[psandersen]

Given a pixel isn't a perfect colour dot, it has subpixel patterns and there is space between pixels; there is an advantage to having substantially higher pixel count than we can perceive.

What we really need are better up-scalers and a smarter dynamic trade-off between spatial/temporal display resolution. E.g. upscale from half res using something like Edge-Directed Interpolation ( https://people.xiph.org/~ds/edi/info.html ), render unchanged elements at native res, reduce rendering resolution when needed to maintain always lock with screen refresh rate.

With cheap hardware neural network accelerators, we could make use of very high spatial and temporal resolution displays without needing to actually render at 8k/240hz etc.