How should I write the prompt for best results?

Keep it empty or describe only the background inside the masked area (e.g., “asphalt road with painted lines, soft shadow”). Avoid mentioning the removed object or unrelated foreground items—this can cause duplication or hallucinations.

My edges flicker or show seams. What should I tweak?

Increase LTXVDilateVideoMask size to cover motion blur and expand beyond the object, then raise GrowMaskWithBlur feather radius to soften transitions. Also re‑seed SAM3 on a clearer frame and re‑propagate if the tracked mask drifts.

Can I test a short section before processing the full video?

Yes. Use VHS_GetImageCount to see the total frames and GetImageRangeFromBatch (optionally with ComfyMathExpression) to select a small frame window for quick iterations, then switch back to the full range once satisfied.

Does the saved video include audio?

SaveVideo typically outputs a silent file. If you need audio, re‑mux the original track in your NLE or a tool like ffmpeg. If your setup routes audio via GetVideoComponents, you can also handle pass‑through externally after render.

LTX Cleanplate for VFX - ComfyUI Workflow

This ComfyUI workflow builds a clean background plate by removing an object from video using SAM3-driven masks and a mask-conditioned LTX-2.3 In‑Outpainting IC‑LoRA. The pipeline loads your clip with VHS_LoadVideo/VHS_VideoInfo, extracts frames via GetVideoComponents, and uses LoadSAM3Model → SAM3VideoSegmentation to seed a mask for the object you want to remove. SAM3Propagate tracks that mask across the shot, and SAM3VideoOutput provides a per‑frame OBJECT_MASK. To ensure seamless fills, LTXVDilateVideoMask expands the mask (e.g., s10/t1) and GrowMaskWithBlur feathers the edges.

The masked region is then green‑filled by LTXVInpaintPreprocess, and CreateVideo re‑encodes those preprocessed frames. That green plate, together with the OBJECT_MASK and your minimal background‑only prompt, feeds the LTX 2.3 In‑Outpainting IC‑LoRA (packaged here as a subgraph node), which is explicitly mask‑conditioned to generate only inside the masked area. Utility nodes like VHS_GetImageCount, ComfyMathExpression, and GetImageRangeFromBatch let you preview limited frame ranges before a full run. Finally, SaveVideo writes the cleanplate result for downstream compositing.

API

Use this workflow from code

Every Comfy workflow is a JSON graph. The payload below is this workflow, exactly as ComfyUI runs it — fetch it from the URL, keep it in version control, or load it in ComfyUI and run it node by node.

GET https://comfy.org/workflows/download/8f2cf0df5da6.json

Fetching workflow JSON…

Run it from Python or TypeScript with the Comfy SDK. The same code targets Comfy Cloud or a ComfyUI you host yourself — only the base URL changes.

# Install (beta)
pip install comfy-sdk        # Python
npm i @comfyorg/sdk          # TypeScript

# Run this workflow (Python)
from comfy_sdk import Comfy

client = Comfy(api_key="comfyui-...")
wf = client.workflows.from_file("workflow_api.json")
job = client.run(wf)
for output in job.get_outputs("<output-node-id>"):
    output.to_file(output.name)

The SDK takes a workflow in API format: open this workflow in ComfyUI and use File → Export Workflow (API). API access requires a Comfy Cloud plan with an API key. SDK docs · Get an API key

FAQ

Frequently Asked Questions

View all workflows