We need displays with 4 color components!!!1

How is it that we can have two displays with similar panel specs (say it's IPS for simplicity) and they show completely different colors. If I calibrate and profile them and then choose color that should be well inside gamut of both displays it still looks very different between these two displays. I can try to manually adjust RGB values on either monitor to try matching colors by eyes but it always proves impossible. It can be simple color like white or gray to make things simpler - still impossible to make colors actually look the same.


Hmmm... could it be our eyes have 4D gamut instead of 3D gamut?

I wrote lengthy post but then it kinda started to sound like I was trying to describe colors to color-blind people... just without actually describing anything... it seems our language is lacking as much as our current color encoding standards - it does seem like these things are directly related. I do not believe however I am seeing anything other people cannot see because from what I did see people do notice differences between displays - it is usually just dismissing it on something else like color calibration. Only when proven it is not calibration as it was performed they are more willing to acknowledge that maybe displays as they are currently made in fact cannot display all possible colors*. Of course the difference between colors here isn't only in displays but its in real world also - and once its noticed its quite obvious... and still I am missing words to describe it...

What do you think?
Unlike the past we we are today technically able to improve color fidelity and imho it would be much more useful change than just adding more pixels which at this point doesn't really add that much to image fidelity. I cannot see things improving in this regards unless we acknowledge we can see more colors than we currently think and start improving our cameras and displays and other aspects of video accordingly.

*) It is obvious even in current 3D gamut system - some colors are just way more saturated than we can fit in our limited color gamut.
I am however talking about different gamut difference. Even if we had perfect coverage of CIE 1931 color space it would still not capture all of the color nuance we can see in real world. For one it would be impossible to make photo of these two displays I was referring to at the beginning of this post and show difference in colors even using cameras and display that support each and every CIE color as this particular difference still falls outside CIE defined color space. We need color hyper-space...

sharp did quattron years ago, no one seem to take to that RGBY setups though.

It was color yellow so totally useless at it is nowhere where our eyes sensitivity is which would be between cyan and green. And we still don't have 4 color components to drive additional color components.

Technically WOLED panels should to some small degree already support 4 color component operation and be able to showcase difference I am describing. White from RGB subpixels vs white from W subpixel depending on filters on RGB subpixels should look different.
From there these panels could be tweaked by tuning color filters. On the other end of OLED spectrum QD-OLED should also have no issues adding specifically tuned subpixel - though for now those panels would not be able to be used in any capacity to showcase difference (edit: though can be used comparing them versus WOLED as QD-OLED should have much less rod stimulation - and really today the only way to see difference is either using different displays and by using different lamp types)

If it was possible to hack WOLED panel it would be possible to try to showcase difference.
The issue is... is it possible?

edit:
Come to think about it it should be possible to make custom proof of concept panel with RGB LEDs and W-LED - then if especially W-LED was one of these that blasted at rods it would be easily possible to showcase difference between different amount of rod stimulation. White from RGB LEDs versus white from W-LED should look different even if they were calibrated according to current CIE 1931 color space specs. All other colors except fully saturated too should be doable by either mixing only RGB or mixing them with W-LED.

Too bad presentation would not work over YT video

Look. Accept that colors are not always the same on different monitors and different media and move on with your life.

Because even if you can get your monitors calibrated to within an inch of their life, nobody in the real world calibrates shit, and it's going to look different on their screens. (Unless you're producing content for IMAX, maybe)

As far as I know, we simply don't interpret the output from rod cells as a colour, which is why everything looks colourless in the dark, and not blue-green. They are also mostly out towards the edge of the eye, whereas colour vision mostly happens in the middle. Therefore I don't think adding a 4th colour would improve things.

There are other things that might explain why two calibrated monitors look different. Maybe they use different wavelengths for RGB, or maybe those colours are more or less monochromatic.

I am not sure how that make sense, if both display are RGB, bot cannot show all colors the same way and they would look the same just without said color (imagine black&white monitor, if 2 side by side look different, one would not say it is because they cannot do all colors, sure but they should be able to do the same black and white image, same for the limited same model RGB monitors, for their colors space).

Isn't not a technical issue, about led, phoshpor coating or other being a bit different... and would still be true regardless of using 1-2-3-4-5 color components ?

Mr Evil said:

As far as I know, we simply don't interpret the output from rod cells as a colour, which is why everything looks colourless in the dark, and not blue-green. They are also mostly out towards the edge of the eye, whereas colour vision mostly happens in the middle. Therefore I don't think adding a 4th colour would improve things.

Click to expand...

This is what materials I read say but I simply don't think this is the case.
The same apparent color - as defined by calibration - is really different, it looks and feels different. The same but different and the more I think about it and analyze the differences the more obvious it is.

Rods being slightly on sides only means we should have rather bad resolution of perception of this color. It is hard to say how good or bad it would be without having display. Looking at different displays the effect on colors is everywhere on the screen and colors don't change presentation looking at them straight or off-angle so I guess that in the brain it would all get mixed in the end and resolution should be good enough to see even small part of the image highlighted by making it have more or less rod stimulation.

Mr Evil said:

There are other things that might explain why two calibrated monitors look different. Maybe they use different wavelengths for RGB, or maybe those colours are more or less monochromatic.

Click to expand...

Using different wavelengths is exactly why there are differences - but I should not be able to differentiate colors on calibrated displays if colors are within apparently supported gamut.
Therefore conclusion that we do not capture whole gamut within our current color space standards - monitors with different colors have completely different non-overlapping gamuts.

toast0 said:

Look. Accept that colors are not always the same on different monitors and different media and move on with your life.

