What model does this workflow use to remove watermarks?

It uses an inpainting pipeline guided by the LTX 2.3 LoRA. Place the LTX 2.3 LoRA in models/loras and ensure a compatible base checkpoint is available in models/checkpoints. The custom processing node (UUIDs 3b019976-c705-40af-8f45-081975d72105 and 83cfc570-fff1-4869-8c43-d39ef189634b) handles applying the LoRA across video frames.

Will it work on moving or animated watermarks?

It’s optimized for persistent, static watermarks (e.g., corner logos or date stamps). Moving, animated, or large-area overlays are harder to inpaint consistently and may require parameter tuning, pre-masking, or manual cleanup.

Does the output keep the original audio?

SaveVideo typically writes the processed video stream only. If you need the original audio, remux it afterward using a video editor or a command like: ffmpeg -i clean.mp4 -i source.mp4 -map 0:v -map 1:a -c copy final.mp4.

How can I improve results on tough watermarks?

Try moderately increasing the LoRA weight and denoise strength, downscaling slightly to reduce processing noise, and ensuring the watermark region is fully covered by the inpainting mask settings in the custom node. For very stubborn cases, process in segments around scene cuts.

Remove Watermark from Video - LTX 2.3 LoRA

This ComfyUI workflow removes persistent watermarks from a video by inpainting over them frame by frame with a LoRA-enhanced diffusion pass. It starts with LoadVideo to ingest your clip and GetVideoComponents to extract frames, dimensions, and timing info. The core processing is handled by two custom nodes (UUIDs 3b019976-c705-40af-8f45-081975d72105 and 83cfc570-fff1-4869-8c43-d39ef189634b), which apply an LTX 2.3 LoRA-guided inpainting step across the sequence to reconstruct clean background detail where the watermark sits. Finally, SaveVideo reassembles the processed frames into a new video file.

Technically, the pipeline performs video-to-video inpainting using a Stable Diffusion-style model augmented by the LTX 2.3 LoRA adapter to bias reconstruction toward plausible content while suppressing the watermark. The custom processing stage maintains per-frame consistency and uses mask-driven edits so only the marked area is regenerated, preserving the rest of the frame. Because the work happens at the frame level and is driven by inpainting, the result avoids typical blur or smudge artifacts of traditional filters and holds up better on semi-transparent, corner logos. It’s a practical, one-click solution for cleaning owned or licensed footage without manual rotoscoping.

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.

fedbcfb05b08.json

Fetching workflow JSON…

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

// Install (beta)
npm i @comfyorg/sdk

// Run this workflow (TypeScript)
import { Comfy } from "@comfyorg/sdk";

const client = new Comfy({ apiKey: "comfyui-..." });
const wf = await client.workflows.fromFile("workflow_api.json");
const job = await client.run(wf);
await job.getOutputs("<output-node-id>")[0].toFile("output.png");

The SDK takes a workflow in API format: open this workflow in ComfyUI and use File → Export Workflow (API).

Comfy Cloud API access requires a plan with an API key.

SDK docs Get an API key

FAQ

Frequently Asked Questions