TMElyralab / MusePose

MusePose: a Pose-Driven Image-to-Video Framework for Virtual Human Generation
Other
2.29k stars 167 forks source link

Include `mmcv`s in requirements and update versions #53

Open jhj0517 opened 5 months ago

jhj0517 commented 5 months ago

Changed:

  1. Include mmcvs in requirements
  2. Update torch versions with CUDA 12.1
  3. Install different decord on MacOS.
  4. Update xformers to latest compatible version
  5. Update README to reflect changes.

Tested with

python pose_align.py --imgfn_refer ./assets/images/ref.png --vidfn ./assets/videos/dance.mp4
python test_stage_2.py --config ./configs/test_stage_2.yaml -W 512 -H 512

in venv and confirmed that it works fine.