video-inpaint-remove
Installation
SKILL.md
video-inpaint-remove
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.
Install the node pack (once)
Copy assets/openframe-nodes/ into ComfyUI/custom_nodes/openframe/. It provides OpenFrameVideoInput, SaveVideo, GreenFill and PadForRefinement, and is MIT, from satianurag/openframe. Mask tracking uses SAM 2, so its weight license applies.
| Need | Workflow | Set |
|---|---|---|
| Light and fast (Wan VACE 1.3B) | wan_vace_inpaint_remove.api.json |
source.filename, source.mask_filename (a white = remove mask on the first frame), positive.text = what should be there instead, sample.seed |
| Higher quality (LTX-2.3) | ltx23_inpaint_remove.api.json |
source.filename, source.mask_filename, positive.text, noise.noise_seed |
Rules
- Describe the clean result ("an empty marble table in sunlight"), not the removal itself.
- Keep
preserve_outside_maskset to true so untouched pixels stay identical to the source. - Check the result with
still-qaon 3 frames. The method notes are inreferences/openframe-workflows-README.md.