LibreAvatrak

What LibreAvatrak is

LibreAvatrak is a free facial motion capture tool for VTubers and virtual avatar creators who use a single camera. It uses MediaPipe to detect face landmarks and calculate head pitch, yaw, and roll. It also measures eye openness with EAR and mouth movement with MAR, while OpenCV handles camera capture and image processing. Tracking data streams to Godot to animate VRM humanoid avatars in real time. A Python and Flask multiprocess backend provides an HTTP data API. The setup instructions call for a Python environment and Godot Editor 4.6, and describe loading a VRM file in Godot and saving the replacement character scene as .scn rather than .tscn. LibreAvatrak is licensed under the GNU Affero General Public License v3. Creators may earn revenue from content made with it, including streams and commercial productions. The project also states that modified instances or publicly accessible network services must share modified source code. It prohibits closing the source, selling packaged copies, and building paid cloud motion capture services based on the software.

Who it suits

LibreAvatrak suits individual creators who want to animate VRM avatars in Godot using facial tracking from one camera. Users should be comfortable setting up a Python environment and Godot Editor 4.6.

Limits

The software is free to use, but its license prohibits closing the source, selling packaged copies, and building paid cloud motion-capture services based on it. Modified instances and publicly accessible network services must share their modified source code.

Getting started

1. Open the LibreAvatrak project website on GitHub
2. Prepare a Python environment and Godot Editor 4.6
3. Load a VRM file in Godot
4. Save the replacement character scene as .scn rather than .tscn

Verdict

LibreAvatrak offers a focused camera-to-Godot workflow for facial animation, with support for VRM humanoid avatars. Review the setup requirements and license conditions before using or modifying it.