Click to expand...

I will need to do it anyways but there is no point in being silent about it. This is very first time I ever mention it on internets. I did only tell few people some of which didn't know what I am saying until I showed rather extreme example. One person even the next thing did buy wide-gamut display - and actually his was quite good to get the effect I was showing him. Not sure he understood why it works - most people have no idea how monitors with current RGB system work so....

toast0 said:

Because even if you can get your monitors calibrated to within an inch of their life, nobody in the real world calibrates shit, and it's going to look different on their screens. (Unless you're producing content for IMAX, maybe)

Click to expand...

You are right - in fact most of the time things are just set by eye.
Still it is not so much about always having accurate colors but ability to have wider expression of colors.
The difference in colors is in 'vibrancy'. Having color that hardly changes color but changes vibrancy without making things seem oversaturated would be actually amazing for creativity - you could for example highlight something and have it have the same color but 'glow'.

You can be sure that if we had 4 color component displays then most ads would have very vibrant presentation - also mixing in colors with rod stimulation for better effect.
In movies and games you could build atmosphere by even overusing this effect. It is once you know what color differences those are doesn't at all seem as insignificant as it might at first seem.

Then on the other hand you would have ability for simply more accurate natural looking photos/videos with real-life colors. With the current setup approximately life-like if scene you look have comparable amount of this rod color - which is something I think PAL color with EBU phosphors absolutely nailed and why old CRTs had such natural looking image on average whereas lots of later "HD" displays were completely off even if contrast and gamut was there.

In either case, no matter if your display has this or that amount of rod stimulation everything will always look the same on it - it is like putting skin-like color overlay over black&white TV to have people look more or less like people but then what isn't supposed to have this color also have it because there is no way to control amount of this color. This is not what anyone wanting fidelity should want. And I can assure anyone they would be blown by image of such 4 color component display- especially once people started making art for this standard.

LukeTbk said:

I am not sure how that make sense, if both display are RGB, bot cannot show all colors the same way and they would look the same just without said color (imagine black-white monitor, if 2 side by side look different, one would not say it is because they cannot do all colors, sure but they should be able to do the same black and white image, same for the limited same model RGB monitors, for their colors space).

Isn't not a technical issue, about led, phoshpor coating or other being a bit different... and would still be true regardless of using 1-2-3-4-5 color components ?

Click to expand...

White is not the same (it does not look as having the same color!) with rods being stimulated and not being stimulated.
We don't need 5 color component but exactly 4. We have 4 types of photo-sensitive cells each with its own peak sensitivity.

Is this supposed to make any huge difference when games are still using Rec.709 colors in 2024? Even a lot of games that have pretty good HDR support are still stuck on Rec.709 colorspace.

XoR_ said:

White is not the same (it does not look as having the same color!) with rods being stimulated and not being stimulated.

Click to expand...

How would rods not being stimulated if there is any light reaching the eye ?

They do react more to low wave (blue-green) but they do react to almost all the visible light spectrum.

But right now, all the reason why 2 same monitor calibrated look different would still be true for the type of monitor you are suggesting, right ? The same things that create different rgb that make for differents white will happen to those w-led, colr temp and phospor mixes will different from led to led and from monitor to monitor of the same skus.

LukeTbk said:

How would rods not being stimulated if there is any light reaching the eye ?

They do react more to low wave (blue-green) but they do react to almost all the visible light spectrum.

Click to expand...

Perceived colors are about ratios of stimuli.
For RGB-LED display that I have I have very narrow peaks of red, green and blue and each slightly stimulate rods - but then you have more cone stimulation than rods thus see different color than if you had more rod stimulation on e.g. typical W-LED sRGB display.

In this sense there is minimum amount of rods stimulated and any use of zero rods stimulation refers to delta from this minimum. RGB-LED might not be at minimum it exactly but its for no-rod colors my reference.

Sorry for confusion.

LukeTbk said:

But right now, all the reason why 2 same monitor calibrated look different would still be true for the type of monitor you are suggesting, right ? The same things that create different rgb that make for differents white will happen to those w-led, colr temp and phospor mixes will different from led to led and from monitor to monitor of the same skus.

Click to expand...

Trend in displays has always been that cheap products would use more random phopshors, CCFL lamps or W-LEDs which were typically on "too much rod stimulation" end while more expensive displays where manufacturer actually cared about how things look would use better tuned and more natural looking tech or even very vibrant looking tech.

Across all different display technologies it is always the same difference - the color cast that I noticed on cheap LCDs is there also on cheapest CRTs - it is more noticeable having much better displays. Better here is generally less rod stimulation but actually there is optimal look where colors are almost natural looking with tiny bit artificial vibrancy. It also matters what ratios or this stimulation comes from each RGB component - cause this is also a thing but it is let's say higher level but yeah, this is also a thing and important to looks of colors and why its hard to match colors between monitors - cause even if you had the same looks of white it doesn't mean colorful picture would look identical

I am pretty sure however that having control over stimulation of rods would allow fine control over how rod color is derived from normal RGB signal and even without having full 4 color channel content there would be many interesting ways colors could be configured and eventually actually optimal way discovered - of course superseded by 4 color channel content once available - which true power would be perhaps defying expectations even more than allowing for more natural looking colors.

As for variance between monitor skus I would expect only cheap stuff have any between various batches. Actually expensive monitors should be consistent - there is well defined gamut to which for example coefficients used for sRGB emulation in firmware are referring. Cannot nilly willy change LED type in this case.

