samples

Browse files

Files changed (19) hide show

README.md +6 -2
images/00150-1399128505_20221221173435.png +3 -0
images/00196-1101562705_20221231155533.png +3 -0
images/00198-2068722232_20221224172330.png +3 -0
images/00246-2352187649_20221231160654.png +3 -0
images/00291-2873121500_20221231161640.png +3 -0
images/00337-3622700117_20221226124912.png +3 -0
images/00364-2593184067_20221224175352.png +3 -0
images/00417-1486510992_20221224081809.png +3 -0
images/00635-3305501895_20221231190252.png +3 -0
images/00652-957333324_20221231190955.png +3 -0
images/00881-3244860476_20221226151114.png +3 -0
images/00976-3769766671_20221226155509.png +3 -0
images/00984-60167261_20221226155731.png +3 -0
images/01350-1074110064_20230101173929.png +3 -0
images/01438-2194281854_20221230190612.png +3 -0
images/01450-118577626_20221230190741.png +3 -0
images/01706-3603346036_20230101182815.png +3 -0
images/01874-4242434330_20221231083451.png +3 -0

README.md CHANGED Viewed

@@ -18,7 +18,9 @@ widget:
 ## Update
-2023-01-02: I wasted more GPU hours to train it a little bit more (overfitting). Check out [bp_mk3.safetensors](bp_mk3.safetensors) and [bp_mk5.safetensors](bp_mk5.safetensors). Prepare yourself own VAE! Update your WebUI if you can't load [safetensors](https://github.com/huggingface/safetensors).
 ## Introduction
@@ -86,7 +88,7 @@ is still applied.
 > SD cannot generate human body properly, like generating 6 fingers on one hand.
-BPModel can generate more proper kitty cat (if you know what I mean) than other anime model,
 but it's still not perfect. As results presented in [Diffusion Art or Digital Forgery? Investigating Data Replication in Diffusion Models](https://arxiv.org/abs/2212.03860), the copy and paste effect is still observed.
 Anything v3™ has been proven to be the most popular anime model in the community, but it's not perfect either as
@@ -165,6 +167,8 @@ Negative prompt: (bad:0.98), (normal:0.98), (comic:0.81), (cropped:0.81), (error
 Steps: 40, Sampler: Euler a, CFG scale: 8, Seed: 2668993375, Size: 960x1600, Model hash: fed5b383, Batch size: 4, Batch pos: 3, Denoising strength: 0.7, Clip skip: 2, ENSD: 31337, First pass size: 0x0
 ```
 ## Usage
 The [`bp_1024_e10.ckpt`](bp_1024_e10.ckpt) doesn't include any VAE and you should using other popular VAE in the community when using with [AUTOMATIC1111/stable-diffusion-webui](https://github.com/AUTOMATIC1111/stable-diffusion-webui) or you would see the

 ## Update
+2023-01-02: I wasted more GPU hours to train it a little bit more overfitting. Check out [bp_mk3.safetensors](bp_mk3.safetensors) and [bp_mk5.safetensors](bp_mk5.safetensors). Prepare yourself own VAE! Update your WebUI if you can't load [safetensors](https://github.com/huggingface/safetensors). [Adds lots of samples!](images)
+![car](images/00976-3769766671_20221226155509.png)
 ## Introduction
 > SD cannot generate human body properly, like generating 6 fingers on one hand.
+BPModel can generate [more proper kitty cat](https://twitter.com/crosstyan/status/1606026536246685696) (if you know what I mean) than other anime model,
 but it's still not perfect. As results presented in [Diffusion Art or Digital Forgery? Investigating Data Replication in Diffusion Models](https://arxiv.org/abs/2212.03860), the copy and paste effect is still observed.
 Anything v3™ has been proven to be the most popular anime model in the community, but it's not perfect either as
 Steps: 40, Sampler: Euler a, CFG scale: 8, Seed: 2668993375, Size: 960x1600, Model hash: fed5b383, Batch size: 4, Batch pos: 3, Denoising strength: 0.7, Clip skip: 2, ENSD: 31337, First pass size: 0x0
 ```
+more samples can be found in [images](images)
 ## Usage
 The [`bp_1024_e10.ckpt`](bp_1024_e10.ckpt) doesn't include any VAE and you should using other popular VAE in the community when using with [AUTOMATIC1111/stable-diffusion-webui](https://github.com/AUTOMATIC1111/stable-diffusion-webui) or you would see the