OpenJutsu

What is OpenJutsu?

OpenJutsu is an open-source tool that replaces people or objects in videos using artificial intelligence. It runs locally on your own GPU and works as a local alternative to cloud services like Higgsfield Genjutsu. The application combines a Next.js interface with a ComfyUI workflow to handle video processing.

Users upload a video and a reference image, then generate a new version where the main subject is swapped with the provided character. The tool preserves the original camera movement, editing cuts, motion, and audio track. It uses MiniMax H3 for video generation, SAM 3.1 for person tracking, and Qwen3-VL for describing scenes and writing prompts.

The software offers simple and advanced motion transfer modes, allowing single or multiple character replacements per shot. It also supports object swapping, background changes, face fixing, and relighting from a reference picture. All models run on open weights, and users can select different vision-language models based on their available VRAM.

OpenJutsu features

• One-click video character replacement using reference images
• Preserves original camera movement, cuts, motion, and audio
• Supports simple and advanced multi-character swap modes
• Object swapping, background removal, and face correction
• Local processing with MiniMax H3, SAM 3.1, and Qwen3-VL
• Configurable vision-language models for prompt generation

Who uses OpenJutsu and for what

• Replace actors in indie films or student projects without reshooting
• Swap clothing or hairstyles on video subjects for quick edits
• Remove unwanted objects or change backgrounds in recorded footage
• Generate consistent character appearances across multiple video clips
• Relight video scenes using lighting references from still photographs

OpenJutsu FAQ

Comments (0)

Comments and upvotes open when this product launches on ProductPerch.