LCD colors change depending on the viewing angle. On a good LCD it is basically unnoticeable in real world content, but if you put 2 LCDs side by side and are looking for it, it's easy to see even if they were perfectly calibrated to match each other.

Switch to 2 perfectly calibrated OLEDs and you won't be able to because the viewing angles are so good there isn't color shift until you get to extreme angles.

So even if an LCD has extremely accurate colors it only matters if you look straight on (and the image isn't in motion) Just another thing that makes OLED superior to LCD.

XoR_ said:

As for variance between monitor skus I would expect only cheap stuff have any between various batches. Actually expensive monitors should be consistent - there is well defined gamut to which for example coefficients used for sRGB emulation in firmware are referring. Cannot nilly willy change LED type in this case.

Click to expand...

Not I am getting lost, was it not the premise of this thread that the same panel side by side would look too different ?

sharknice said:

LCD colors change depending on the viewing angle. On a good LCD it is basically unnoticeable in real world content, but if you put 2 LCDs side by side and are looking for it, it's easy to see even if they were perfectly calibrated to match each other.

Switch to 2 perfectly calibrated OLEDs and you won't be able to because the viewing angles are so good there isn't color shift until you get to extreme angles.

So even if an LCD has extremely accurate colors it only matters if you look straight on (and the image isn't in motion) Just another thing that makes OLED superior to LCD.

Click to expand...

It is not viewing angles. I can always use two CRTs with different phosphors and see similar differences e.g. between SMPTE-C and EBU phosphors - most of the gamut is overlapping and yet its not possible to make colors really look the same. Colors are obviously in some sense the same when everything is calibrated/profiled and displayed with color remapping but at the same time its not the same color.

I am also pretty sure colors on QD-OLED will look very different to WOLED panels - though I have not yet have pleasure of testing that. I did compare lots of different display tech including such that don't have any viewing angle issues so no - its not viewing angles or something else just not incomplete video standard.

That said there are of course other differences between displays which can also cause some small visual difference but I am pointing at green-cyan elephant in the room that we can all see and don't call it anything because... it does make sense we don't have names for things like these but that doesn't mean we should make subpar video and display standards because of that.

LukeTbk said:

Not I am getting lost, was it not the premise of this thread that the same panel side by side would look too different ?

Click to expand...

Not the same panel but panels with different light spectrum - it necessitate e.g. different backlight (and/or color filters) or differences for self-emitting panels.

When panels from even different skus use the same tech its easily possible to calibrate them to look identical.
If you had wall of PVMs (Trinitron CRTs) with e.g. SMPTE-C phosphors all would look identical and calibrating them to nail the look would be fairly easy. If you had there model designed specifically for PAL it would stick out and you can run signal through image processor able to change anything you want about it and you would not be able to make colors the same even if you limited signal to only be what should be displayable by both types of tubes - and this difference is due to rod stimulation being slightly different between SMPTE-C and EBU phosphors and this what our current video and color space / gamut standards don't define or treat in any way - its literally free floating parameter and each display can be different, at times without rhyme or reason why certain look was chosen - just one common trend is that cheaper displays are badly calibrated in the direction of overstimulating rods while really high end stuff typically has much less of this rod stimulation and if such expensive display support gamut emulation trying to emulate gamut of other monitor exactly will still result in this expensive monitor having much more vibrant and nice/pleasant looking colors. Panels between these two can be very similar in contrast, viewing angles, etc.

I am relying on experience of using displays - best many over many years - to know what I am talking about. I cannot make photo of the effect - though cameras will pick differences between displays, especially cheap cameras, just not the right difference - it cannot be represented by our current tech and this is exactly why we need the change - not necessarily to make photos of monitors but because the same differences are seen in real world - which once you recognized colors is quite easy to notice.

My experience with what colors I did see was pretty consistent - at least where it comes to trends and what colors I refer to saying things like "color cast from rod overstimulation". It is also consistent with colors in real life.

MistaSparkul said:

Is this supposed to make any huge difference when games are still using Rec.709 colors in 2024? Even a lot of games that have pretty good HDR support are still stuck on Rec.709 colorspace.

Click to expand...

If we had this RGBX standard or however it would be called it would probably take many years until we would see it implemented with OS and then games.

First we need to know we see more colors than we think we can see
Even if that topic doesn't magically contribute to future standards at least someone might consider light spectrum of their next display - which would still be good thing for them. That said most gaming monitors these days are pretty good already as necessitated by needing wide-gamut for proper HDR support which by minimum means DCI-P3 an DCI-P3 capability already at least shifts look of image toward much better and more natural looking colors - of course when clamped to actual color standard used which is sRGB/Rec.709 in most cases as you pointed out.

XoR_ said:

Using different wavelengths is exactly why there are differences - but I should not be able to differentiate colors on calibrated displays if colors are within apparently supported gamut.

Click to expand...

The wavelength difference is important because it can make calibration inaccurate. Colorimeters need to be corrected for the wavelengths used by the display. Without the appropriate correction, they might say that two displays are the same when they will actually be perceived differently. More accurate calibration will remove the difference.

It it bacause of observer metamerism, people have different color matching functions. You are right, adding a 4th subpixel would mitigate the issue.

There are lots of colors especially in cyan range which can't be displayed with RGB pixels.

Adding a cyan subpixel would be especially useful in OLED displays where it's been difficult to develop a long-lasting efficient blue emitter because of the high energy of blue photons. If the cyan pixel could be used instead of the blue most of the time, the lifetime requirement of the blue could be lessened. Therefore UDC, a major OLED material manufacturer, has been advocating RGBB pixel architecture.



https://www.oled-info.com/udc-discusses-its-rgbb-display-architecture-and-its-advantages-color

Later said:

It it bacause of observer metamerism, people have different color matching functions.

