The compute power it would take to do that in realtime at the framerates required for VR to be comfortable for two separate perspectives would be absolutely beyond insane. But at the rate hardware improves and the breakneck speed these AI models are developing maybe it’s not as far off as I think.