Welcome to the MetaHuman Engine documentation.
MetaHuman is a browser-native 3D digital human engine that provides a complete interaction loop: 3D avatar → voice → vision → dialogue.
| Capability | Technology |
|---|---|
| 3D Avatar | Three.js + React Three Fiber |
| Voice | Web Speech API + Edge TTS |
| Vision | MediaPipe Face Mesh & Pose |
| Dialogue | OpenAI-compatible API |
| Document | Description |
|---|---|
| Architecture | System design and data flow |
| API Reference | Backend API contract |
| Development | Setup and deployment guide |
- Zero-config by default — Works out of the box with local fallbacks
- Graceful degradation — External failures don't break the experience
- Modular architecture — Avatar, voice, vision, dialogue are independent
- Minimal re-renders — Focused state stores prevent unnecessary updates
- Browser-first — All processing happens client-side when possible
- Virtual customer service demos
- Interactive kiosks
- Educational assistants
- VTuber/streaming avatars
- Therapy companions