Ugh, I hope this doesn’t become popular until the technology allows for extremely fast rates like 1000Hz. I am sensitive to flicker, and I can spot a DLP projector from a mile away when it’s a mostly dark background with a few bright objects, as the famous rainbow effect.
That's exactly why it has not caught on. I'm not sure why so far, nobody has tried that in a device that's primarily monochrome with colors mostly just for minor things like accents, for example, an ebook reader. (The tablet tried to be a normal tablet with "eink mode".)
How would that make any difference? It's not motion
in the image that causes the rainbow effect, it's eye movements of the viewer.
I mean, you could use very fast backlight pulsing to change the backlight color of an e-reader, but what use is that, or rather, what about that wouldn't be achievable with regular LED backlighting, just with RGB LEDs or various shades of white, if it's just adjusting the color temperature.
You could say a 4K lcd actually has 12K horizontal pixels whereas a monochrome one would only have 4K. Wouldn’t the colour one give better resolution if it used sub pixel rendering?
In theory, yes. But it comes with its own tradeoffs, like the fact that white areas have three times as many gaps in them as a monochrome LCD would.
Subpixel rendering is only useful for bitonal objects like fonts (and most implementations only support fonts).
There's no reason for it to apply only to bitonal things — if anything, it only really works well for things that have variations in brightness, which is why we invariably apply subpixel rendering to antialiased font rendering, not bitonal rendering. Without the antialiasing, the color fringing is too distracting.
Most implementations of subpixel rendering are very limited in where they can apply, insofar as I don't know of any that are applied to anything but text, and many of those only apply it to text on white or black backgrounds. But it works beautifully on color backgrounds, too, and there's no reason it couldn't be applied to complex backgrounds*. I suspect that if high-resolution (as in, pixel density) displays hadn't become available, we might have invested the effort to make our graphics drawing engines perform subpixel rendering on everything. But with the high-res displays with pixels so small as to be invisible becoming commonplace (and many of them using complex nonuniform subpixel arrangements), it became easier to simply brute force it in hardware.
*and this can be tested by taking subpixel-rendered text on a white background, and then photoshopping that onto an arbitrary background image. The result is actually flawless. (At least, in my experiments of essentially using it as subpixel "ink" onto a light background. I haven't tried anything where the text is lighter than the background.)