode, and there are another map in shader tab. Program crashes. \n\n1. In default (or other) scene go to Texture paint window, create new texture for paint. \n2. Apply this texture to diffuse color object in shader editor. (or in object properties)\n3. Go back to texture window. We make sure that everything is painted correctly\n4. Go to shading window and drag&drop any other texture to node window. (NOT APPLY TO DIFFUSE COLOR (or APPLY))\n5. Go back to texture paint window. For some reason, the dragged texture itself was assigned to the duffuse color.\n6. Change linced texture in image editor (in left window)\n7. Try paint on RIGHT window (3D viewport)\n8. Click on left texture editor (try paint in 2D ) program crash. \n\nVideo steps: 0Ophy4Lmx0A\n\nDebug logs:\n[blender.crash.txt](blender.crash.txt)\n\n[blender_debug_output.txt](blender_debug_output.txt)\n\n[blender_system_info.txt](blender_system_info.txt)\n\n[Blender_crash.webm](Blender_crash.webm)\n\n", "bake hair dynamics issues\nOperating system: wind 10\nGraphics card: gtx 1070\n\nBroken: blender-2.80-79f67acccb36-win64\nWorked: idk\n\nunable to bake hair dynamics. i click the button but nothing happens. in particle edit mode particle hair still looks as if it was not simulated.\n\n", "Hair particles: cannot delete hair particles in particle edit mode until you comb it\nOperating system: Linux-3.10.0-957.10.1.el7.x86_64-x86_64-with-centos-7.6.1810-Core 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 418.56\n\nBroken: version: 2.81 (sub 4)\nWorked: 2.79\n\nYou cannot delete all (hit a) hair particles until you comb them. Even more, if you select one hair or several and hit delete, they disappear, but if you then select all the rest and hit delete, all the particles deleted come back.\n\n\n- Open this blend file: [#69487-hair-delete-279.blend](T69487-hair-delete-279.blend)\n- Press {key X} and choose Particles\n\nIn current Blender (2.82 alpha @ 756b9acaf5) hairs get deselected but are not deleted. In 2.79b the hairs are deleted.\n\n**Original steps to reproduce:**\n\nCreate a new hair particle system\nGo into particle edit mode\nSelect all, hit X delete particles -> nothing happens\nDeselect all\nSelect a couple, hit X -> the selected hairs will disappear\nSelect all, hit X -> all the hairs come back visible again\n\nThere's a hair particle system newly created in the attached file.\n[hair_particles_delete_bug.blend](hair_particles_delete_bug.blend)", "Blender crash on combining textures (intel)\nOperating system: Windows 10\nGraphics card: Intel HD Graphics 4400\n\nBroken: 2.93.2\nWorked:\n\nWhen I tried to combine 2 textures like this blender is crashing.\n  \n\n[Test.blend](Test.blend)\n[Test.blend1](Test.blend1)\n[system-info.txt](system-info.txt)\n[Test.crash.txt](Test.crash.txt)\n[2021-11-12 09-11-32.mp4](2021-11-12_09-11-32.mp4)\n", "Crash with certain file. Sculpt mode. Vertex color fill + edge.\nOperating system: Windows-8.1-6.3.9600-SP0 64 Bits\nGraphics card: GeForce GTX 660 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 461.40\n\nBroken: version: 2.93.0 Alpha\n\nWhen I try to apply vertex color Fill (Sculpt mode) on edge in this particular file, I get crash\nAnd all fine for shapes/triangle.\n\n[VC + edge crush.blend](VC___edge_crush.blend)\n\n[debug_logs.7z](debug_logs.7z)\n", "Regression: Particle mirror broken for particular mirrored mesh.\nOperating system: Windows-10-10.0.19044-SP0 64 Bits\nGraphics card: NVIDIA GeForce GTX 1080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 511.65\n\nBroken: version: 3.1.2. also version: 3.3.0 hash: `8d9d5da13706`\nWorked: 3.0.0, commit hash: `8b44b756d850db38ab4d83513682f2012d88a3f4` from branch 3.0.1 to be exact.\n\nCaused by 02ab4ad991\n\nParticle hair x-mirror results in hair going in all directions instead of being properly mirrored, see the following screenshot: \n\n1. Start from the attached blend file which has an emitter sourced from a base model and mirrored by applying a mirror modifier. I could not reproduce the issue from scratch with simpler geometry and I might have reconnected the hair at some point .