Commit History

  1. Fix: increase generation timeout from 10 to 20 minutes for T4 with CPU offload

    ce07a6b
  2. Fix: remove invalid --guide_scale arg from generate.py command

    59ab431
  3. Fix: add diffusers, transformers, accelerate, tokenizers to explicit pip install

    1b2c013
  4. Fix: move generate.py patch into standalone Python script, remove inline -c

    2c40a1b
  5. Fix: install ALL deps + patch both wan/__init__.py and generate.py for optional imports

    e99b3c1
  6. Fix: patch only problematic submodule imports (speech2video, animate)

    694d214
  7. Fix: use external Python script for patching wan/__init__.py (no shell escaping)

    55549d3
  8. Fix: simple regex-based try/except wrapper for all wan submodule imports

    b1025e1
  9. Fix: comprehensive deps + monkey-patch wan/__init__.py to tolerate missing optional modules

    57fa6f9
  10. Fix: add librosa (transitive dep from speech2video import chain)

    11e2761
  11. Fix: add decord + all Wan2.2 deps for proper import chain

    62fce73
  12. Fix: add missing einops dependency

    0628ecf
  13. Fix: install Wan2.2 deps directly (skip flash_attn/dashscope which fail in Docker build)

    61027c5
  14. Fix: download model at startup via huggingface-cli, remove build-time download

    3d1b476
  15. Switch to official Wan2.2 inference code for native T2V+I2V support

    40328c3
  16. Fix: enable CPU offload + attention slicing for T4 (14.75 GB VRAM limit)

    eb6e5e7
  17. Fix: total_mem → total_memory for PyTorch 2.8 compatibility

    4f91cdd
  18. Fix build: add git dependency, install diffusers from source separately

    38951d1
  19. Upgrade UI: gallery, progress tracking, tabs, image preview, polished design

    b7691c7
  20. Add Wan2.2 TI2V 5B FastAPI app: text-to-video + image-to-video

    d47df1b
  21. initial commit

    Verified signature
    8fab102