SysMocap is a real-time motion capture system that drives 3D virtual character animation from video input for VTuber, live streaming, augmented reality, and virtual reality applications.
The tool solves the problem of capturing performer movements and translating them into 3D character animation without specialized motion capture hardware. It uses video-driven pose estimation to detect body, hand, and facial movements in real time, then maps those movements onto imported 3D models. The system supports full-body, half-body, and facial-only capture modes, allowing flexibility in setup depending on available camera angles and performer positioning.
Developers and content creators building VTuber applications, live streaming setups, or AR experiences should consider this tool if they need accessible motion capture without expensive equipment. The project supports multiple 3D model formats including VRM files and Mixamo FBX models, with automatic skeleton detection for common formats and manual bone mapping for custom skeletons. It integrates with OBS for live streaming and supports WebXR API forwarding over HTTPS for web-based applications. The cross-platform availability on Windows, macOS, and Linux makes it accessible across different production environments. The drag-and-drop model import and Material Design 3 interface are designed to lower the barrier to entry for users without deep technical expertise.
Development activity shows consistent maintenance with automated testing through GitHub Actions and regular releases. The project maintains multilingual support including English and Chinese interfaces, indicating attention to international user needs. The codebase is written in JavaScript, making it accessible to web developers and allowing potential community contributions in a widely-used language.
