A beginner-friendly AI creative workspace combining an infinite canvas, local generation engines, cloud creation, model management, and persistent projects.
The product direction and first-release scope are approved. The repository contains an early prototype; the target capabilities below are not claims of completed implementation. See the documentation and milestone acceptance gates before assessing readiness.
Double-click Berry.bat in the repository root:
.\Berry.batLaunches the standalone desktop application window (powered by Tauri & WebView2). No web browser tabs or command windows required.
# Run Tauri desktop app with live reload:
pnpm --prefix frontend tauri dev
# Or run frontend and backend independently:
cd backend && .\.venv\Scripts\python -m uvicorn app.main:app --reload
cd frontend && pnpm dev- Windows release with cross-platform code architecture.
- NVIDIA local generation and cloud-only creation with user-provided API keys.
- Optional isolated ComfyUI and Stable Diffusion WebUI installation and existing-engine connections.
- Berry's own infinite canvas and creation controls, with native engine entry points retained.
- Text-to-image, image-to-image, inpainting, and upscaling.
- Local model scanning/import, directory management, compatibility and dependency guidance.
- Persistent projects, assets, generation records, and reliable task execution.
Video, Agent features, clarified CLI integration, non-NVIDIA discrete GPU inference, and additional OS releases belong to later phases.
| Document | Purpose |
|---|---|
| Product Vision | Approved positioning and product boundaries |
| First-Release PRD | Requirements and release acceptance scenarios |
| Architecture | Target technical design and invariants |
| Roadmap | Implementation milestones and evidence gates |
| Product Alignment | Decision history and prototype inspection findings |
| Coding Agent Handoff | Copyable implementation prompt |
The prototype uses React, TypeScript, Vite, Tailwind, Zustand and React Flow on the frontend, and FastAPI/Pydantic on the backend. Generation engines remain optional and isolated from the lightweight core.
For development setup, see CONTRIBUTING.md. The setup instructions require validation against the implementation baseline; no build or runtime verification was performed during product planning.
Branch from dev and target dev with pull requests. Never commit directly to main. Use English for all project-facing written artifacts and Conventional Commits for commit messages.
Read AGENTS.md, Branching Strategy, Contributing, and Code of Conduct.
Apache License 2.0. See LICENSE.