\n2. Go into particle edit mode on the `Eyebrows_Particle` object.\n3. Enable x-mirror in the top-right `Options` menu.\n4. Observe that the particles change directions.\n\nHere is what it looks like when it's working in the same particle edit mode with x-mirror enabled. The exact commit that introduced the issue is: `77694b571f5bd297ea7228b30f591ebbd333e0ea`.\n\n\n\n[mirror_bug.blend](mirror_bug.blend)\n\n[repro.blend](repro.blend)\n- open file\n- `Particle` > `Mirror`\n- observe wrong mirroring", "Blender 3.5 crash on render with Intel 4600 integrated GPU\nOperating system: WIndows 10 x64\nGraphics card: Intel 4600\n\nBroken: 3.5 \nWorked: 3.4.1\n\n\nBlender crashes when attempting to render in Evee. Cycles seems to work.\n\n", "Blender 3.5.0 Win11 RTX3090 Fatal Crash on activating Cycles\n@deadpin thanks for replying.\n\nDriver Version: 31.0.15.3161 (NVIDA 528.24) DCH Win1164 installed\n\nCycles now working without issue thank u.\n\n_Originally posted by @Jaymi-McManus in /blender/blender/issues/106953#issuecomment-923094_\n\n>>>>>\n\nUpdate: \n\n1. Closed the ticket too soon, after adding some simple color, specular, metallic, changes in BSDF and then navigating between layout and other tabs, fatal crash occurs again\n\n2. Updated NVIDIA drivers to 31.0.15.3161 (NVIDIA 531.61) DHC / Win11 64 the issue still persists\n\nUpdate logs attached.", "rendering crashed\nOperating system:\nGraphics card:\n\nBroken: (example: 2.80, edbf15d3c044, master, 2018-11-28, as found on the splash screen)\n\n\nBased on the default startup or an attached .blend file (as simple as possible).\n\n", "Cycles GPU/CPU render difference with more than 64 transparent bounces and coplanar faces\nOperating system: Windows-10-10.0.18362 64 Bits\nGraphics card: GeForce GTX 1660 SUPER/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 441.66\n\nBroken: version: 2.81 (sub 16)\n\n\n\nWhen using particle plane with transparency ( to fake volume scatter ), we have a color mismatch between CPU and GPU rendering. \n\n\n see attach file\n\n\n\n[cpu_gpu_issue.blend](cpu_gpu_issue.blend)\n \n", "Particle Edit mode for non hair particles - 2 Crashes\nOperating system: Windows-7-6.1.7601-SP1 64 Bits\nGraphics card: GeForce GTX 760/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 390.65\n\n\nBroken: version: 2.80 (sub 69)\nWorked: (optional)\n\n\nParticle Edit: 2 Crashes\n\nParticle Edit: 2 Crashes\n\n- Open the blender file and click on the buttons \"Particle select and display point\"\n- Crash 1 (Screenshot)\n\n- Open the blender file and click on the tool \"Weight\" and click into the particles\n- Crash 2 (Screenshot)\n\n- Also had a crash with \"Puff\" \n- file not included\n\n- There are some tools that do nothing while others work\n- User gets no feedback / error message but give no feedback\n\n\n\n[2019-05-19-particle-edit-crash.blend](2019-05-19-particle-edit-crash.blend)\n\n", "Interpolated hair curves clipping into meshes with subdivision modifiers unapplied\nOperating system: Windows 10\nGraphics card: RTX 2080 Super Max-Q\n\nBroken: 3.51\nWorked: N/A\n\nInterpolated hair curves clip into meshes with unapplied subdivision modifiers. Will stop clipping once subdivision is applied.\n\n1. Create a mesh (i.e. suzanne) and add a subdivision modifier with any amount of subdivision. \n2. Add 'fur' hair curves and do a basic groom\n3. Notice (hopefully) that there a bald spots. No amount of extra guide curves will remove these bald spots\n4. Applying the subdivision modifier should remove these bald spots.\n\n", "Joining meshes (having equally named attributes on different domains) leads to name collisions for attributes (leading to e.g. errors on gltf export)\nOperating system: Windows-10-10.0.17763-SP0 64 Bits\nGraphics card: NVIDIA GeForce RTX 3060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 512.95\n\nBroken: version: 3.6.0 Alpha\n\n\nIf I have two attributes with same name, and try to export it as glb, exporter produces huge page of inner errors,\nsay that there is name collision, but do not say what object cause the issue.