Awarded VEIV EngD
This project aims to explore the possibilities of video-based rendering to enable arbitrary camera viewpoints for applications such as sports matches. The research will attempt to improve the capture and reconstruction pipeline, involving wide-baseline calibration and acquisition, raw image processing, geometric reconstruction, texturing, relighting, animation and rendering. Specific attention will be paid to the reconstruction of human action. The goal of the project is realistic depiction in real-time.