OLED yellow vertical edge artifacts

raymod2 I just bought this monitor (PG42UQ). I am absolutely in love with it, but I noticed the exact same issue. I almost RMA'd it until seeing this post. Glad I found it. I will not return the monitor. Despite this issue, its a gorgeous screen.

Thank you for the detailed images. Knocked the screenshots I was taking for ASUS out of the water.
 
I have a 55" LG G2 TV, which I use for watching movies and also as a PC monitor.
The problem with the yellow/orange/lime colored objects is very obvious and highly disturbing in Windows 11.
For a video content it's negligible, but any static element makes it clearly visible, and depending on TV/signal settings it might be somewhat weaker or stronger, but it never disappears or gets faint enough.
The main problem is that the manufacturer (LG) obviously didn't give a sh*t about solving this issue even if the TV switches to game mode and/or the source is clearly signed as a PC. WRGB is not a new technology, and using TV as a monitor is also functioning idea, at least in the last 3-4 years. Although it is clearly a hardware based problem, a highly satisfying software/firmware solution is undoubtedly possible. There is no excuse for not solving this problem as a manufacturer (LG) and as an OS developer company (Microsoft, Apple). Considering the fact that nowadays more and more monitors appear on the market with the very same panel technology and with the same issue, I really can't understand how is it still not solved?!

I attached some examples.
 

Attachments

  • 20230308_120717.jpg
    20230308_120717.jpg
    446.7 KB · Views: 0
  • 20230308_120811.jpg
    20230308_120811.jpg
    640.3 KB · Views: 0
  • 20230308_120951.jpg
    20230308_120951.jpg
    901.7 KB · Views: 0
MaTaTom said:
I have a 55" LG G2 TV, which I use for watching movies and also as a PC monitor.
The problem with the yellow/orange/lime colored objects is very obvious and highly disturbing in Windows 11.
For a video content it's negligible, but any static element makes it clearly visible, and depending on TV/signal settings it might be somewhat weaker or stronger, but it never disappears or gets faint enough.
The main problem is that the manufacturer (LG) obviously didn't give a sh*t about solving this issue even if the TV switches to game mode and/or the source is clearly signed as a PC. WRGB is not a new technology, and using TV as a monitor is also functioning idea, at least in the last 3-4 years. Although it is clearly a hardware based problem, a highly satisfying software/firmware solution is undoubtedly possible. There is no excuse for not solving this problem as a manufacturer (LG) and as an OS developer company (Microsoft, Apple). Considering the fact that nowadays more and more monitors appear on the market with the very same panel technology and with the same issue, I really can't understand how is it still not solved?!

I attached some examples.
Click to expand...

This problem is inherent in the pixel structure and the way the subpixels are lit to show different colors at different levels of brightness. The pixel design would need to be different for a productivity monitor use case. For that, LG had to go with a completely different panel in their professional OLED monitors, which an RGB panel from JOLED.
 
I'm sure any attempt to reduce visible fringing can only be done by either adding or taking away one of the subpixels in certain situations.

This would have the effect of either shrinking or expanding the on-screen element in question, plus the smarts required to know when and when not to do this is likely beyond the capability of the firmware and/or image processing of the panel.
 
Guys go back and read the thread again. I proposed a very simple firmware solution. Simply change the logical grouping of subpixels from RWBG to WBGR. This puts the red and green subpixels next to each other where they can blend to form yellow. This is not a perfect solution but it would be vastly superior to what we currently have. They could make this an optional mode so the user could choose whether to use it or not. It is unacceptable that LG and Asus refuse to fix this problem.
 
Senn said:
I'm sure any attempt to reduce visible fringing can only be done by either adding or taking away one of the subpixels in certain situations.

This would have the effect of either shrinking or expanding the on-screen element in question, plus the smarts required to know when and when not to do this is likely beyond the capability of the firmware and/or image processing of the panel.
Click to expand...
Even changing the width by adding or removing subpixels would be a good solution for the somewhat larger object starting from 5-10 pixels. Or at least an option in the settings would be great, so that everybody could decide if a red-green fringe or a somewhat distorted object is a better solution.

When to do such a modification is really simple do decide, it depends only from the number of the side by side pixels with affected colors. The processors of the TV are nowadays capable of making unbelievable changes of the picture within milliseconds, so making the calculations for standing images should not be a problem at all. I really can't imagine that it is impossible, and can't accept such an excuse.

Even the solution mentioned by raymod2 should be possible, I don't see any reason not doing it. Recalculating the pixels for a SDR desktop use on the fly if there are no other picture altering algorithms active should not be a problem.

I simply can't accept that it is so hard to do. It's not something that you need for every signal source, only in a special situation, when using the TV/Monitor with PC in SDR desktop mode.
 
On the other hand I also see a responsibility of the OS developer companies, and maybe even the graphic card manufacturers. They should also integrate in their drivers special solutions as options for different display types, so the user can choose depending from the panel type and personal preference what he wants to see.
 
lol you go tell nvidia, amd and microsoft that they have to be able to control the tv/monitor's pixel layout. let us know what they say....
 
I have no idea how you think a firmware is able to change the physical pixel structure of a given display. What you're saying is the "fix" has to be done at the manufacturing level. Firmware can't make red OLEDs, (or blue or green) magically become a different color.
 
pendragon1 said:
lol you go tell nvidia, amd and microsoft that they have to be able to control the tv/monitor's pixel layout. let us know what they say....
Click to expand...
Not to control the pixel layout, but to make similar tricks like Clear Type depending on the chosen display type.
 
UnknownSouljer said:
I have no idea how you think a firmware is able to change the physical pixel structure of a given display. What you're saying is the "fix" has to be done at the manufacturing level. Firmware can't make red OLEDs, (or blue or green) magically become a different color.
Click to expand...
Simple, instead of addressing only one pixel with RWBG subpixels, addressing two pixels using WBG+R subpixels. You can think about it like shifting the whole image 1/4 pixel or one subpixel to right.
I don't think it's too complicated to imagine how it should work. Physically there is no extra gap between the pixels sideways, so it would not make the picture worse.
 

Attachments

  • pixels-alt-large.png
    pixels-alt-large.png
    2.6 MB · Views: 1
I think this probably varies a good bit too. For example, I can notice it on my OLED, but for whatever reason, it doesn't really bother me, whereas things like blooming on other panel types really did. Still, it'd definitely be nice if there was a way to mitigate the issue. I'm curious if there's an easy way to address this like suggested why nobody has yet, tho'.
 
From pixel structure alone its obvious they designed these panels foe TVs and PC use wasn't considered at all. Making gaming monitors using the same panel tech is an obvious afterthought.

Without completely removing white subpixel they could make RGBW pixel layout and up until RGB subpixels can create enough brightness they whould be used and W subpixel only engaged when its absolutely necessary.
This wouldn't mitigate issues completely and would drive power consumption sightly but at lower brightness levels text and edges would look correct. Also near black chrominance overshoot would be gone - arguably the worst issue of LG panels.

As it is however it doesn't seem likely they will treat their afterthoughts seriously so let's hope Samsung will do QD-OLED panel with proper RGB. From how things stand their tech is closer to perfection.
 
Last time I read reddit suggestions to use the LG C2 42" TV as a monitor.
I just got this TV and immediately noticed how blurry text and ugly yellow color has an extra dash of red and green.
At least it's great for gaming or watching movies.. So I will keep it.


1.png
2.png


3.png
4.png
 
Honestly, versus the headache inducing pixel bleeding issues that bugged me with all the current 43" LCDs, I could live with that.

Get rid of static image burn in concerns on OLED (not happening, I know) and I would for sure.

For now, the self-emissive waiting game and continued contentedness with what I have.
 
pastuh said:
Last time I read reddit suggestions to use the LG C2 42" TV as a monitor.
I just got this TV and immediately noticed how blurry text and ugly yellow color has an extra dash of red and green.
At least it's great for gaming or watching movies.. So I will keep it.
Click to expand...
Enabling grayscale font rendering helps with ugly discolored text
 
Although I am also looking to converge to a solution manufacturer-side, I'm also seeing if I can push things Microsoft-side too, in a "who gets there first" feat.

You may wish to upvote this possible "ClearType 2" PowerToy for custom OLED pixel structures.

https://github.com/microsoft/PowerToys/issues/25595

I have created a photoshop image demo, that "RBG" simulated ClearType works on LG WOLED, as one example; since RWBG translates to RBG ClearType if you ignore the W subpixel that software can't control directly. It does not address the yellow fringing behavior (display-side changes needed)

custom-RBG-cleartype-for-WOLED-displays.png

IMPORTANT: ClearType contrast setting is currently at an extreme setting and may produce minor sharpen-style artifact; a real font renderer will be adjustable.
 
Last edited:
Chief Blur Buster said:
You may wish to upvote this possible "ClearType 2" PowerToy for custom OLED pixel structures.

https://github.com/microsoft/PowerToys/issues/25595
Click to expand...
Reading this and I have no idea of what "feature regression" you are talking about.
ClearType of old wasn't any more configurable and able to accommodate different subpixel structures in previous versions of Windows.

