ddcat-ai /open-ai-canvas

About open-ai-canvas

Yingce (影策) is a free canvas for planning AI films and short dramas. You lay out scripts, characters, scenes and storyboards as nodes. Text, image, video and audio jobs run from one task list. It's self-hosted and still changing fast, and documentation is only in Chinese.

Best for: Hobbyist directors and small teams running their own canvas workspace for AI short dramas.

• Draws nodes, links, zoom, minimap, undo/redo, import/export and read-only sharing
• Keeps scripts, characters, scenes, style boards, reference assets and storyboards
• Runs text, image, video and audio jobs with reference images and batch runs
• Queues jobs with progress, logs, cancel/retry, an asset library and cross-device sync
• Cuts clips on a timeline, splits, trims, transcribes subtitles and exports the final video
• Manages channels, models, credits, feature switches and object storage from an admin panel

1. 1
   Clone the repo and make local data folders: the command below, cd open-ai-canvas, the next command

   git clone https://github.com/ddcat-ai/open-ai-canvas.git

   Copy

   mkdir -p .local/project-workbench-debug .local/cache/go-build .local/cache/go-mod

   Copy
2. 2
   Start the backend in one terminal: cd backend, then go run ./cmd/server with the env vars shown in the README
3. 3
   Start the frontend in a second terminal: cd ../web, the command below, bun run dev

   bun install --frozen-lockfile

   Copy
4. 4
   Open http://localhost:3000, register the first admin account and set up model channels in settings
5. 5
   On Windows PowerShell, run ./scripts/start-local.ps1 from the repo root instead
6. 6
   For Docker, run the command below or the next command

   docker compose -f docker-compose.dev.yml up --build

   Copy

   docker compose -f docker-compose.local.yml up -d --build

   Copy