I had a conversation with Kurozumi once where he expressed the same opinion. I wonder how he’s doing these days. Moved on, I suppose.
I guess all 3 of us feel the same then. I looked at those shots you posted up above and had no idea the rgb structure of those qd-oleds looked like that, I could never use that for crt shaders. I hope one of these companies takes it back to the basics and at the least include some kind of option to switch over to standard rgb triad structure in some future release.
Ive got even more bad news for you: even if they had an RGB layout the colours of the sub pixels would be wrong as theyd now be matching rec. 2020. This means the straiggt line from the whitepoint to green does not go through the green of rec 601 and so you have to turn on other subpixels to get the right colour (red in this case). This happens in the blue and red too. So basically we’re stuffed for 100% RGB masks on any modern display going forwards.
This is very horrible news. So now we’ll all be hoarding RGB displays the way people hoard CRTs, I guess.
I take it you have a similar problem with rec 709 and DCI-P3? Curious if working with sRGB or Adobe RGB is easier for achieving CRT colour accuracy.
Also curious in your testing how much the RGB structure of the s95b is messing with the final image compared to your PVM if you had the time, Cyber mentioned how OLED owner’s opinions have evolved over time along with techniques to mitigate issues.
Here’s a little roundup I did, of some of the comments of OLED TV users after things started working at a more acceptable level.
I really enjoy RetroGaming on my OLED TV. I get it that everything may not be perfect at the subpixel level but using a 55" screen, I don’t sit 1 foot away from the screen and I can’t see the subpixels at all. Furthermore the only compromise I really have to deal with when it comes to subpixels at least on my display is the phosphor order of Red-Blue-Green instead of Red-Green-Blue and that’s more psychological or academic as they say than anything else at normal, real world viewing distances.
So in summary for those who are into zoomed in pixel pr0n these may not satiate any appetite but for those who actually still enjoy playing games and having fun they’ll probably be in for a treat.
Back in the day, we played on whatever we had available, wherever we were. I knew about composite connections so I was the one to insist on that over RF. Actually, it was what I had to use at home since I gamed on a Commodore 1702 on my NES back in 1991!
So for me, I see this whole thing as a glass half full rather than half empty thing. At first we were getting really crazy results using WOLED and people were screaming Oh, the White LED, after accidentally discovering that Guest’s BGR fit and worked perfectly with WOLED displays, where’s the white OLED is the devil talk now?
So who knows what else might figure out over time once we keep our minds open enough? Do we really appreciate and understand how microscopic these OLED subpixel elements are to begin with?
@thephonypope, if you’re interested, you can read up on some of the ground work that went into getting some major improvements for OLED TV users over the past year.
You’ll see how far we came from when RGB Mask Layouts were being used and getting horrible alignment issues (but still looking good from normal viewing distances) and the thinking then was that the most suitable Masks for OLED TVs were BW masks.
I’m not sure what you’re talking about, as far as I know you still get weird triads. And if we’re going to use viewing distance to solve everything, hell, let’s just not use shaders and stand 20 feet back, raw pixels look fine at that distance. It’s all about your particular situation/setup and what you’re sensitive to.
I can absolutely see these differences when sitting 2-3 ft from my desktop monitor, which is a common way to play games. For couch mode gaming? Probably not as big of a deal, but then you could probably just do without the mask, too.
This is how triads look on my OLED TV. If you think that’s weird then that’s okay.
A bit extreme, I hope you didn’t miss my point.
My point exactly. They say beauty is in the eye of the beholder.
I wouldn’t say so because even if you can’t see the individual elements which make up the picture from certain distances, the final image you see only looks the way it looks because it’s made up of those elements.
Anyway, my main point was that over time things have improved considerably for OLED users and I’m thankful for that and was sharing it with @thephonypope based on a statement he made.
At times I’m a pixel sniffer but at other times I’m a couch gamer. I too can see and notice things, many things. I’m also an optimist and I like being happy but that’s just me and that has nothing to do with TVs, Shaders and Emulation. This is not a jab at anyone else, just stating the obvious that we’re all different and the way we handle and perceive things may be reflected in our opinions towards these things.
Yeah, the spacing is weird and you can see lit subpixels where there shouldn’t be any (this is evident toward the bottom of the screen), much like what too much bloom does to the image.
By that some token, if the subpixels aren’t right…
Anyway, no need to drag this out further. If you’re happy, good for you. Some us just want our CRTs back but they’re all dying and modern technology isn’t good enough and it makes us sad.
Maybe this shouldn’t be when using Sony Megatron Colour Video Monitor but that screenshot example wasn’t taken using Sony Megatron Colour Video Monitor and the extra subpixels you see could be due to the use of shader generated bloom as well as other effects and blending.
