Demostrating Pimax 8k has real RGB subpixels with an 4k panel resolution. And explaining the source of its image problems


#182

[quote=“neal_white_iii, post:177, topic:8629”]…The RGB subpixels do not form a square, so (to me) it still looks like an individual pixel is either green or red/blue…
[/quote]

I am more inclined to think of it like the other thing somebody suggested, that we do indeed have full RGB pixels, in a vertical layout, but only half the vertical resolution.

Now, one could argue that if one had to sacrifice resolution in any direction, the vertical would be it, since it is the horizontal one you need for stereopsis.

Horizontal resolution being true still won’t help the 8k, with the scaler spoiling things, so that should still take an 8kX, but you can’t help but think that if all this rings true, then: A) Is vertical resolution rendered and transmitted unnecessarily? …and B) Could that wasted bandwidth have gone into filling up detail on the horizontal instead, with a better suited scaling mode?


#183

The answer to both of those appears to be yes. Furthermore, we don’t know how much diffusion the panel does since it’s claimed to take an actual 4kx2k image, from which it must necessarily lose half the data. Panel makers (e.g. pentile marketing) tend to add some sort of redistribution layer there just so they can argue they’re not wasting that video signal bandwidth; the 8 bits they threw away theoretically contribute to half a bit of image… in all neighbouring pixels, so none of them display what was sent (unless the image is flat).


#184

Yes, given what you said, I suppose the vertical staggering means every other horizontal pixel is likely indeed from the next line down, so that vertical detail is preserved in that way, in a checkerboard pattern - ye olde square root of two, and that a degree of influence from the values of the physically omitted pixels would likely be filtered into the surrounding “true” ones… Seems like so much could have been optimised…


#185

Not only scaler but also sub-pixel matrix. One problem might be caused by the fact, that scaler does not know that it is not scaling to full RGB, and because of it produced colours get smeared to surrounding pixels by the panel (blurring image). Knowing this in advance we can perhaps (just theory not sure if possible) prepare (pre-process) source image in a way that result of scaler will not be so blurry anymore. Like some people suggested applying sharpening filter before outputting to the scaler. Maybe instead of generic sharpening filter something aware of exact subpixel-matrix & scaler algorithm would work even better. On the other hand, even if possible, it would probably make things worse for movies, so it should be optional (turn on or off) feature if implemented.


#186

Great shot! Looks to me the subpixel structure is something like this:


#187

Exactly my thought, the green pixels seem about 50% larger than the red and blue. In which case it would be 2/9th less info than RGB (1/9th less red, 1/9th less blue). But it’s hard to tell, we’d need even better photo’s.


#188

Oh, and also, still nobody opened up his 8k, to measure panel size? I’m planning to open up my 5k+ as soon as I have it (end of next week), to measure panel size.


#189

People who care about this stuff are exactly the ones who didn’t buy a 8K. :see_no_evil:


#190

I’ve tweaked your image to more closely match the photo. I think this is important, because it’s more representative of the SDE.


#191

Hmm interesting, I think this might actually be it. I felt that maybe red just bled more, but it only seems to ‘bleed’ to the right, making me think you’re right.

Also this gaping hole to the right of green probably explains the SDE. But why would they make it like this? It doesn’t make sense or does it?


#192

I’m sure sooner or later someone will open the HMD. But yeah it seems most ‘techies’ went for the 5k+.


#193

Thanks, yeah that looks pretty much spot on. Very strange layout.


#194

Yes, indeed it is. And you can see why the 5K+ has such a similar SDE. A full RGB stripe does a better job of filling in the edges.


#195

I’m guessing that there is some cost benefit (e.g greatly increased yields) to this design. I can’t think of an application that would benefit from the panels being designed this way (which would be the other explanation).


#196

Take a look at this image comparisson ; left Pimax 4K (Xperia z5p) and right the pic from @some1, the green pixels looks the same size in both (maybe a little bit more bigger in 4K), only IMO; the screen layout is different; zigzag in 4K and diamond 8K, dots and squares… :sunglasses:


#197

what happens is that the 4K do not have the same separation between the pixels in both dimensions (vertical and lateral) that can be seen as an error of perception at the beginning of using them, over time you can see that they are practically the same only 4K have a smaller measure in one dimension but higher in the other.

personally I like more the 8K, not having to see that zigzag pattern…


#198

A user explained usually find the diagonal pixel layout in panels in some projectors


#199

Good picture, @some1! The image

that @neal_white_iii has made from this picture explains to me why I saw the 8K SDE as I saw it, since it seemed to me to see only one corner, which would coincide with the space to the right and below the green subpixel.


#200

I’d love to know if this will ever be the case if Pimax could weigh in. Just got my choice email so have to decide soon.


#201

@Pimax-Support @PimaxVR @xunshu @matthew.xu
Why don’t you release some technical details on the screens and components of the 8k and 5k? Someone is going to open his eventually to let us know, it would be better to have that released officially :slight_smile: