OleehyO e084a4a270 feat: auto-extract first frames as conditioning images for i2v model ...

When training i2v models without specifying image_column, automatically extract
and use first frames from training videos as conditioning images. This includes:

- Add load_images_from_videos() utility function to extract and cache first frames
- Update BaseI2VDataset to support auto-extraction when image_column is None
- Add validation and warning message in Args schema for i2v without image_column

The first frames are extracted once and cached to avoid repeated video loading.

2025-01-07 06:43:26 +00:00

..

__init__.py

style: format import statements across finetune module

2025-01-07 05:54:52 +00:00

bucket_sampler.py

style: format import statements across finetune module

2025-01-07 05:54:52 +00:00

i2v_dataset.py

feat: auto-extract first frames as conditioning images for i2v model

2025-01-07 06:43:26 +00:00

t2v_dataset.py

style: format import statements across finetune module

2025-01-07 05:54:52 +00:00

utils.py

feat: auto-extract first frames as conditioning images for i2v model

2025-01-07 06:43:26 +00:00