I'm interested in virtual reality, but according to some sources, less than 1% of computers in use today have the necessary performance to run modern VR games, granted many of them are not intended for gaming, that's still a huge barrier for the widespread adoption of VR headsets into the market. So I was wondering what could be done to reduce the performance requirements of a VR game that hasn't been implemented with such optimizations in mind.
I'm talking about technologies like foveated rendering, which (ab)uses the fact that we only ever perceive a few degrees in the center of our field of vision as sharp while everything else is blurry. This can be combined with eye tracking so that only the areas where the player is looking are rendered in full quality, while we can reduce the render quality and therefore the performance requirements everywhere else. This seems like a great advantage and I honestly wonder why none of the major VR companies have implemented eye tracking into their headsets and foveated rendering into their SDKs into their headsets. The technology exists and I can't imagine that it would add much to the price tag. Paper about foveated rendering
There's also instanced stereo rendering which basically renders things to both eyes at the same time rather than rendering it for each eye individually.
It's also possible to use just one frustum for both eyes rather than two as discussed here. VR and frustum culling
But are there other approaches to the performane problem?