Funded by UT Dallas Bass School Collaborative Research Grant
Awarded research grant funding in December 2025 ('Advancing Research and Pedagogy in Volumetric Video through Student Collaboration') to engineer real-time streaming pipelines for volumetric performance capture.
The Volumetric Video Capture & Playback Pipeline is an ongoing research and development initiative at the VAST Lab investigating cutting-edge techniques to ingest, compress, and render 4D holographic video captures of human performers in real-time game engines.
Evercoast Integration & Multi-Camera Ingestion
In partnership with industry volumetric capture platform Evercoast, the research pipeline processes multi-camera synchronized RGB-D video arrays, reconstructing continuous photorealistic 3D point clouds and dynamic mesh sequences.
Real-Time Playback & 4D Gaussian Splatting
Traditional volumetric video produces immense per-frame mesh data, straining GPU memory bandwidth. This project investigates:
- Compute Shader Mesh Decoding: Custom GPU compute pipelines that decompress temporal delta geometry directly in VRAM.
- 4D Gaussian Splatting: Exploring neural radiance fields and splatting representations for volumetric actors with view-dependent specular highlights and natural hair transparency.
- Pedagogical Tooling: Developing user-friendly Unity and Unreal Engine editor importer plugins so undergraduate animation and theatre students can integrate live holographic acting into interactive productions.