That's why rasterization makes use of mip-mapping, bi/tri/ansiotropic, right.
In texture space you still can do this to reduce cached texture memory requirements in the distance.
Yeah, but i've already included this in my guess about a factor of 8. It's really a guess - not sure how much resolution we need behind the camera.
But as you mention this, shading both mip levels affecting a trilinear fetch would be an interesting option to eliminate shader aliasing, so better image quality. Makes sense maybe for a selection of problematic materials or detected fireflies.
The more you think about it, the more complications arise.
NVs texture shading support with Turing comes surprisingly too. Maybe they already have plans to extend this towards caching. I guess so and i would welcome this kind of 'new fixed function stuff'... exceptionally
I wonder how they handle the issue of specular reflections for VR. Probably the eye position is always the center between our physical eyes and showing the same reflections to both is acceptable.
I'm glad to read your enthusiasm. Gook luck with your work and please don't forget to share your findings!
Thanks! I hope i get there before others or at least within my lifetime