Click to expand...

That's the same as the colorimeter problem I mentioned before, except with a real observer instead of an instrument. It is likewise correctable, though I have no idea how you'd go about measuring the response of your own eyes. It's still not because of rod cells anyway, in case anyone was thinking that.

Mr Evil said:

The wavelength difference is important because it can make calibration inaccurate. Colorimeters need to be corrected for the wavelengths used by the display.

Click to expand...

That is true. Because colorimeters only approximate the way humans see colors they need to be calibrated to given color spectrum of the display - which in practice makes them inherently inaccurate unless you have spectrometer also and can do such calibration. In this case colorimeter is useful because of greater light sensitivity as spectrometers are pretty mad at calibrating near-black colors - which can be done but in worst case scenario like OLED would need extremely long times to collect enough light to provide good calibration.

Mr Evil said:

Without the appropriate correction, they might say that two displays are the same when they will actually be perceived differently. More accurate calibration will remove the difference.

Click to expand...

You can have best colorimeter and best spectrometer and do everything by the book and still you won't be able to make image look identical between some monitors because you cannot correct light spectrum of the monitor itself and displays don't control all color components that we humans can see.

Like our video standards color calibration was never designed to be perfect but merely good enough. You can only correct what you can correct. That never stopped manufacturers of calibration hardware/software to use ad slogans suggesting colors will be the same across your whole workflow if you buy and use their stuff.

What I always wondered how much people see these differences - definitely even I can see that between typical W-LED and RGB-LED when both are calibrated the colors are pretty much the same but only in a sense - it does look like the same color but not in absolute sense. It is the same only in the sense like if it was print and I viewed it with different light - but one which somehow only affects rod stimulation so even more similar than typical print under different lighting. Still nod identical.

XoR_ said:

You can have best colorimeter and best spectrometer and do everything by the book and still you won't be able to make image look identical between some monitors because you cannot correct light spectrum of the monitor itself and displays don't control all color components that we humans can see.

Click to expand...

If you had such calibration devices, you could calibrate the displays with different spectrums so that they would look the same for an average person, or to any one observer, but never so that they would look the same for a number of observers, because individuals perceive colors differently.

If a normal and wide color gamut display were calibrated so that they would look the same for an average person, I don't know how big a percentage would consider them similar and how big different.

Later said:

It it bacause of observer metamerism, people have different color matching functions. You are right, adding a 4th subpixel would mitigate the issue.

There are lots of colors especially in cyan range which can't be displayed with RGB pixels.

Adding a cyan subpixel would be especially useful in OLED displays where it's been difficult to develop a long-lasting efficient blue emitter because of the high energy of blue photons. If the cyan pixel could be used instead of the blue most of the time, the lifetime requirement of the blue could be lessened. Therefore UDC, a major OLED material manufacturer, has been advocating RGBB pixel architecture.

View attachment 674413

https://www.oled-info.com/udc-discusses-its-rgbb-display-architecture-and-its-advantages-color

Click to expand...

If we're going to go as far as making one subpixel alternate between colors, why not just replace the whole subpixel structure with a single pixel that outputs color based on the YCbCr calculation and take RGB completely out of the equation? I know that is a much more drastic change and more expensive to implement than what is being suggested here, but it seems that would completely eliminate the problem.

Armenius said:

If we're going to go as far as making one subpixel alternate between colors, why not just replace the whole subpixel structure with a single pixel that outputs color based on the YCbCr calculation and take RGB completely out of the equation? I know that is a much more drastic change and more expensive to implement than what is being suggested here, but it seems that would completely eliminate the problem.

Click to expand...

How could you make a pixel without the sub pixels?

Also for the post you replied to, that's outdated information as far as TVs go. Both LG OLEDs and Samsung QD OLEDs use one specific OLED color for every sub pixel and then just put color filters over them. LG uses white oleds with filters, and actually uses 4 already, red, green, blue, and then a plain white. Samsung uses blue OLEDs with filters to have red, green, and blue.

Mr Evil said:

That's the same as the colorimeter problem I mentioned before, except with a real observer instead of an instrument. It is likewise correctable, though I have no idea how you'd go about measuring the response of your own eyes. It's still not because of rod cells anyway, in case anyone was thinking that.

Click to expand...

You could try set up the same psychophysical experiment that led to the original color matching functions, but it would be a pain in the ass, and you'd need equipment to generate arbitrary narrow band sources for your "ground truth"

sharknice said:

How could you make a pixel without the sub pixels?

Click to expand...

Create an emitter that can dynamically radiate different wavelength of light with a photoreactive gas or dye to reflect it. There is research on this happening right now.

Armenius said:

Create an emitter that can dynamically radiate different wavelength of light with a photoreactive gas or dye to reflect it. There is research on this happening right now.

Click to expand...

Sounds cool, have any links?

Imho fixing colors all at once while covering all cases for all people is impossible due to technical difficulties and unlike prototype for scientific research both display and camera tech need to exist at the very least and then be cheap enough to be viable. Proposed solutions with pixels displaying whole visible spectrum or controlling hue don't sound either technically ready or cheap.

Moving from B&W to color video was no brainer so expensive new tech was justified - and it was in some aspect very hacky.
Moving from RGB to RGBX will have this advantage that it won't be that expensive or technically hard in comparison. Technically we could do this decades ago while moving to HD or when introducing HDR. We already use cameras and displays using matrix of subpixels and digital stuff hardly adds any cost. One more subpixel and fancier color filters is all that is needed on the hardware. It is all some cost so full fledged implementation will take years from the moment someone actually makes presentation that will sell that idea to the TV lords.

