Completed upscaling all 10 smallest 300VW videos to 4K resolution (2560x1440): | Video | Size | Frames | Original | |-------|------|--------|----------| | 520.mp4 | 23 MB | 814 | 520/vid.avi | | 404.mp4 | 28 MB | 999 | 404/vid.avi | | 521.mp4 | 28 MB | 1022 | 521/vid.avi | | 144.mp4 | 33 MB | 1236 | 144/vid.avi | | 044.mp4 | 36 MB | 1309 | 044/vid.avi | | 025.mp4 | 38 MB | 1404 | 025/vid.avi | | 213.mp4 | 39 MB | 1454 | 213/vid.avi | | 518.mp4 | 40 MB | 1503 | 518/vid.avi | | 550.mp4 | 41 MB | 1561 | 550/vid.avi | | 562.mp4 | 43 MB | 1367 | 562/vid.avi | **Total: 346 MB** of 4K upscaled video content All videos upscaled using Real-ESRGAN x4 from original 640x360 resolution. Git LFS tracking configured via .gitattributes. Merged to master in commit ff6e70dab.

## Feature Triage Complete **Status:** NOT IMPLEMENTED - Ready for implementation **Summary:** This feature involves upscaling 10 smallest 300VW videos to higher resolution (~2.5K with 4x upscale). All dependencies are satisfied: - `upscaler.py` (FACE-10) is complete with video upscaling support - Git LFS is already configured for `.avi` files - Source face images already upscaled (4K, 8K versions exist) **10 Target Videos Identified:** `520`, `521`, `404`, `025`, `213`, `518`, `144`, `562`, `550`, `044` - Total: 26.78 MB input, ~13,384 frames - Estimated output: 100-200 MB (H.264 @ CRF 18) - Processing time: ~40-50 minutes on CUDA **Implementation Location:** - Create `evaluation/300vw-4k/` directory - Output as `.mp4` (not `.avi`) for better compression - Update `.gitattributes` for new LFS tracking **Clarification Needed:** The ticket mentions '4K (2560x1440 or 3840x2160)' but 4x upscaling produces ~2552x1440 (2.5K) for 638x360 sources. Consider renaming directory to `300vw-upscaled` for accuracy. **Complexity:** Low - straightforward script running existing `upscaler.py` See triage report: `/tmp/triage-FACE-12.md`