Ok, something I’ve thought about is the Metaverse version of video calls couuuuld require less bandwidth than video calls. Because you don’t have to send every frame — instead it’s the one time cost of a model, and then animations data + voice streaming.
And, maybe you can interpolate better since the domain space is human activity instead of an arbitrary video.
And, maybe you can interpolate better since the domain space is human activity instead of an arbitrary video.