In fact I am perplexed HDR doesn't have this in its specs. Bright daylight scenes and dark scenes have usually different light characteristics - the effect in current HDR is not that convincing imho - not like I imagine it could be.

EDIT://
Edited post to clarify some aspects.

Also to throw a bone: LG could modify WOLED panels to support this for demos:

1. Make white have more light that hits rods.

2. Improve color filters to filter out rod stimulation as much as possible

3. Allow either mixing RGB or like these panels work currently always try mixing W subpixel. This would be axis of control.

It isn't the same as having cyan or green-cyan subpixels but I think it would make more sense from the point of view of making actual display. I can only imagine separate cyan subpixel would make it somewhat hard to color correct standard 3D gamut and for simplicity we want one parameter that controls if light has rod light or not. With RGBW fully saturated colors (as in those which don't use white subpixel) in this case would have no rod stimulation control possible and control would be the highest for white. I guess that is good compromise because technically adding cyan light would make image look cyan because just like RGB components stimulate rods the light for rods would stimulate cones sensitive to its light - well, it would look exactly as color we choose for it.

What it would allow is handling most used colors. Also for HDR we usually need to represent daylight and daylight is rod stimulation heavy while dimmer artificial lit scenes might need less of this rod stimulation. In this case for HDR daylight highlight having white subpixels such stimulation would work. Nailing specific look would be done by mixing in RGB to create additional white light. Dark scenes would mostly use RGB mixing + some W subpixel to control the exact amount.

---
As for this WOLED and SMPTE-C comparison - SONY GDM-FW900 pretty much uses SMPTE-C and so did many VGA CRTs which is where I can even see these phosphors - and they do not look like EBU phosphors used in PAL TVs. I like PAL much more and especially since I grew up with PAL it looks to me most natural but also just is touted as generally having superior colors... to everything? By some people including me - I do consider CRT TV made with PAL specs to have most life-like colors - at least for most scenes.

It took me seconds to get impression that white on my WOLED monitor is like on these SMPTE-C phosphor using CRTs. Of course WOLED has DCI-P3 and no matter how you slice it RGB display with DCI-P3 should have less rod stimulation overall. WOLED panel does however use RGBW subpixels and it can achieve SMPTE-C looks for SMPTE-C colors while providing wider gamut. It could also without any modification to actual OLED panel support control of rod stimulation between how much this stimulation is when mixing RGB to get white and mixing RGW, GBW and RBW subpixels - which would be current range - which still should be imho enough for proof of concept to showcase some differences. If that would be wide-enough range to convince most people that this tech shows promise I don't know but imho it might just as well be.

What I thing could or would work is making it so display can display Rec.709 like best looking color-wise displays like Pioneer Kuro. I have this beast of a display next to my WOLED monitor and let me tell you: this plasma phosphors are tuned in such a way that gives much more vibrant and 'glowing' colors - all the while colors don't look oversaturated one bit. That is just the effect when ratios between all photosensitive cells are fine tuned to give pleasant stimulation - colors look more vibrant whole not looking more saturated.

So with the potential possible range of stimulation that should be quite big actually it might be possible to have presentation where e.g. half of the picture shows in more fine-tuned colors for this 'glowing' while other half is done like it currently is. Maybe add something in-between like closer to EBU looks and display the same thing in all parts of the screen with different ratios.

Imho it should already be enough to showcase some possibilities of this tech.

---
Also and this is also why I did make this topic: if it should be possible to modify WOLED panel to have some of this control... why do we use W subpixel for all colors?
If these monitors used RGB as much as possible and only add W subpixel when its actually needed what we would get is:

1. Mitigate near-black chrominance overshoot
It happens because there is a sort of kick to OLED subpixels going beyond some luminance level and since W subpixel is so bright its also quite visible. All three RGB subpixels together should be much dimmer so these flashes would be much less visible if we didn't use W subpixel for darker colors. Adding it later would have it flash but no one would then see this flash. To mitigate this issue even more there could be random offset where W subpixel is mixed in so transition is smoother. Lots of tricks can be used here.

2. The whole issue is that dark scenes need less rod stimulation compared to bright daylight scenes - if RGB was used for as much as possible and W subpixel mixed in later we would get both more natural looking dark scenes and more natural looking bright daylight scenes. This is unlike something like I imagine QD-OLED should be based on its spectrum - more vibrant overall but also much less convincing daylight scenes

Note: less natural doesn't mean some people might not prefer looks of these less natural colors - in fact I think I would like QD-OLED much more than WOLED. If however WOLED used its strengths better then who knows. I actually like how natural looking daylight scenes are on this panel and Kuro plasma with its candy-like colors while looking amazing does seem at times too overdone. In reality during the day my rods eyes are totally blasted with light blinding them and saturating them making no colors 'glow' - unless something can block all that light... and its usually objects with very saturated colors and those will on WOLED as it currently is do have much less W-subpixel and look more vibrant but its not like colors need to be always very saturated - some things do even in full daylight glow while still having color toned down...


...anyways - I am not sure if my explanations make any sense at this point.
Color perception in humans is quite bit more complex than what displays use to sell the illusion of seeing what we should see. It is in some sense making us seeing things through cameras+displays color blind. Subtle aspect of tonality of colors is being lost. I do also believe that if we make effort to restore these subtler aspects the images will be much more enjoyable and effort will be appreciated - especially when people having direct control over this aspect of light will be able to see how brush with rod color set to white or black can change presentation of colors while not changing the color itself. I also believe that anyone with such experience will inevitably have much greater awareness of these tonal differences happening in real life.

