中文版 | English Version
A private-deployable, lightweight AI video generation workbench built on Seedance 2.0. This Lite open-source version uses local SQLite3 for data persistence and supports multi-modal reference inputs with zero mandatory cloud dependencies. OSS is optional — reference images work out of the box without any cloud storage.
| Feature | Lite (Open Source) | Pro |
|---|---|---|
| Seedance 2.0 video generation | ✅ | ✅ |
| Conversation-based workflow | ✅ | ✅ |
| Media library | ✅ | ✅ |
| Local SQLite storage | ✅ | ✅ |
| OSS-free local mode (reference images) | ✅ | ✅ |
| Reference video / reference audio | Requires OSS | ✅ |
| Multi-user / Workspaces | ❌ | ✅ |
| User quota management | ❌ | ✅ |
| Supabase cloud database | ❌ | ✅ |
| ByteDance real-person asset library | ❌ | ✅ |
| REST API access | ❌ | ✅ |
Need the Pro version (multi-user, quota management, Supabase)? Contact: giordanoshai@gmail.com
Lite:https://sd-lite.smartled.cc non-api Pro:https://sd.xinrentong.com with seedance api
- 🚀 Model Support: Full support for Seedance 2.0 multi-modal protocols (image, video, audio references).
- 💾 Local Storage: Lightweight SQLite3 database — no complex infrastructure required.
- ☁️ OSS Optional: Without OSS, the app automatically switches to local mode — reference images are sent as base64 directly to the API. Zero barrier to getting started.
- 🖼️ Auto Preview: Automatically extracts keyframes and generates thumbnails for all generated videos.
- 🔄 Async Processing: Built-in background worker polls task status and saves results to local storage.
- ⚡ Lightweight: Single-user mode, no login required, ready out of the box.
- Python 3.12+
- FFmpeg installed and added to your system PATH (required for video metadata and thumbnail extraction).
git clone https://github.com/giordanoshai/seedance2.0_api_web_interface_Lite.git
cd seedance2.0_api_web_interface_Litepython -m venv .venv
# Windows
.venv\Scripts\activate
# macOS / Linux
source .venv/bin/activate
pip install aiosqlite fastapi httpx jinja2 oss2 pillow pydantic python-dotenv python-multipart uvicorn[standard]Copy .env.example to .env and fill in the required fields:
# Required: Volcengine API Key
SEEDANCE20_KEY=Your_Seedance2.0_Key
# Optional: Alibaba Cloud OSS (leave empty to use local mode)
OSS_KEY_ID=
OSS_ACCESSKEY=
OSS_BUCKET_NAME=| Mode | Trigger | Reference Image | Reference Video | Reference Audio |
|---|---|---|---|---|
| Local mode (default) | Leave OSS fields empty | ✅ sent as base64 | ❌ | ❌ |
| OSS mode | Fill in OSS_KEY_ID / OSS_ACCESSKEY / OSS_BUCKET_NAME | ✅ | ✅ | ✅ |
Modes are detected automatically — no manual switch needed.
python main.py
# or
uvicorn main:app --host 127.0.0.1 --port 8001 --reloadDefault URL: http://127.0.0.1:8001
├── app/
│ ├── routers/ # API Routes (task creation, status query, media lib)
│ ├── static/ # Frontend UI (HTML/JS/CSS)
│ ├── template/ # Jinja2 Templates
│ ├── config.py # Config module (auto OSS_ENABLED detection)
│ ├── database.py # SQLite3 interactions
│ ├── oss_client.py # Dual-mode storage (OSS / local + base64)
│ ├── task_worker.py # Background polling worker
│ └── volcano_api.py # Volcengine API wrapper
├── data/ # SQLite database
├── outputs/ # Local videos, thumbnails, and uploaded assets
├── screenshot/ # UI screenshots
├── main.py # Entry point
└── .env.example # Configuration template
If you find this project helpful, please give it a Star or submit a Pull Request.
This project is developed based on Volcengine and Seedance APIs. All generative rights belong to the respective creators. Please comply with relevant laws and platform terms during use.






