Skip to content

Conversation

@code-scrollright
Copy link

No description provided.

- Updated requirements.txt for Python 3.11 compatibility
  - Changed torch from ==1.13.1 to >=2.0.0
  - Changed torchvision from ==0.14.1 to >=0.15.1
  - Updated other packages to use minimum version requirements for flexibility

- Added automatic CUDA/CPU device detection in app.py
  - Auto-detects if CUDA is available, falls back to CPU if not
  - Uses appropriate dtype (float16 for CUDA, float32 for CPU)
  - Enables the project to run on systems without NVIDIA GPU

- Added SETUP_NOTES.md with detailed setup instructions and troubleshooting

- Added test_setup.py script to verify installation and dependencies

These changes make the project compatible with:
  - Python 3.11 (and 3.9, 3.10)
  - Systems with CUDA GPU
  - Systems without CUDA (CPU-only)
- Updated diffusers from 0.14.0 to >=0.25.0 to fix compatibility with newer huggingface_hub
- Made annotators (pose/edge/depth control) optional in utils.py
  - Wrapped annotator imports in try-except to handle basicsr build issues on Python 3.11
  - Added graceful degradation when annotators are not available
  - Core text-to-video functionality works without annotators

- Updated SETUP_NOTES.md with:
  - Known limitations on Python 3.11
  - Clear documentation of which features are available
  - Test results showing successful initialization

The project now successfully initializes on Python 3.11 with CPU-only mode.
Core features (text-to-video, video instruct-pix2pix) are working.
Pose/Edge/Depth control features require Python 3.9 due to basicsr dependency.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants