- Add optional MASK input/output to FaceWrapper node
- Implement mask deformation for all modes (Debug/Un-Wrap/Wrap)
- Add mask conversion methods between tensor and numpy formats
- Process mask using same landmarks as main image
- Return untouched mask in Debug mode
- Return empty mask when no face detected
Core changes:
- Normalized boundary points to 0-1 range in MediapipeBaseLandmarks
- Fixed scaling inconsistencies between landmarks and boundary points
- Updated edge coordinates to use normalized values
- Improved y_translation handling for boundary landmarks
Files modified:
- core/base_landmarks.py