For now its like we see somethings in real life that might not even be that saturated but do stick out in how vibrat they are. On displays it to some degree happens too so it only makes it kida consistent - and why there is less dissonance. It is however something that can be noticed making photo and viewing it - which color is glowing and which not depends on the display and not which were glowing in reality. Photo/video of Pioneer Kuro plasma isn't representative of how it looks in reality. Watching cheap sRGB monitor on Kuro plasma it has nice vibrant colors - in reality it doesn't. We should have color system, cameras and displays which can capture and display these color nuances.

Got QD-OLED today and expected its colors will be very close to RGB-LED IPS and they in fact are but even closer monitor I had is old two decade old CRT
with Trinitron tube IBM P275.

Contrast is better and this Trinitron having SMPTE-C phosphors had slightly different gamut.
I also have Dell P1110 which is also similar. Then SONY GDM-FW900... not at all. Totally different colors - imho worse despite obviously sharing SMPTE-C phosphors with the other two. Then my FW900 underwent AG removal procedure which changed its colors which were not like the other two but more like Iiyama LM704UT but not quite like that. Colors on the Iiyama are similar to some colors WOLED can display. Going further in relative similarity there sRGB mode on WCG-CCFL Dell U2410 which colors are then similar to Pioneer Kuro. Then I would say from notable displays would be LG 27GP950 and then going further QD-OLED and these CRT's and lastly RGB-LED - though I need to do more direct comparisons to nail which has the darkest color cast.

Cause QD-OLED for example feels like it has almost black color cast.
Typical CCFL and W-LED panels have white color cast.

Another thing: this color cast as I call it - which is because its literally how it looks - is directly related with how black level bothers me. On RGB-LED which doesn't have good contrast ratio black level didn't bother me much compared to displays with better contrast ratio but white color cast. SONY GDM-FW900 with rather bright color cast (always looked like light gray) has much more visible irritating black level issues than IBM P275 which always felt very dark. I put polarizer on FW900 and it made color very dark but didn't affect this quality at all, didn't change colors and black level issues still bother me. IBM P275 with not lighter screen color - still so much better looking image. FW900 had its original AG when I got it and it had different colors than after removing it but visually nowhere near IBM P275.

I literally have in my room bunch of displays which show spectrum between pretty much black to white in this color cast and each monitor shows completely differently looking colors - with displays with similar quality of this color cast showing similar colors.

I really thing we need to just put fourth subpixel and control this color.

----
Also case of WOLED

These displays are driven incorrectly
LG engages W subpixel as soon as they can. Not only this causes near black chrominance overshoot but also because W subpixel is very bright in this rod level it causes extremely big contrast difference between pure black (which obviously doesn't have this color at all since there is no light at all) and dark shades causing black to stick out visually too much causing impression there is sharp edge between still dark gray and black.

QD-OLED doesn't show this issue. Black doesn't at all stick out and blends nicely with surrounding dark shades.

The issue WOLED in question is much more visible when looking at image not directly but slightly off center where we have like hundred times more sensitivity of rods because we have lots of rods compared to fovea where there is barely any rods. So how much I see this issue depends on how I focus my visual consciousness. It is simply best to not try to see anything other than looking just straight and avoid anything that even remotely look black.

What could be done is using RGB subpixels much more for dark shades and only engage W subpixels much later on when they are actually needed to bring luminance up. This would work especially well since WOLED light characteristics with this bright apperance works extremely well for bright daylight scenes which look absolutely great on WOLED. Instead W subpixel is engaged too early leading to this issue - but also to the so called near-black chrominance overshoot - which BTW name makes no sense for what it describes which is flashing W subpixel as it starts being used. Anyways using RGB subpixels in this case would mitigate the issue - not eliminate it but past some level things can be non-perfect. W subpixel is just too bright and it flashing is way too visible. All three RGB subpixels should be much darker together and make this effect not really visible.

As for RGB subpixels used versus this issue with black sticking out from surrounds - I am actually not sure how much it would help because it is hard to assess how RGB image looks using RGB subpixels alone without W subpixel on these panels. Judging by fully saturated colors on WOLED panel I would say this issue should be somewhat less visible.

Perfect display would have QD-OLED RGB subpixels and something like W subpixel and of course all four would be controllable. This is exactly what I mean by 4 color components display.
LG made panels which have this capability to some (probably rather small degree) but don't even attempt utilizing it and even introduce more obvious because using W subpixel is more efficient energetically and using W much brighter subpixel at lower brightness versus using three RGB subpixels also reduces burn-in.

This is however LG and their monitors/TVs. Apparently some other (at least TV) manufacturers do use RGB subpixels to eliminate this flashing of W subpixels. How much they use them to mitigate other issue or even just set look of colors is unclear to me. What is clear to me is that on at least my monitor LG 48GQ900 LG totally fails to drive panel correctly and since this W subpixel flashing is in all their TVs I guess they just cannot help themselves to drive panel incorrectly.

Interesting study about role of rod stimulation in color perception https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2630540/
To be honest language used there is way above my paygrade. I can only notice differences in colors and light level in displays which seems to stimulate rods differently from their RGB (and also W for WOLED displays) subpixels. At least I find rod stimulation theory to be the best match to what I see.

Otherwise yesterday I did some more testing with WOLED panel and ambient light (read: strong ceiling lamp about 4 feet distant in both horizontal and vertical axes - so also at angle of ~45 degrees) and the jarring perceptual difference between black and dark greys is the more mitigated the lamp luminance is higher. Since my lamp has two sets of W-LEDs with different color temperature I checked if temperature affects the effect and yes it does. Colder light seems to work much better than warm light. I don't particularly like very cold light so I settled at the middle 6500K-ish temperature.

For the WOLED vs QD-OLED debate to me then it is obvious WOLED is more suited for very bright viewing environments while QD-OLED is suited for very dark viewing environments.
Ideally display should accommodate both even without being true 4 color display. Imho WOLED to a degree could already do it and when there is less ambient light RGB subpixels should be used for darker tones. How much help this would be on current displays I am not sure and there is no easy way to check that without having an actual control over individual subpixels.

Next step for me in this particular madness is getting spectrometer and measuring actual light spectrum for different displays for each RGB subpixels and white and try to figure out how different displays achieve similar rod stimulation levels despite using at times completely different technologies. Also what is the difference between monitors like SONY GDM-FW900 and IBM P275 which should for all intents and purposes have the same look using SMPTE-C phosphors but are about as far from each-other as WOLED is from QD-OLED - or at least displaying color white.

I wrote this post after reading certain explanation about why Pioneer Kuro and OLEDs (WOLEDs - only OLED available at the time) display colors which are so different despite being both calibrated.
Some people seem to have realized the difference is in the light spectrum but still ignoring cyan elephant in the room - or at least not mentioning either it or rods directly. Apparently it isn't so easy to figure out

Note: Since it is hard topic I asked GPT to refactor my post. Original if someone is up to it in the spoiler below.

--------------
On Color Tuning with Past and Current Displays

Phosphor-based displays, like plasma, allowed engineers to manipulate RGB subpixels with a mix of different phosphors. This enabled precise control over the light wavelengths emitted by each subpixel, aligning them with the typical stimulation of photoreceptive cells found in natural scenes.

LCD displays, however, use backlights with simple color filters. This means that, by nature, you can't get significant cyan light from a red subpixel. Some control is possible indirectly through gamut correction. Even with the same color filters, you can adjust the backlight and mix RGB subpixels to control the cyan light emitted, mainly by adjusting the green component. Cyan-green light is crucial here.

WOLED panels include a white (W) subpixel, always mixed so that only three subpixels are used at the same time. This suggests a lack of consideration for stimulation ratios. If LG engineers had focused on this, they might have utilized four subpixel configurations. The W subpixel's value in reducing burn-in and power consumption would diminish if it was only used for bright colors. Thus, RGB subpixels need to be used together as much as possible, leading to effects no better than a typical W-LED display.

The best way to control cyan (or cyan-green) light is by adjusting the peaks of narrow light emitters. In an RGB-LED backlight display, while RGB color filters mix colors, the narrow peaks of RGB diodes can be tuned to achieve the right cyan light. QD-OLED displays offer more control, as quantum dots can precisely control the light's peak without needing color filters.

P.S. "Gamut clamping" means displaying the same colors. These displays mix RGB subpixels to achieve a wide gamut, but through gamut clamping, you can expect the same cone stimulation.


WOLED Displays by LG

LG designed WOLED displays to make OLED technology viable for mass production, but this has compromised color quality. While the colors can be better than typical LCDs in terms of contrast and viewing angles, they might not appear the most natural. You might be impressed initially by the black levels and novel colors due to the unfiltered W-subpixel, but this effect fades quickly. When compared to a high-quality display like the Pioneer Kuro, these OLEDs can seem unimpressive.

No such issues with QD-OLED displays. Placing two QD-OLEDs side-by-side shows colors that, while slightly different, are very close to each other.


The Perfect 4-Component/Color Display?

To achieve this, minimize the cyan light from each RGB subpixel. Use longer wavelengths for red, shorter for blue, and move green away from cyan (but not too much, or green becomes yellowish). Add a white subpixel with as much cyan light as possible while keeping white still white.

Ideally, we’d have multiple greens, but just enough to reproduce necessary saturation. For instance, moving green slightly can hit the standard most useful for various color spaces. Compared to QD-OLED, if we only cared for Rec.709, we could move green more than if we also wanted DCI-P3.

With optimized blue, red, and green, and a white subpixel focusing on cyan light (giving a 6500K white), it would be possible to color correct such a display using RGB subpixels and control rod stimulation by mixing the white subpixel. This way, you could emulate Pioneer Kuro LX5090 colors by adjusting the W subpixel to RGB ratio. With a touch of a button, switch to the WOLED panel display style.

Ultimately, full RGBW subpixel control unlocks true potential, not just RGB emulation.

----
Original:

On color tuning with past and current displays.

What phosphor based displays like plasma allowed engineers to relatively easily do and which other technologies like LCD and WOLED don't allow to such degree is to have each RGB subpixel have combination of different phosphors making it possible to have better control of how much different light wavelengths come from R, G and B subpixels and match them to typical average stimulation of photoreceptive cells you would see in typical scene from common light sources. LCD for example has some light spectrum from backlight and then you have simple color filters - so by the nature of it you cannot e.g. have much cyan light coming from red subpixel. There is however some control possible - in this case indirect and coming from gamut correction - even with same color filters you can choose adjust backlight to together with gamut clamping which mixes RGB subpixels get different amount of cyan light coming from each subpixel by mostly controlling how much green you need to add - it is cyan-green color that we care the most anyways.

