Open-source pipeline: Quest 3 room capture -> Gaussian Splat -> VRChat world
Capture any real physical space with a Meta Quest 3 headset and import it as a photorealistic 3D Gaussian Splat scene into VRChat. VRScan automates the full pipeline so VRChat world creators don't need to touch a terminal.
Meta Quest 3 (wearing headset, walking room)
| OpenQuestCapture -- captures RGB, depth, pose -> exports .zip
| quest-3d-reconstruction -- converts to COLMAP sparse point cloud
| LichtFeld Studio -- trains 3DGS model -> outputs point_cloud.ply
| ALCOM + VRChatGaussianSplatting Unity package
| VRChat world upload via VRChat SDK
- Meta Quest 3 headset
- Windows PC with NVIDIA GPU (RTX series recommended, driver 570+)
- ALCOM -- open-source VRChat package manager
- Unity (version recommended by VRChat SDK)
- Python 3.10+ and Miniconda
- Git
Start with Issue #1 -- Dev Environment Setup
Each issue contains a numbered checklist. Complete every task in order before moving to the next issue. All 8 issues form the complete pipeline.
This project connects the following open-source tools:
- OpenQuestCapture by samuelm2 -- MIT
- quest-3d-reconstruction by samuelm2 -- MIT
- LichtFeld Studio by MrNeRF -- GPLv3
- VRChatGaussianSplatting by MichaelMoroz
- ALCOM by anatawa12 -- MIT
MIT -- see LICENSE