Implement standalone PyTorch-based RIFE interpolation that runs in a
dedicated virtual environment to avoid Qt/OpenCV conflicts:
- Add PracticalRifeEnv class for managing venv and subprocess execution
- Add rife_worker.py standalone interpolation script using Practical-RIFE
- Add RIFE_PRACTICAL blending model with ensemble/fast mode settings
- Add UI controls for Practical-RIFE configuration
- Update .gitignore to exclude venv-rife/ directory
The implementation downloads Practical-RIFE models on first use and runs
interpolation in a separate process with proper progress reporting.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Split monolithic symlink.py into modular components:
- config.py: Constants and configuration
- core/: Models, database, blender, manager
- ui/: Main window and widgets
New features included:
- Cross-dissolve transitions with multiple blend methods
- Alpha blend, Optical Flow, and RIFE (AI) interpolation
- Per-folder trim settings with start/end frame control
- Per-transition asymmetric overlap settings
- Folder type overrides (Main/Transition)
- Dual destination folders (sequence + transitions)
- WebP lossless output with compression method setting
- Video and image sequence preview with zoom/pan
- Session resume from destination folder
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
PyQt6 application to create sequenced symlinks for image files.
Features:
- GUI and CLI modes
- Multiple source folder support with drag & drop
- SQLite database for tracking symlink sessions
- Multi-select and reorder files
- Two-column file view (filename + source path)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