WOLED panels also have W subpixel which is always mixed to only ever use three subpixels at the same time - and this proves to me one thing: there is zero consideration for ratios of stimulation in these panels! If LG engineers cared they would certainly have use cases for four subpixel configurations. I just feel in this case value of W subpixel in reducing burn-in and power consumption would disappear because they would need to only use it for really bright colors and just use RGB subpixels together as much as possible and still get effects no better than typical W-LED display - which WOLED panels are very close to - in fact you could make WRGB W-LED display and get similar colors.

Easiest and best way to control amount of cyan (or cyan-green light) is by moving peaks of narrow light emitters. In case of e.g. RGB-LED backlight display you have imperfection of RGB color filters which mix colors - but since peaks are so narrow and you can (at least in theory - not so much in reality) tune peaks of RGB diodes you can arrive at combinations which together with filters and gamut clamping arrive at right amount of cyan light. You have even more control in case of QD-OLED displays - quantum dots being literally something with which you can control peak of resulting light and you don't have to worry about color filters.

p.s. "gamut clamping" is short at displaying the same colors - these displays just have wide gamut and to get to the same colors need to mix RGB subpixels - with gamut clamping I can just display the same RGB image and expect cone stimulation to be the same.
---
WOLED the way LG designed these displays just to make OLED viable for mass production has pretty bad colors. Compared to most typical LCDs colors can be better - at least contrast and viewing angles are better but even then it would not be the most natural looking display in the shop. You would however be likely wow-ed to get one seeing amazing black levels and colors themselves maybe if they look inaccurate are quite specific to these displays and something novel even compared to most W-LED based displays due to using unfiltered W-subpixel. It is however something that goes old very fast and when comparing it to good display like Kuro these OLEDs look pretty unimpressive.

No such issues with QD-OLED.
I can put QD-OLED next to each-other and colors while slightly differently tuned are pretty much very close.

---
Perfect 4 component/color display?
Would need to minimize how much cyan light comes from each RGB subpixel - so make red use longer wavelengths and blue shorter. Move green away from cyan - just not too much because then green becomes yellowish. Then add white subpixel with about as much cyan light as you can shove there while making white still white.
For green we ideally had multiple greens but we really only need to reproduce certain amount of saturation so just move it slightly before we hit standard which is most useful. Compared to where green is for QD-OLED if we would only cared for Rec.709 for compatibility we could move it more than if we also wanted DCI-P3 - so we could hit lower level of rod stimulation for this more yellowish greens.

Anyways, with blue and red further apart from rod stimulation and optimized green to also move away slightly and then with white that optimizes cyan light (which would imply using such filter which has all colors close and focused as close to cyan-green as possible and giving 6500K white) it would then be possible then color correct such display using RGB subpixels and by mixing white subpixel control amount of rod stimulation.

Then it should be possible to display colors exactly like I see them on Pioneer Kuro LX5090 just by adjusting ratio of W subpixel to RGB and then with a touch of a button display them like WOLED panel does.
Of course it would be just RGB emulation - with true power being unlocked only when all RGBW subpixels can be controlled directly.

I had lengthy chat with GPT about eyes and perceived colors.
GPT rather put its bets in SPD (Spectral Power Distribution) basket. It was hard to convince it

Conclusion and Potential Next Steps​

In conclusion, it seems likely that rods do play a role in the phenomena you're describing, even in photopic conditions. The white color cast you're experiencing could result from additional rod-driven luminance signals overlaid on the cone-driven color image, and different SPDs might stimulate rods differently across displays. While SPD differences do affect cones to some extent, rod stimulation in photopic conditions seems like a plausible explanation for the desaturation and "whitening" effect you're observing.

Further research into how rod signals integrate with cone signals in real-world lighting and display scenarios could shed more light on this issue. It might be interesting for you to experiment with lighting conditions or displays with known SPDs to see if you can isolate the white color cast phenomenon.

Click to expand...

SPD in this case apparently cause different responses from cones and rods depending on light energy. I brought point that SDP might contribute to sharp edges of sensitivity diagram and quickness of tiring of eyes depending on energy and less so to hue because as GPT said there is aggregation step in ganglion cells. I assume brain would be overworked processing different types of signals compared to actually useful information such design would provide. We had to differentiate red apple from green apple and sky from ground.

As GPT the SPD effects also happen in rods. Something I didn't really think about much but might make some sense to explain some things. Generally I don't dismiss SPD totally but imho for making better color display focusing on such things is silly and only puts emphasis on mostly useless but hard to engineer solutions. Adding W subpixel we can already do. Just make RGBW subpixels with e.g. quantum dots tuned for blasting white with tons of rod light and mix it as needed. LG in the case of their displays could at most slightly adjust light spectrum to slightly change dark tones and not adjust it in useful range. It would still probably be beneficial but how much it is hard to say without spectrometer and measuring individual RGB subpixels.

There was also point about rod stimulation causing change in hue in direction of colder colors. I had to explain step by step why this doesn't happen and why it happening doesn't make any sense.

White color cast in conclusion... or rather quoted as 'whitening' is as it sounds. Just to not think it has anything to do with white let me just mention that true white is when both cones and rods get the same stimulation from sufficiently uniform reflective source illuminated by the Sun. That white color cast that rods cause is at most white-like. Displays are hit or miss. Displays like RGB-LED or QD-OLED so those which have low rod stimulation display is also white-like and in this case different type. In fact it misses exactly this white color cast from rods. Should be obvious really.

p.s. Convincing GPT isn't any accomplishment of course. I just updated information in case anyone was interested.

It seems multi-primary color displays are gaining attention.
https://en.ubiresearchnet.com/multi-primary-color-display-technology-2026/