In the last few years we have worked very hard to push the state of the art of 3D capture systems. Some of you may have attended our tutorials at CVPR 18 and CVPR 19, where we demonstrated some of our work towards a high quality performance capture pipeline.
We are now very excited to share The Relightables, the first 3D capture system that bridges the gap between image based renderings and volumetric videos. Traditional approaches in volumetric capture lack true photo-realism and struggle to capture high frequency details of the performers. Our method instead relies on a Light Stage combined with novel high quality depth sensors to generate renderings of full body humans with an unprecedented level of photo-realism.