saganaki22/gnm-studio — explained in plain English
Analysis updated 2026-05-18
Animate a 3D avatar head live using your webcam for streaming or recording.
Try photo guided custom head fitting to build a personalized avatar shape.
Export recorded facial and head motion for use in Blender.
Run a fullscreen avatar output as a camera source inside OBS.
| saganaki22/gnm-studio | able-rip/cc-visionrouter | aisatan/poe2dire | |
|---|---|---|---|
| Stars | 29 | 29 | 29 |
| Language | JavaScript | JavaScript | JavaScript |
| Setup difficulty | easy | easy | easy |
| Complexity | 3/5 | 2/5 | 1/5 |
| Audience | vibe coder | developer | general |
Figures from each repo's GitHub metadata at analysis time.
Download the portable Windows executable and run it, or use the browser edition with no install at all.
GNM Studio is a desktop and web application that lets you create and animate a digital head using your webcam to capture facial movement. It runs as a portable Windows program built with Rust and Tauri, and a browser based version is also available so you can try it without installing anything. Both editions use MediaPipe, a face tracking tool from Google, to follow your expressions and turn them into movement on a 3D character. The app offers two avatar choices: the Google GNM head, a detailed mesh with thousands of vertices, separately colored eyes, teeth, and tongue, and an alternative called FaceCap, an MIT licensed avatar with 52 adjustable facial controls similar to those used in ARKit style systems. You pick between them from a card based selector, and individual expressions can be locked in place so they stay fixed while you adjust others. Getting tracking working involves a calibration step: you hold a neutral expression while the app checks your position, then it sets a baseline for your face and head orientation. From there, smoothing settings reduce jitter from tiny movements while still reacting quickly to real ones. There is also an experimental feature that can build a custom head shape by analyzing one or two photos of your face, with the analysis happening on your own machine rather than being uploaded anywhere. Other options include mirrored or transparent camera views, orbit and pan camera controls similar to those found in Blender, a fullscreen output mode useful for streaming with tools like OBS, and recording of motion and video that can be exported for later editing, including into Blender itself. An experimental skin material system with adjustable tones and texture maps is included as well, though it is turned off by default. The project runs locally: video and audio from your camera stay on your device instead of being sent to a server, in either the desktop app or the web edition, though the desktop version includes native performance features the browser build does not have. No signup or account is needed to use either version.
A local app that turns your webcam into a facial motion capture tool for animating 3D avatar heads.
Mainly JavaScript. The stack also includes Rust, Tauri, React.
Use freely for any purpose, including commercial use, as long as you keep the copyright notice.
Setup difficulty is rated easy, with roughly 5min to a first successful run.
Mainly vibe coder.
This repo across BitVibe Labs
double-check against the repo, no cap.