About the Project
This project introduces a new system for 3D live streaming of sports events, using volumetric capture, Gaussian Splatting, and synthetic datasets rendered in real-time with Unreal Engine 5.
For more technical details, visit the official project page:
humansensinglab.github.io/basket-multiview/data/
Project Video
My Role
- Created synthetic datasets using Unreal Engine 5 and Maya.
- Built optimized 3D environments and procedural assets for multiview capture.
- Developed Python-based automation pipelines for data generation.
- Integrated cinematic sequences into real-time workflows.
Project Breakdown Videos
Cameras / Illumination Setup
The scene features a total of 89 strategically placed cameras arranged around the stadium, enabling comprehensive multiview capture from synchronized viewpoints.
The lighting setup replicates realistic basketball court illumination using physically accurate light sources.