Fascinating. So not only is it feeding it an 8K / 30 (60?) FPS image, it's feeding it numerous incident angle variations and displaying all of them simultaneously?
I think that is where the custom compression algorithm comes in. If you think the fact that human body and face doesn't change much, and the fact that it's a 3d model based, the compression ratio could be very high.
I only know what I saw from the IO stream, but I think it might send a compressed 3D mesh + texture across the network and render the light field locally.
I think what they are transferring is not a video but 3d model and the skin texture applied on the model (all derived from the realtime video / depth recording on the other side). The receiving and then renders it as a 3d model on the screen.
Is existing Looking Glass Factory tech the same though? Not so sure about that. Those displays are typically monitor-sized at the largest and not really aimed at displaying a live feed of a person. This looks to be a more seamless experience on a larger screen.
Sounds like a monster data rate.