anymouschina /TapCanvas

About TapCanvas

TapCanvas lays your script, characters, storyboards, images, video and audio on one canvas. Agents read the real project state, pick tools, and run multi-step production. It's built for AI short dramas and comic dramas with a traceable pipeline. You set up your own model keys, and one-click film isn't live yet.

Best for: Comic drama teams that want a traceable canvas pipeline for AI short dramas.

• Organizes documents, characters, storyboards, images, video and audio as canvas nodes
• Runs multi-step production through agents that read your project and canvas state
• Routes text, image, video and audio models through Luban API, a built-in model gateway
• Keeps versions, generation history, task status and delivery records for each asset
• Reuses DAG workflows (a graph of steps) with async workers and failure recovery
• Previews 3D scenes, character blocking and camera moves in the director desk

1. 1
   Install Docker, Docker Compose and OpenSSL. You don't need Node.js, pnpm, Bun or Go on your machine.
2. 2
   Clone the repo: the command below

   git clone https://github.com/anymouschina/TapCanvas.git

   Copy
3. 3
   Enter the folder: cd TapCanvas
4. 4
   Start everything: ./scripts/dev.sh docker (add --fresh-build to ignore build caches and re-pull base images)
5. 5
   Open the web UI at http://localhost:5175, the API at http://localhost:8788 and the Luban API console at http://localhost:4455
6. 6
   Add your own model keys in the Luban API console before you generate anything.

• TapCanvas vs Toonflow-app