The high res requirement is because the screen should ideally cover your entire field of vision. You'd need to sit very very close to a 50" TV to do that. At normal view distances it covers only a fraction so 1920*1080 is quite good.
Our local IMAX theater has 2K projectors. Even that screen isn't covering the entire FOV and yet I can see pixels quite clearly on it. Moderately aliased realtime graphics would probably have lots of very obvious jaggies and all.
You can also turn your eyes around while wearing a head mounted display, so the entire FOV is even bigger than that of the eye (about 120 degrees or so?). However you only focus on a small set of what you see and that's in the middle of the eye's field. Everything else will never require 100% detail, it could theoretically go down to like 20 or 10% or even less. But because of eye rotation you do need the ppi for the entire screen.
Eye tracking could however tell the graphics system where you're looking on that given screen and only use full detail at the center of that. It could make utilization several times better - you'd still render the entire 4K or so image, but the cost could perhaps be even as small as rendering a regular 2K frame. And even without eye tracking you could probably scale the detail around the edges of the screen anyway.
I guess this is one of the reasons Carmack is so interested and enthusiastic. He's been thinking about a lot of this tech before and now the perfect platform to utilize them is finally in sight.
Our local IMAX theater has 2K projectors. Even that screen isn't covering the entire FOV and yet I can see pixels quite clearly on it. Moderately aliased realtime graphics would probably have lots of very obvious jaggies and all.
You can also turn your eyes around while wearing a head mounted display, so the entire FOV is even bigger than that of the eye (about 120 degrees or so?). However you only focus on a small set of what you see and that's in the middle of the eye's field. Everything else will never require 100% detail, it could theoretically go down to like 20 or 10% or even less. But because of eye rotation you do need the ppi for the entire screen.
Eye tracking could however tell the graphics system where you're looking on that given screen and only use full detail at the center of that. It could make utilization several times better - you'd still render the entire 4K or so image, but the cost could perhaps be even as small as rendering a regular 2K frame. And even without eye tracking you could probably scale the detail around the edges of the screen anyway.
I guess this is one of the reasons Carmack is so interested and enthusiastic. He's been thinking about a lot of this tech before and now the perfect platform to utilize them is finally in sight.