\nExpected: short message, more readable, says «object _objName_ have name collision in attributes. Attribute_name appears more than once, can`t export»\n\nIn case when you have hundreds of objects with dozens of attributes it is hard to find where the issue is and fix it. And such long page of errors only confuses user.\n\n\n- on one mesh object, create a color attribute on the `Vertex` domain (doesnt matter if `Color` or `Byte Color`) called `Color`\n- on another mesh object, create a color attribute on the `Face corner` domain (doesnt matter if `Color` or `Byte Color`) called `Color` as well\n- join the two mesh objects\n- leads to name collisions (two `Color` attributes on the same mesh)\n- leads to errors on e.g gltf export\n\nThis is also true for other attributes (does not have to be color attributes really)\n\n**Original report**\n\nOpen file [atrcollision issue.blend](attachment), try to export as gltf/glb, check error log.\n\n", "Some particle hairs that emitted from the vertices do not rotate with the armature.\nOperating system:MacOS\nGraphics card:Radeon Pro\n\nBroken: 2.93.2\n\n\nSome particle hairs that emitted from the vertices do not rotate with the armature.\n\n\n\n\n\n\n\n\n\n[Hair Rotation bug.blend](Hair_Rotation_bug.blend)\n\nSimple example wit a cube:\n[#89749.blend](T89749.blend)\n\nIssue only happens when emitting from Vertices (emitting from faces is fine)", "different generation of graphic card won't utilise together.\nOperating system: Windows 10\nGraphics card: Radeon R9 295X2 (secondary chip) + Radeon RX560 4GB\nAMD Ryzen 1700X\nMSI X370 Gaming Pro Carbon\n3 * 8GB DDR4-2400\n\nBroken: 2.82.7 to 2.83.3 <--- where x is from 0 to current stable version\nWorked: 2.8, 2.83.4 (partly worked, may end up sudden wattman reset)\n\nRadeon Version: 20.7.2\n\nafter the 3rd step, the render did not show anything, switching back to material viewport, the whole program was frozen. not sure if that is graphic card problem, but that way i could not even render anything. on 2.82.7, this also happened and the render kernels failed to work.\n\n\n1. open up a file of as shown\n2. set the preferences that would use OpenCL to render with R9 295X2 (and RX560)\n3. preview in Cycles\n4. unable to load the preview, switch back to material viewport.\n\n[zeta school 1.blend](zeta_school_1.blend)" ]

[ "Hair instability with collision\nOperating system: Windows 10\nGraphics card: Geforce GTX 1070\n\nBroken: 82c3fdd559b32d88881883b99cf441935401cf05\n\nIf dynamic hair comes in contact with collision object, the hair \"explodes.\"\n\n1. Open attached file.\n2. Bake hair simulation of cube. \n3. Hair explodes as soon as it contacts suzanne.\n\n - Reproduced this with several emitter / collider meshes, particle settings etc. \n\n[hair_collision_bug.blend](hair_collision_bug.blend)\n\n[system-info.txt](system-info.txt)", "Hair Dynamic collision pulsing\nOperating system: Linux-4.18.0-17-generic-x86_64-with-Ubuntu-18.10-cosmic 64 Bits\nGraphics card: GeForce GTX 980 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 415.27\n\n\nBroken: version: 2.80 (sub 55)\nWorked: 2.79\n\n\nThe hair keeps jumping on collider and not seems to take the collider properties into account like dumping, friction...\n\nJust open the attached file and hit spacebar to play the animation and run the simulation.\n\n[HairDyn_Jumping.blend](HairDyn_Jumping.blend)\n\n" ]

[ "Draw mode - separation in Grease Pencil coping layers that are empty\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1650 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.71\n\nBroken: version: 2.91.2\n\nIn Draw mode - separation in Grease Pencil on object coping layers that are empty from old GP_object. This made missunderstanding.