tigerowo /infinite-canvas

About infinite-canvas

Self-hosted canvas for planning shots and making AI images, video and audio. You connect nodes, run an agent, and build 3D scenes on a director's stage. It uses OpenAI-compatible APIs with your own keys. It's still in development, so data formats can change, and docs are Chinese-only.

Best for: Independent creators building a self-hosted canvas for AI images, video and scene setups with their own API keys.

• Drag, zoom and connect nodes on an infinite canvas with undo, redo and export
• Runs an agent that reads your nodes, plans a task and adds results back
• Places characters, models, panoramas and cameras in a 3D director's stage
• Writes camera, lens, focal length and aperture settings into your prompts
• Generates images through OpenAI-compatible Images and Responses APIs
• Reuses prompts and assets from a library built from open-source GitHub projects

1. 1
   Clone the repo and enter the folder: the command below, cd infinite-canvas

   git clone https://github.com/tigerowo/infinite-canvas.git

   Copy
2. 2
   Copy the env template and set your account and password: cp .env.example .env
3. 3
   Start it with Docker: docker compose up -d --build
4. 4
   Or run from source: go run ., then in another terminal cd web, bun install, bun run dev
5. 5
   Open http://localhost:3000; the prompt library admin page is at http://localhost:3000/admin/prompts

• infinite-canvas vs Toonflow-app