GitHub radar
Openjutsu: swap people or objects in your video on your own PC
Openjutsu replaces the main person in a video with a character you pick, for free — camera moves, cuts and sound stay untouched.
It's an open-source tool: upload a video and a character photo, and the person in the footage becomes that character.
When it helps
Good for a one-off character swap if you own a strong NVIDIA GPU; for frequent use, budget time for setup and rendering.
Pros
- Free, local alternative to the paid Higgsfield Genjutsu service
- Keeps the original camera movement, editing and soundtrack intact
- Supports multiple characters, object swaps, and a face-fix step
Cons
- Needs an NVIDIA GPU with 16+ GB VRAM — no Mac support
- Setup is involved: install ComfyUI, then about a dozen extra modules
- A 6-second clip takes 6–13 minutes to render, longer at high quality
How to set it up — step by step
- 1Confirm you have an NVIDIA GPU with 16+ GB of VRAM — the tool won't run without one
- 2Install and start ComfyUI, the open-source editor that runs these generative models, on your machine
- 3Open your coding agent (e.g. Claude Code) in the folder where you'll keep the project
- 4Send the agent the text from the block below
- 5Let the agent install the extra ComfyUI modules and model files, then start the app
- 6Open http://127.0.0.1:3000 in your browser, upload a video and a character photo, and press Generate
Text for your agent
Copy this and send it to your agent — Claude Code, Codex, any of them:
Set up Openjutsu from here: https://github.com/D3voz/openjutsu — I already have ComfyUI running with an NVIDIA GPU (16+ GB VRAM); follow the README: install the required extra ComfyUI nodes and model files, start the app, and open http://127.0.0.1:3000. Ask me if you need any model files or extra permissions along the way.
Heavy — needs an NVIDIA graphics card; the author tested it on an RTX 4060 Ti with 16 GB of VRAM and 64 GB of system RAM. Won't run on a Mac or without a dedicated NVIDIA card.
Open on GitHub▌ More finds