\n\n1. Add GP Blank or Stroke\n2. Made 5 or more drawnings, eg. lines on 5 diferent layers\n3. Separate 2 layers to other GP_object to leave 3 layers\n4. Check the Object Data Properites for Stroke1 and Stroke2 after separation - the new Stroke2 have all layes but empty after separation just the 2 layers separated have objects (this is corect) like Stroke1 - no sense to import layers if object from these layers are not separated\n - this made conflict in some way of:\n1whbbc/ GP - move layer betwen objects (by menu and outliner)\n0whbbc/ GP - duplicate layer for stroke by RMB on layer or outliner\n3whbbc/ GP - Draw - Merge Layers by RMB on Stroke layer and outliner\n\nIn other way user should have option to separate with whole layers ot without (as option maybe)\n\n\n", "Lattice UVW input Glitch\nOperating system: Darwin-18.7.0-x86_64-i386-64bit 64 Bits\nGraphics card: AMD Radeon Pro Vega 64 OpenGL Engine ATI Technologies Inc. 4.1 ATI-2.11.21\n\nBroken: version: 2.81 (sub 16)\nWorked: (optional)\n\nClick dragging vertically through the UVW parameters of a Lattice will let you select all 3 parameters (as expected).\nBut, entering a new number will only change the 1st of the 3 selected parameters.\nWeirdly, click drag selecting all three parameters and entering nothing-at-all and then clicking away will somehow enter in the correct number that was entered one attempt back in time.\n\nStart a new General file and delete all objects.\nAdd a Lattice.\nGo to the Lattice Properties.\nClick and drag down through the U, V and W fields.\nEnter a new number.\nNotice that only 1 of the parameters changed (when all 3 should have changed).\nClick and drag down through the U, V and W fields again.\nClick away.\nNotice how the other 2 parameters \"catch up\" to the first 1.\n\n", "Inconsistent Grease Pencil auto-keying behavior\nOperating system: Linux 64-bit\nGraphics card: GTX 1080\n\n3.1 and prior\n\nWhen auto-keying is enabled and you're transforming a GP stroke in edit mode on a new frame, a keyframe is added at the beginning of the operation before it is confirmed. If you right-click to cancel the operation, the redundant keyframe remains. This is inconsistent with auto-keying in every other area, in which Blender waits for the user to confirm the operation before adding the keyframe. \n\nAlso, deleting a stroke fails to add a new keyframe. I apologize if these are two different bugs but they seem related.\n[GP_autokey_bug.blend](GP_autokey_bug.blend)\n - Open the attached file.\n - Grab (G) a stroke, then right-click to cancel moving it. The keyframe was inserted, even though no change was made.\n # Go to another frame, then delete a stroke. No keyframe was inserted this time, even though the GP drawing has definitely changed!\n\n", "Bake of the second particle system after duplication and making it unique makes mess.\nOperating system: Windows-7-6.1.7601-SP1 64 Bits\nGraphics card: GeForce GTX 660 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 430.86\n\n\nBroken: version: 2.80 (sub 74)\nWorked: version: 2.79 (sub 7), branch: blender2.7, commit date: 2019-05-23 08:19, hash: 054dbb833e15, type: Release\nbuild date: 24/06/2019, 17:18\n\nIf you duplicate particle settings, make them unique, and tune, bake of dynamica will broken.\n[2019-07-16_10-03-29.mp4](2019-07-16_10-03-29.mp4)\n\nFinally I found how to reproduce it.\n\nCreate plane, \nCreate particle system that emits cubes (color A) from -50 to 50 frame with short lifetime (20).\nDuplicate particle settings, make settings unique, rename it.\nChange instance to another cube (color B), change working period to 50 — 100 frames.\nCheck work in the viewport. Bake all dynamics, Revert file.\n\nMessed:\n[duplicated particle settins.blend](duplicated_particle_settins.blend)\n\nBug is not appere only if you create separate fully independent settings in the fresh file.\nWith independently created particles systems\n[fresh separate created.blend](fresh_separate_created.blend)\n\n\nPerhaps it can be connected with #61963\nLooks like some data just stucks\n\n\n", "Many Grease Pencil operations cannot be performed without an \"Active Layer\", though it shouldn't be needed\nOperating system: Linux 64-bit\nGraphics card: GTX 1080 ti\n\n2.93.5 and also 3.1 alpha (Nov 19th build)\n\nWhen working in a scene with more than one GP object, it's easy to go into edit mode for one GP object while having selected the layer of another in the dope sheet. Selecting a GP stroke may automatically select a layer for a different GP object. When the active GP object in edit mode has no active layer, a lot of operations are disabled, and it makes no sense as to why they should need an active layer at all.\n[GP_bug.blend](GP_bug.blend)\n - Open the attached file, which is in Edit Mode for the Suzanne GP object on the left\n - With no active layer belonging to Suzanne selected in the Dope Sheet, not a single operator in the \"Stroke\" menu can be performed. Selected strokes can be transformed and extruded, but they cannot be assigned materials or subdivided. This is odd because none of these operators seem like they should need any particular layer to be selected\n # Deselect all layers in the Dope Sheet and select a stroke. A layer is automatically selected, but it belongs to Suzanne.001, which is also baffling.\n\n", "Breaking nodes API changes for 4.0\n- [x] Remove `Node.width_hidden`\n- [ ] Complete \"Spline\" to \"Curve\"/\"Curves\" rename in geometry nodes\n- [ ] Stop storing node locations relative to their parents\n- [ ] Use a separate type for internal links (in RNA too) (see [D16960](D16960))\n- [ ] Use a separate type for node group interface inputs and outputs and node sockets\n- [ ] Use dynamically allocated `char *` array to store string socket default values", "After duplicating a collection with rigidbody objects, the simulation doesn't apply to the new rigidbody objects\nOperating system: Windows 11\nGraphics card: NVIDIA GeForce RTX 3080 Laptop GPU\n\n\nBroken: `master` branch\nWorked: Don't know if it ever worked\n\nAfter duplicating a collection with rigidbody objects, the simulation doesn't apply to the new rigidbody objects\n\n**Steps to Reproduce**\n\nIn a new blender project, add a new collection. Add a cube to the collection. Add a rigidbody to the cube. Verify that playing the simulation causes the object to fall under gravity.\n\nCopy and paste the cube to create a new cube inside the collection. Move the new cube so both are visible. Verify that playing the simulation causes both cubes to fall under gravity.\n\nIn the outliner right-click on the collection and select \"Duplicate Collection\". Move the two new cubes so all four cubes are visible. Run the simulation.\n- The two cubes in the original collection will fall under gravity.\n- The two cubes in the duplicate collection do not move and are not affected by the simulation.\n\n \n\nIt appears the two cubes in the duplicated collection are not added to the RigidBodyWorld collection.\n\n", "As a user I would like the option *add* to my existing pose with the Pose Library instead of *replacing* it.\nCurrently with the pose library, I can select some controls and then blend to a new pose. This is very handy for many cases, but sometimes you may want to use a pose in an additive method.\n\nTwo example use cases.\n\n1. I have a frame where I've posed my hand to look like a loose fist and now I just want to squeeze the fingers together a bit more. I have a pose in the library called \"finger squeeze\" where each finger finger is just brought together in a single axis. With the current library, if I drag on the finger squeeze pose it would **blend to** that finger squeeze pose.. removing the loose fist. If we add the ability to **add** to the pose instead of replace, then I could keep my fingers in this loose fist, but additively squeeze them together. \n2. I've posed my character's face in a worried expression and now I want to slightly lift the brows higher. With an **additive** mode to using the pose library, I can go to the \"brow raise\" pose and raise the brows w/out them loosing the worried shape. ", "Adjusting multiple properties at the same time doesn't work for all property types\nOperating system: Windows-10-10.0.18362-SP0 64 Bits\nGraphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 456.55\n\nBroken: version: 2.90.1\n\nMost properties in blender can be changed for all selected objects by holding ALT while changing it. However, this doesn't work for all properties. For instance, the 'Rig Type' property in the Rigify add-on doesn't work. We can make it work by right click and selecting 'Copy to selected' but it should be possible with the ALT shortcut for consistency.