4 ms·

The major (unpopular) reasons for VR plateau: - Fresnel lenses! - No haptic feedback? - Closed hardware/software. Unfortunately in the future electricity pr

by bullen 6y ago

The major (unpopular) reasons for VR plateau: - Fresnel lenses! - No haptic feedback? - Closed hardware/software. Unfortunately in the future electricity prices will mostly limit VR to non-realistic graphics, f.ex. Cyberpunk 2077 can't render (full blast) at 60 FPS on a 350W 3090, so to get 90FPS on both eyes you would need (if it was possible, it's not) more than 3x 3090 at a whopping 1000W! And this is at peak everything: memory speed and litography; with cheap- energy, hardware, (food, rent too) etc. The final medium is not VR, it is regular 3D MMO action potentially with server-side neural-network brained mobs/npcs! How is that progressing Mr Carmack?

kraftman 6y ago
foveated rendering should solve this if it can be done well, as you're only rendering high quality where the eye is looking.
- numpad0 6y ago
  Last time I did a napkin math, cone size required for foveated rendering with current latency and framerate to match slew rate of human eyes basically equals to the whole FOV of typical headset that it’s pointless to implement.
entropicdrifter 6y ago
Considering the ridiculous pace of progress for GPU performance we've seen in the last few years, coupled with NN-based upscaling of frames, foveated rendering with eye tracking, and the fact that you've cherry-picked a game whose performance is wildly unoptimized with graphics that are on the absolute cutting edge of what PCs can handle, I imagine there are still many possible scenarios that could play out over the next decade or so where VR with reasonably realistic graphics by today's standards is not only possible to do on a normal PC power budget, but even with a mobile/tablet chipset, as in the Oculus Quest or some similar standalone headset. Things would get really interesting if Apple decided to enter the standalone VR/AR arena with their cutting edge ARM chipsets

by bullen 6y ago

kraftman 6y ago

foveated rendering should solve this if it can be done well, as you're only rendering high quality where the eye is looking.

numpad0 6y ago
Last time I did a napkin math, cone size required for foveated rendering with current latency and framerate to match slew rate of human eyes basically equals to the whole FOV of typical headset that it’s pointless to implement.

entropicdrifter 6y ago

Considering the ridiculous pace of progress for GPU performance we've seen in the last few years, coupled with NN-based upscaling of frames, foveated rendering with eye tracking, and the fact that you've cherry-picked a game whose performance is wildly unoptimized with graphics that are on the absolute cutting edge of what PCs can handle, I imagine there are still many possible scenarios that could play out over the next decade or so where VR with reasonably realistic graphics by today's standards is not only possible to do on a normal PC power budget, but even with a mobile/tablet chipset, as in the Oculus Quest or some similar standalone headset. Things would get really interesting if Apple decided to enter the standalone VR/AR arena with their cutting edge ARM chipsets