Looking for a spot in the 8k macro pictures in the video by SweViver where all 3 RGB Pixles are “on”’ at the same time and in focus was very difficult. I found this small spot in the DSC cockpit dial.
Let’s suppose that that is a good representation of the panel. It would confirms asymmetrical pixels (Hight > Width) in a diamond pattern like this.
I tried to approximate the actual resolution of the Pimax 8k using the geometry of that patern as follows
Horizontal Pixels = 3840 * Sqrt(2) = 5430 pixels
Vertical Pixels = 2160 / Sqrt(2) = 1527 pixels.
That is that the Pimax 8k panels maintain a pixel density of a 4k panel or 8.3MegaPixels total.
To clearly show the difference between this “diamond” in the 8k and the squared matrix in the 5k I prepared the following pictures with a scale of 1:100
He’s desperately trying to make it look like the 8k panel are actually full RGB, while it’s pretty clear from the other thread that it isn’t. We can’t say 100% though, the photo’s just aren’t clear enough, but this doesn’t look like full RGB at all
It’s a good point of view, but I think it has been shown that the panel does have
Vertical Pixels = 2160 / Sqrt (2) = 1527 pixels.
But it has also been shown that it has a 1.5 relation with the horizontal pixels of the 5k + bone that would have 3840 horizontal pixels.
3840 * 1527 = 5.863.680 in lugra of 8294400. Also it is necessary to add that the panel is used approximately 80% = 4690944 is 40% more than a 1440p panel.
That is why it is said that the panel is not Full RGB or is not Full 4k.
What you have not taken into account is that next to the red also appears the blue and red color as a whole seems to be as big as the green led. suggesting that the panel is 1 pixel green / 0.5 pixel red +0.5 pixel blue. What produces a huge SDE effect since there are gaps between the pixels without turning on because it is not Full 4k RGB
I agree with this, basically. I’ve done a load of pixel counting this morning and the 8k appears to be using 1.5 times as many horizontal pixels (as expected) but around the same number of vertical pixels (if you assume each green dot has an associated blue and red dot). Which means when I count the total number of pixels by area for a given character or feature I am ending up at 25-30% more pixels (or sub pixels if you like).
The problem with this claim that the panel is G/RB is that this then wouldn’t add up - the height of a given character should contain less green pixels because couting a green and then a RB as individual pixels means the height of the character is doubled, when it should only be 1.5 times the 5K+ (or should be less if you count the G/RB as a single pixel because the height of the panel would be half 2160 = 1080, so it shouldn’t be “the same” number of subpixels being used.
The other thing is that people keep calculating horizontal or vertical pixel counts and then multiplying by 0.8 because of the “80% utilisation” but it totally doesn’t work like that - VR HMD’s lose more pixels in the corners, so the horizontal and vertical utilisation should be much higher than 80% but then you lose more on the diagonal making the average 80% (as you’ve done).
So yeah, to my eye the 8K is using quite a bit more actual pixels (including all 3 types of subpixel) than the 5K+, and this claim that the panel is only half is completely wrong.
We still don’t know what the utilisation is because Pimax are seemingly refusing to give us basic information.
And we don’t know 100% for sure that the distance in the VR world was EXACTLY the same for both of this pictures (unless SweViver has 100% confirmed this somewhere?). If he was slightly further for the 8K this would of course make it use less pixels. How well controlled was the testing?
So to go counting pixels and use formulae which incorporate utilisation etc is totally futile.
All we can say is that the layout of the pixels in the 8K is different to the 5K+ layout and it accounts in some way for fine details to appear a little blurry and ill defined when really zoomed in.
I have heard that if you split a pixel in half it causes a chemical chain reaction of confusion followed by an aftershock of rage and eventually, when the noise settles people will analyze the fallout 4
@LoneTech, what do you think of all this? Could it be that pattern of subpixels or checkerboard?
The Pimax 4K I’m pretty sure they have the same pixel pattern as the Pimax 8K, because I could compare them with the same spherical lenses of the Pimax 4K, in situ.
Without a doubt, a combination of better lenses and better 4K/UHD screens would allow to have an exceptional Pimax 8K+ or 8KX.
@SweViver, could you take pictures where the subpixels are seen more clearly? It might even be a good idea to disassemble the Pimax 8K and take pictures of the subpixels on the screen using a microscope. This would help a lot to understand the subpixel pattern of the Pimax 8K screens.
Yes. I’ve not used any “utilisation” calculation, I am assuming sweviver had his headset positioned the same for both pictures (his flight sim pics seem to suggest it at least) - in that situation it seems to me the 8K is using approx 30% more (sub)pixels for the same feature, which to my eye removes the possibility that this is a “half” resolution panel.
There’s no need to look for one point with all pixels on. Look for places you see points of each colour in clear patterns; they’re all the same checkerboard pattern, red, green and blue. The key here is that their relative placement is distorted by lens chromatic aberration, which is why we see a lot of different overlaps between them. While they might be arranged as antiflash illustrated, they’re not that wide. The fill ratio isn’t as high as the illustration. This is effectively checkerboard for each colour, and while they’re physically displaced one way or another, that gets lost when viewed through the lens. The fact that we see the precise same distribution in all colours means the only difference between them is an offset; it would be typical for that in turn to be along the columns as illustrated.
You can view this pattern (and process it) multiple ways. One way is a 3840x1080 or 1920x2160 panel with RGB subpixels, where the subpixel order is altered every other column. Another is that it’s 45 degrees rotated, and a perfectly typical RGB grid, but the borders are at 45 degree angles to the grid. This isn’t how it’s fed data, though.
The link I fail to follow in antiflash’s reasoning is that the 8K’s panel would have to maintain full subpixel count, while outperforming the claimed resolution. Nobody said it was a 5.4K panel; it’s advertised as 4K. Compare e.g. 3840x1440 panels (which I’ve seen suggestions some Pimax 4Ks use); those are certainly not advertised as 2560! And if we count pixels of any one colour, counting the gaps as one also, our comparison shots have matched expected counts. Please find me a place in the comparison shots where the 8K shows twice the resolution of the 5K in one axis; that’s what would support this hypothesis. What I’ve seen shows about 1.5x in both axis, then half the dots missing.
The silly part of it all is that there are dot distributions which could yield higher average resolution for the same number of dots; those where the nearest dot center pattern yield regular hexagons, not squares. That’s the true delta pattern with a sqrt(2) pixel aspect ratio, I think (the result is that you can trace lines equally in 3 directions). Such a thing would be more subtle to compare, because the lens magnification might distort the angles. By having Pimax stay mum about real specifications, the display manufacturer simply leads us to conclude they are trying to fool us.