\n\n**Steps to reproduce**\nWhile in Pose Mode select multiple bones\ngo into Bone Properties -> Rigify Type -> Rig type\ntry to change the Rig Type for all of them by holding ALT", "GPencil: Shortcuts in Draw Mode are hardcoded\nMultiple keys in the `gpencil_draw_modal` function are hardcoded and (some) don't show up in the info bar. This makes it impossible for the user to know what certain keys do as well as impossible to change them in the keymap. All of those keys should be in the keymap with the default keys set as they are now.\n\nHere is a list of these (hardcoded) shortcuts:\n\n - {key Enter}, {key Pad_Enter}, {key Esc}, {key Space}, {key Ctrl Z} (?): Confirm\n - {key Mousewheel_Up}, {key Pad_Plus}: Increase size of eraser\n - {key Mousewheel_Down}, {key Pad_Minus}: Decrease size of eraser\n - {key O}: Set center of guide\n - {key V}: Turn off speed guide\n - {key M}: Alternate or flip guide direction\n - {key L}: Turn on line guide; {key Ctrl L}: Turn on line guide with angle = 0; {key Alt L}: Turn on line guide with last angle\n - {key C}: Turn on circular guide\n - {key K}: Turn line guide clockwise\n - {key J}: Turn line guide anti-clockwise\n", "Making 'singleuser' for a group with too smart NodeCustomGroup breaks user counters of both source and target nodes\nOperating system: Linux-5.0.0-37-generic-x86_64-with-debian-buster-sid 64 Bits\nGraphics card: GeForce GT 730/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 418.56\n\nBroken: version: 2.82 (sub 6)\nWorked: (optional)\n\nWhen node tries to copy its nodetree in `copy` api function, user counters get broken when a group with such node is singleusered.\n\nIt seems like the same issue as with cloning materials #72317\nMost likely it will also happen when singleusering a material.\n\n1. load attached blend file and execute embedded script -- the script registers 2 custom nodes, node foo uses shared nodetree, node 'bar' tries to copy nodetree\n2. navigate to compositor -- it already contains 2 copies of a group containing these nodes\n3. observe node groups in data api view\n```\n initial disposition: NodeGroup (2 users), foo (1 user), bar (1 user)\n```\n4. click 'make single user' on one of the groups\n3. observe node groups in data api view\n```\n expected: NodeGroup (1 user), NodeGroup.001 (1 user), foo (2 users), bar (1 user, source node), bar.dup (1 user, new node)\n observed: NodeGroup (1 user), NodeGroup.001 (1 user), foo (2 users), bar (0 users), bar.dup (2 users)\n```\n\n[test_group_singleuser.blend](test_group_singleuser.blend)\n", "Batch Rename Operator's lacking additional features for user-defined organisation of numerical (e.g., 001, 002) reassignment on the Bones in the Selection\nOperating system: Windows-10-10.0.22621-SP0 64 Bits\nGraphics card: AMD Radeon(TM) Graphics ATI Technologies Inc. 4.5.0 Core Profile Context 23.4.3.230420\n\nBroken: version: 3.3.7\nWorked: Never developed feature I guess.\n\nThis problem arises when we are trying to Batch-Rename a bigger Selection of Bones, and we want to assign numerical naming to them (such as Bone.001, Bone.002, Bone.003...), but, additionally, we want to logically order/reorder which Bone will start with \"001\" and which other Bones in 3D space will then get the other names \"002\", \"003\", etc.\n\nParticularly, when we Batch Rename the same Selection of Bones (all the Bones in given file) with the follow settings:\n**Selected: Bones\nType: Set Name\nMethod: New**\n**Name:** e.g., **Bone.001** or **Bone**\nand that we want to order these number names to 'swap' between the Bones in different, useful orders (for example, imagine, opening the file, that we just want to Batch Rename the Bones so that they are named Bone.001 to Bone.007, but on an inverse order: that is, in the 3D Viewport, from right to left instead of from left to right). So, what I've learned by accident is taht we can actually Batch Rename [Ctrl+F2] the same Bone Selection, with exactly the same settings, applied over and over again (iterations of the Batch Rename Operator); and this will make those number names to swap (and sometimes this is very helpful in [re]organising a project), however, if we keep doing that (iterations), what we see is the emergence of a strange cyclic pattern, that is actually very specific and hard to understand at first; unfortunately, this pattern is not that useful for multiple reasons in reorganisation, as it seems to be controlled only by (what I assume could be) the Bones' Indices (\"IDs\"), which are not very accessible nor very controllable, and we certainly do not have features in Batch Rename which relate in a clear and directive way to the Bone's Indices. Originally, I did expect Bone Parenting hierarchy, if any, would help in some sort, in terms of smart automation, of the Batch Rename feature; or, perhaps, the order in which the Bone Selection was made for a current use of the Batch Rename Operator would have an effect; or, at least, the Active element in that Bone Selection would have and influence. However, apparently, none of these factors seem to apply or be consistent: only the order in which the Bones were originally Added to the Armature Object [so, an ID/Index thing for the Bones?] seems to have an effect in the pattern which emerges from the iteration of the Batch Rename Operator on such settings for the same Bone Selection.\nMoreover (annother very peculiar and annoying limitation): the \"001\" will NEVER swap place! differently from the other number names. So, to give the name \"Bone.001\" to, for exemple, the Bone on the opposite side (which is not the 1st Bone with the earliest Index in my file), we can't just Batch Rename to get this job done (especially not, if there is already a Bone named \"Bone.001\" name in the Armature Object and that we want to keep this exact Name structure \"Bone.###\").\n\n1) Please, open the Blend file.\n2) Select All the Bones and Batch Rename with the following settings:\n**Selected: Bones\nType: Set Name\nMethod: New**\n**Name:** give something like **Bone.001** (could be just **Bone**) or **Test.001** (could be just **Test**)\n3) OK on the Operator, see how the Bones were Renamed;\n4) Once again, command another Batch Rename Operator [Ctrl+F2], keeping exactly the same settings (so this time just press OK);\n5) Observe the output;\n6) Keep re-iterating, and try to understand the pattern that shall emerge from it (is this really about Bones' Indices/ID only? Can we spot this data for Bones in the UI? Can't we swap Bones' Indices as well between themselves?). See if the Bones' Names 'swapping' could attend to different, some very specific reordering that we might wish to obtain in an automated way (like, in my example of an inverse ordering.\n7. So, in clonclusion: it took me months to come up with an hypothesis about the emerging pattern and to finally assume it might be relevant to Report it as... 'something odd and worth Reporting'. I understand it is probably not a bug after all, but originally it sounded very random. But most importantly I thinl: whatever logic the pattern is based upon, it is not a versatile pattern in relation to all the cool possibilities the Batch Remame Operator on a Multi-Selection can do; so it does look like it could require some assessment. Thank you for your attention and sorry for the excessive amount of information on such very specific use of the feature.\n\n", "macOS: touch input fails to work on a Wacom Intuos Pro tablet\nOperating system: macOS 11.1 Big Sur build 20C69\nGraphics card: n/a\n\nBroken: Top of tree git pull from today\nWorked: probably never since 2.8\n\nWacom Intuos Pro tablets support touch in addition to pen entry. When using touch, two fingers will scroll up and down. In Blender, two finger touch should work like an Apple touch pad or Apple magic mouse: two finger movement will orbit the view; holding down command zooms, holding down shift pans. Instead, it simply does nothing.