Not trying to drag on but just stating this for the record so no-one would think that they’re necessarily a fault of the TV.
The photo does show that the TV is capable of producing acceptable (at least to me) BGR triads. How good, bad or indifferent they look is debateable.
Objectively, I would say that the spacing between the Red and Blue phosphors is a hair wider than the spacing between the blue and the green but that’s about it really. I use hair figuratively because it might actually be thinner than a hair.
This is what a slotmask actually looks like, though. In your photo you’ve got huge black spaces between the individual subpixels where the inactive OLED subpixels are. It’s a far cry from the best photos in this thread, which are taken from a standard RGB LCD. The LCD photos could fool an expert.
I don’t think any of the photos in this thread are that zoomed in, or so in focus. It’s pretty easy to take a picture where you can’t make out the individual dots and they appear as a stripe instead.
Also, I don’t think that phenomenon of how the subpixels are made up is exclusively an OLED thing. I’ve seen several different looking examples of subpixels with various shapes sizes and patterns all with black space between the coloured shapes.
Final thing I’d like to add is that to the naked eye even at extremely close distances from the screen you would not see the OLED subpixels appearing like in that zoomed in picture. Also, to match the dimensions of many real CRT masks, you would definitely have to use more than one pixel per phosphor colour so even if you’re using an LCD display there would be at least 2 out of 3 subpixels turned off which would create black space in addition to the black spaces between subpixel elements which would vary in shape, size, quantity as well as orientation depending on the panel design.
So what you see as dots in the picture might actually look closer to this than you are realizing when you put your eyes right up to the screen.
Many things look completely different (to the naked eye) compared to how we see them when viewed under a microscope and if in reality those fine dotted elements didn’t optically blend together in order to form an image, then why do OLED TVs have the best image quality available on the market today? This brings me back to Apple’s Retina display marketing.
You say the LCD photos could fool an expert but It might all depend on how that photo is taken. I’m sure that I can take photos (and I have in the past) of an LCD that show up the sub elements in much the same way as my OLED photos which most certainly should not fool any expert.
So one would take a photo differently depending on the context for which it is being used.
The smallest element which makes up a CRTs image is the phosphor. Modern LCD and OLED displays contain much smaller elements. You can take many of those smaller elements and combine them in order to recreate or simulate a larger element.
Take a look at these zoomed in photos from a random LCD screen. You can see many different things depending on what you’re looking for.
These are some more shots from the OLED which aren’t as zoomed in. The more you zoom out, the more the individual elements blend together. They also blend as a result of movement and imperfect focus in the photo.
I’ll see if I can take a couple photos of some gameplay from close up to the screen. I have some available from when I took those photos but those were taken to illustrate an issue with the shader. The pictures that I’ve shared were to illustrate that the issue was fixed.
The last three shots are certainly better, but there’s still that pesky spacing between the red and blue phosphors which will be hard to eliminate and I wonder what it will look like once displaying different colors. For example, the whites look like they’re just activating white subpixels, instead of being a triad that’s 100% RGB, which is pretty much what you’d expect with a WRGB display.
This is probably you seeing what you want to see. Those non-RGB white letters and the horizontal line are from the Shader Parameters Menu where there’s no RGB mask. Don’t you recognize the text?
However, those 100% Red, 100% Blue, 100% Green triads you see are all white areas in the game.
This is an uncropped version taken from the same place in the same game.
Photo of UN Squadron SNES - Section of white box surrounding pilot. Taken from LG 55OLEDE6P.
No active white subpixels visible in the in-game (CRT Shader) content.
Shader Parameters Menu text and lines appear 100% white, possibly due to the white subpixel being activated there, where no CRT Shader Mask is being applied.
Rec 709 doesn’t have the issue as the primaries are the same colour and the same is true of sRGB. However I think a P3 monitor or any expanded gamut would do the same. Unless they fake it i.e the vice versa of what I said - turning on different sub pixels feom rec.709 primaries to get P3 primaries and adding extra brightness.
It would be interesting if an LG WOLED was actually a little better at creating standard RGB phosphors than a Samsung QD OLED, even with the white subpixel
I don’t understand. When you say standard are you talking abour RGB instead of BGR or are you talking about Aperture Grill instead of Slot Mask - which is what you’re seeing in the example?
BGR Aperture Grill works just as well as BGR Slot Mask.
Based on how those TVs work there are never 4 subpixels on at the same time so if you’re using an application that has the Red, Blue and Green Subpixels active then there shouldn’t be a need for the white RGB to be turned on and one of the other colours turned off.
