With increasing demand for realistic volumetric videos for AR experiences (along with the capabilities and services they can capture), researchers at Google have discovered how to improve the format.
The team has a "The Relightables" system consisting of a spherical cage with 330 programmable LED lights and about 100 volumetric video capture cameras.
According to the team, three changes have been made with the Relightables system. First, the system's cameras operate as depth sensors and capture 12.4 megapixel depth maps of the subject. Second, the team created a geometry and machine learning reconstruction pipeline to synthesize video from the acquired data.