\n\nBlender.app, default cube. Place two fingers down on a Wacom Intuos Pro tablet and move them around; it should orbit, but nothing happens. Well, the bottom status bar flickers back and forth as though command/shift or alt is pressed. \n\nThere are two reasons why this happens. \n1. Blender has incorrect Cocoa/AppKit code. I worked on AppKit for ~13 years at Apple. Below is a diff that addresses some of the issues. The code attempts to keep track of m_multiTouchScroll, but this isn't needed. It should key off of hasPreciseScrollingDeltas instead. I tested this diff with: Apple Magic Mouse, old school Apple scroll wheel mouse, Wacom tablet; all things work the same (except Wacom, which has the bug I'm reporting fixed). There is also a comment about intuos5, but that seems to be incorrect and outdated, but I don't have an Intuos 5 to test with.\n\n[0001-Fix-scrolling-with-wacom-tablets-using-them-as-a-tou.patch](0001-Fix-scrolling-with-wacom-tablets-using-them-as-a-tou.patch)\n\n2. Wacom's tablet touch drivers do link-checked behavior based on the bundle identifier \"org.blenderfoundation.blender\". In other words: they change the touch behavior when they see the target application is Blender. Wacom needs to fix this, but I'm not sure how to log a bug report for them.\n\nTo prove this, I created the sample application:\n\n[TabletTest.zip](TabletTest.zip)\n\nThe sample application has the blender bundle identifier set, as seen in this screen shot:\n\n\n\nIn ViewTest.m I log the flagsChanged:\n\n``` - (void)flagsChanged:(NSEvent *)event\n```\n{\n NSLog(@\"%@\", event);\n}\n```\n```\nIf you run the app and do a two finger scroll in the window you will see some flagsChanged logged, which is very strange because Shift/Alt/Cmd have not been pressed! Wacom is faking them for some reason, and changing the scrolling data, which messes up Blender.\n\nChange the applications's bundle identifier to \"org.blenderfoundationXX.blender\" (you can simply edit the info.plist) and run it again; it will then behave correctly.\n\n\n\nI don't know how to fix item 2 for Blender; I'm fixing it in my own build of Blender by changing the bundle identifier to org.blenderfoundation.bl3nder after it is built. I looked into identifying the events, and I can could identify the source as coming from the wacom driver but I can't fix them. \n\nI'm trying to find a contact at Wacom to have them fix this, but once it is fixed we still need item 1 fixed. \n\n\n\n", "Impossible to move with G key an object scale to 0 and parent to an empty\nOperating system: Windows 7\nGraphics card: GTX 780\n\nBroken: 3.5.1\nWorked: \n\nWhen an object is parented to an empty, and this object is scale to 0, you cannot move it anymore with the G key. But you can move it in the transform menu location.\n\nBased on the default startup or an attached .blend file (as simple as possible).\n1. Open blend.\n2. Try to move object with G key.\n3. It is possible with Transfom menu.\n\nThanks a lot !\ndono", "Bug on Rigid Body Simulation when an object is parented to another\nOperating system: Windows 10\nGraphics card: GTX 1060 6G\n\n2.80\n\nBug on Rigid Body Simulation when an object is parented to another\n\nPlease check the video......\n[BL280BUG.mp4](BL280BUG.mp4)\n\nMore, I have some questions on Rigid Body. I want to ask something, coding and algorithm, to the developers who work on it...\nHow can I reach them?\n\nThanks." ]

[ "Metaball names with digits in the end are invisible (also issue with renaming/duplicating metaballs)\nOperating system: Windows-7-6.1.7601-SP1 64 Bits\nGraphics card: GeForce GTX 660 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 430.86\n\n\nBroken: version: 2.80 (sub 74)\nWorked: (optional)\n\n\nIf you will try to use names like MetaballFamily2, Family will disappear. \nNames like Aaaa1aa and 1aaaaa — checked too, causes no problem\n[2019-07-11_19-49-29.mp4](2019-07-11_19-49-29.mp4)\n[untitled.blend](untitled.blend)\n\n\nMake metaball, add 1 to the family name\n" ]