Just saw this: https://www.cinemersivelabs.com/
Volumetric-type scenes generated from 2D stills and video. Could be amazing for Passthrough backgrounds, with real world imagery!
A range of positions within a single environment could be captured to adequately match standing/seated/laying. And a slew of lewd and non-lewd background scenes could be offered. Or the ability to create your own.
Would processing power would be reduced compared to a full 360 degree 3D scene?