Hacker News Reader

Comment by intrasight | Hacker News Reader

Advertisers say that 90% of their ad spend has no results the problem is they don't know which spend that is. Graphics cards have the same problem. 90% of the pixels aren't being looked at. The problem is the card doesn't know which ones they are. This is the problem to solve.

View on HN

My estimate is 70% is peripheral and the eye can move across the screen in 105ms or so. (60° FOV 18° excluding peripheral and 400°/sec speed)

That makes things really hard when your absolute minimum input latency (that nobody gets in practice) is already 24ms at 60Hz.

Maybe once we start talking about higher frequencies it can make sense. VR is trying hard for higher refresh rates since we are more sensitive to input latency for things we are experiencing and is hoping that tech helps get there.

I do think it'll get there. Really it has to - especially for battery powered AR/VR

Certainly headsets could use the tech but the difficulties in using it for monitors are enormous.

Also the performance benefits aren't as enormous as you would think. For instance unless you are pixel bound (which to be fair most things are in 4k games) it actually hurts your performance by slowing down the CPU.

My fear is it will end up being yet another tool that gets abused to where everyone associates it with a lack of quality.

After all looking to your left and seeing and old image would be very much against the goal of increasing performance.

I’ve been reading about eye tracking and selective rendering for at least twenty years now.