Even if we’re using a mask pattern like RRBBGGX, which uses 2 pixels for one phosphor colour and we want to produce white, then the Red subpixel in the first and second pixel would be active with the blue, green and white turned off, the blue would be active and the others turned off for the 3rd and 4th pixel, the green active and the others off for the 5th and 6th pixel and for the seventh pixel all will be inactive.
This is all that is needed to produce white using a CRT Shader Mask. Why would there be any need for the white subpixel to come into the picture when you’re trying to produce red, blue, green and black?
As we can see from images I shared when the TV wants to produce white in the traditional way it can definitely use that white subpixel to have a pure, bright white.
Fortunately for us, CRT Shader Masks (pure mask simulations i.e.) don’t use white to produce white.
I’m just speaking from my personal observations and experiences by the way, not trying to provide any expert scientific advice.
Take a look at this image here to see what’s really going on with these WRGB subpixels.
I’m seeing blue, green and red adjacent to one another followed by the white subpixel which can be anywhere from white to grey to complete black. This represents an entire pixel. it might look look it’s wide and spread across a large area but do we really appreciate how tiny this one pixel is and how microscopic the distance across is from the leftmost subpixel to the rightmost?
Also take into consideration that as each subpixel is activated it will emit a glow. I have a feeling that unless you have some specialised equipment it would be near impossible to tell if the glow is coming from the left, right or center of the pixel.
That image was taken from this article below. It’s the only one I’ve found where they’ve composited it in a way that shows all of the OLED subpixels active at the same time.
This is another lovely photo showing WRGB OLED TV pixel structure taken from a website that you can subscribe to for the latest news and information regarding OLED and other emerging display technologies.
This is from a much older early Full-HD TV so the layout seems a bit different from what we have today.
That’s a much better photo, I’m still a bit disappointed in the black gap between the red and blue, but overall it’s not as bad as the first photos had led me to believe. The only other problem I’m seeing is that the subpixels are tuned to different RGB values, which @MajorPainTheCactus already mentioned.
Well if you’re referring to the gap between the red and blue columns of subpixels in the CRT Mask emulation, if there’s a discrepancy between the size of that gap and the size of the gap between the blue and the green columns of subpixels it’s very minute even with such high levels of magnification. At actual screen size and “normal” viewing distances for a screen of such a size, the disparity is even less apparent. In the more focused photos it’s harder to tell that there’s a difference in the size of the gap.
If you were refering to the OLED subpixel layout photo from RTINGS.com, then I look at it as just a BGR layout instead of an RGB layout with the white subpixel taking the place of the “X” or the gap that’s common between pixels even in LCD flat panel display pixel arrays. Do remember that the large black spaces in the RTINGS photo are turned off subpixels.
Well at least we’ve learned what might have been contributing to those strange stray subpixels being turned on in some of the pictures we’ve seen in the past. Maybe some fresh or out of the box thought on how best to mitigate this might be in order for the rest of 2023 and beyond.
It might involve some hacking or some culling of colors which might be triggering the extra subpixels. Maybe colour accuracy might suffer slightly but you might still be able to produce enough colours using the display’s subpixel primaries that it really doesn’t matter that much.
In any case this is also way above my pay grade. When it comes to certain things I sometimes come up with imaginary theories on how things might be able to be overcome.
The subpixels only matter as long as they’re visible to the eye, which is a function of viewing distance, pixel density and visual acuity (within reason; videophilia/audiophilia “golden” body parts notwithstanding). Beyond that threshold, we just draw the mask in whole pixels and move on. Based on Trogglemonkey’s calculations, we should be able to do that at around 6K res for non-weird CRTs (i.e., very high pitch).
Based on Apple’s “retina” display designations, it looks like a PPI between 200 and 400 at a distance of ~50 cm / 20 in is what we’re looking for. According to this chart, that should be doable with a ~40" screen at 8K or a ~20" screen at 4K. (obviously, 4K < 6K, so we’re not quite to Trogglemonkey’s threashold there)
There’s still the question of how well-rendered the mask needs to be (e.g., that high-res image of the CRT subpixels; do we need to see the rounded corners?), so there will always be room to turn one’s nose up And that’s without getting into the motion / refresh rate stuff.
But, if we overlay Cyber’s OLED subpixel shot on top of the high-res CRT phosphor shot, it looks pretty good to me, as far as spacing and layout (note: I only rotated his image to make it level and scaled it to match in size; I didn’t stretch anything to fit):
RGB vs RBG is obviously going to make some pixel-pr0n look off, but as with MajorPainTheCactus’ subpixel primary issue, how much that matters depends on what you’re trying to do (i.e., in his case, full-image color accuracy vs subpixel behavior accuracy).
@MajorPainTheCactus would you like for me to spin this out into another thread (or append it to one of the existing general subpixel-focused threads), since it’s fairly off-topic? Or do you care?