up

README.md

@@ -73,6 +73,11 @@ VMS uses `Finetrainers` under the hood. In theory any model supported by Finetra
 
 In practice, a PR (pull request) will be necessary to adapt the UI a bit to accomodate for each model specificities.
 
+
+### Wan
+
+I am currently testing Wan LoRA training!
+
 ### LTX-Video
 
 I have tested training a LTX-Video LoRA model using videos (not images), on a single A100 instance.
@@ -90,6 +95,22 @@ Do you want support for this one? Let me know in the comments!
 
 ## Limitations
 
+### No AV1 on A100
+
+If your dataset contains videos encoded using the AV1 codec, you might not be able to decode them (eg. during scene splitting) if your machine doesn't support hardware decoding.
+
+Nvidia A100 don't support hardware AV1 decoding for instance.
+
+It might be possible to convert them on server-side or use software decoding directly from Python, but I haven't looked into that yet (you can submit a PR if you have an idea).
+
+My recommendation is to make sure your data comes in h264.
+
+You can use FFmpeg to do this, eg:
+
+```bash
+ffmpeg -i input_video_in_av1.mp4 -vcodec libx264 -acodec aac output_video_in_h264.mp4
+```
+
 ### One-user-per-space design
 
 Currently CMS can only support one training job at a time, anybody with access to your Gradio app will be able to upload or delete everything etc.

vms/config.py

@@ -41,7 +41,7 @@ DEFAULT_PROMPT_PREFIX = "In the style of TOK, "
 # This is only use to debug things in local
 USE_MOCK_CAPTIONING_MODEL = parse_bool_env(os.environ.get('USE_MOCK_CAPTIONING_MODEL'))
 
-DEFAULT_CAPTIONING_BOT_INSTRUCTIONS = "Please write a full video description.  Be synthetic, don't say things like ""this video features.."" etc. Instead, methodically list camera (close-up shot, medium-shot..), genre (music video, horror movie scene, video game footage, go pro footage, japanese anime, noir film, science-fiction, action movie, documentary..), characters (physical appearance, look, skin, facial features, haircut, clothing), scene (action, positions, movements), location (indoor, outdoor, place, building, country..), time and lighting (natural, golden hour, night time, LED lights, kelvin temperature etc), weather and climate (dusty, rainy, fog, haze, snowing..), era/settings."
+DEFAULT_CAPTIONING_BOT_INSTRUCTIONS = "Please write a full video description. Be synthetic and methodically list camera (close-up shot, medium-shot..), genre (music video, horror movie scene, video game footage, go pro footage, japanese anime, noir film, science-fiction, action movie, documentary..), characters (physical appearance, look, skin, facial features, haircut, clothing), scene (action, positions, movements), location (indoor, outdoor, place, building, country..), time and lighting (natural, golden hour, night time, LED lights, kelvin temperature etc), weather and climate (dusty, rainy, fog, haze, snowing..), era/settings."
        
 # Create directories
 STORAGE_PATH.mkdir(parents=True, exist_ok=True)

vms/tabs/monitor_tab.py

@@ -44,7 +44,7 @@ class MonitorTab(BaseTab):
     def __init__(self, app_state):
         super().__init__(app_state)
         self.id = "monitor_tab"
-        self.title = "4️⃣  Monitor"
+        self.title = "5️⃣  Monitor"
         self.refresh_interval = 8
     
     def create(self, parent=None) -> gr.TabItem:

vms/tabs/split_tab.py

@@ -32,7 +32,7 @@ class SplitTab(BaseTab):
                 with gr.Column():
                     self.components["video_list"] = gr.Dataframe(
                         headers=["name", "status"],
-                        label="Videos to split",
+                        label="Videos to split (note: Nvidia A100 cannot split videos encoded in AV1)",
                         interactive=False,
                         wrap=True
                     )