So instead of pixels used as the texture of triangle surfaces different colour quad's are used for the surfacing. they can rise and indent collectively but from a birds eye view they are the same as pixels in dimension maybe not with respect to lighting and all the lines of the quad surface must join up, at the border of the triangle where there is pixel conflict between 2 surfaces or more an optimal pixel decision is made.
Now Take the flat quad surfacing add some inner colouring pixels to each of the quad pixels and efficiently enhance it with very basic curved modelling maths leaving the border pixels.
Now bounce light off in a way that has a few complex light bending calc's(for full body lighting) and a lot of simple refraction calc's (for base lighting data).
Say Hello to PS6 graphics maybe who's 3 main features might be Full body lighting, Curved graphics and AI enhanced physics.
say 2026
Probably about 128GB shared graphics ram embedded stacked on the main chip.
A 1.4nm 12 core CPU part on the die alongside the GPU@200TFLOPs.
8TB main drive with secondary m2 drive for upto 32TB upon release.
1.2TB average size of a big game.
What exactly would the PS7 have to achieve. By the end of this PS6's gaming development life cycle it will already look a lot better than ready player 1.
By 2029
This technology will be in budget cards.
By 2032 Mobile VR/AR will look real with PS6 mobile tech and more people will talk about simulation theory.
Hopefully nanotech battery will be getting closer to the limit with aluminium ion batteries and you get at least a good 15 hours worth of game play.