video-upscale
Installation
SKILL.md
video-upscale
Run with python scripts/comfy_client.py --url $COMFY_URL run workflows/<file> --set NODE.INPUT=VALUE --out out/, and upload inputs first with comfy_client.py upload <file>. Before the first run, check the server with python scripts/preflight.py --url $COMFY_URL workflows/<file>. Node packs, model files, VRAM and weight licenses are listed in references/requirements.md. Use open-weights models only. LTX weights are under the Lightricks LTX license, which is not an OSI-approved open-source license.
| Need | Workflow | Set |
|---|---|---|
| Upscale a video clip | seedvr2_video_upscale.api.json |
1.file, 109.seed. Read the target size from the SeedVR2 preprocess node, and see references/rosesfield-upscale-video-README.md |
| Upscale a still | seedvr2_image_upscale.api.json |
1.image, 54.seed |
Rules
- Only upscale the approved take, as the last step on each shot. Never upscale before
still-qaorcontinuity-check. - Upscale every shot in a sequence the same way, so the grain and sharpness match across cuts.
- The graph uses tiled VAE encode/decode, so reduce the tile size if you run out of memory. The SeedVR2 weight license is unverified.