Regarding
"ClearType 2" PowerToy
Click to expand...
would require major rework of ClearType itself. While it is possible MS will do it I wouldn't bet on it.

Test image itself imho 'Special "RGB" "ClearType"' doesn't look good at all. There is red edge on left of black white part and text looks like it had slight sharpening filter on it. Grayscale looks better.

On that matter on my RGB monitor grayscale looks as good as RGB ClearType. Testing it in native gamut to give ClearType best chances.
I run both ClearType and MacType with grayscale and imho its best readily available solution to all subpixel issues.
 
XoR_ said:
Test image itself imho 'Special "RGB" "ClearType"' doesn't look good at all. There is red edge on left of black white part and text looks like it had slight sharpening filter on it. Grayscale looks better.
Click to expand...
It's an artifact of the limitation of Photoshopping it quickly -- in reality, it will be much more adjustable like the ClearType "Contrast" setting.

ClearType vs Greyscale is also a matter of personal preference, vision differences between humans, and other factors.
So for those of you with github accounts, go ahead and upvote it anyway.

XoR_ said:
would require major rework of ClearType itself. While it is possible MS will do it I wouldn't bet on it.
Click to expand...
If initially incubated as third party font renderer, you can simply supersample it (e.g. render the font say, about ~16x larger in memory, and subpixel downsample). With sufficient caching of the resulting supersampler's downscample outputs, and modern GPUs, the performance should be quite more than adequate.

The algorithm is surprisingly simple GPU shader, given an appropriate subpixel texture, you run the large size glyph through it, and out comes the subpixel scaled version. There'll be some issues bypassing the TrueType/OpenType small-font hinting systems for certain tiny font sizes, but the subpixel access will tend to more than outweigh that (at least in my internal tests).

Major rework would be needed at the optimization stage, but at incubation, it's rather simple. Could easily be done initially as a reader application (like 1998's Microsoft Reader application-level ClearType support) -- so simple a future GPT4+ AI probably can almost write it already. (Just so you know, GPT4+ can already output correct small shader code samples that works in ShaderToy, upon request of simple filters like brighten-10% or pixel-merging or pixel-filter/sorting-task, etc)

That's how I kind of PhotoShopped it anyway. I also posted the generic algorithm (just today) in the other github.

And it probably will improve further if I did some spatial-correction by supersampling it via R-(blank)-G-B in quarters that is spatially-correct, rather than R-G-B in thirds that is slightly spatial-incorrect (the way I did it in a quick and dirty way). Sampling W would have been ideal, but the W is a hardware-controlled subpixel as a weighting between R+G+B with no software control, so the algorithm can only control the subpixels that software does have control over.

XoR_ said:
Reading this and I have no idea of what "feature regression" you are talking about.
ClearType of old wasn't any more configurable and able to accommodate different subpixel structures in previous versions of Windows.
Click to expand...
Feature regression isn't about nonstandard subpixel structure --

It's about RGB versus BGR.

BEFORE: In some past versions of Windows, you could configure multimonitor -- e.g. RGB on laptop/monitor and BGR on TV.
AFTER: Right now you can't. It's either RGB or BGR globally for all displays.

Subpixel structure, either standard or nonstandard, should be a per-display configuration setting.

I've edited the github to clarify this potential misunderstanding / misinterpretation;
 
Last edited:
Chief Blur Buster said:
It's an artifact of the limitation of Photoshopping it quickly -- in reality, it will be much more adjustable like the ClearType "Contrast" setting.

ClearType vs Greyscale is also a matter of personal preference, vision differences between humans, and other factors.
So for those of you with github accounts, go ahead and upvote it anyway.


If initially incubated as third party font renderer, you can simply supersample it (e.g. render the font say, about ~16x larger in memory, and subpixel downsample). With sufficient caching of the resulting supersampler's downscample outputs, and modern GPUs, the performance should be quite more than adequate.

The algorithm is surprisingly simple GPU shader, given an appropriate subpixel texture, you run the large size glyph through it, and out comes the subpixel scaled version. There'll be some issues bypassing the TrueType/OpenType small-font hinting systems for certain tiny font sizes, but the subpixel access will tend to more than outweigh that (at least in my internal tests).

Major rework would be needed at the optimization stage, but at incubation, it's rather simple. Could easily be done initially as a reader application (like 1998's Microsoft Reader application-level ClearType support) -- so simple a future GPT4+ AI probably can almost write it already. (Just so you know, GPT4+ can already output correct small shader code samples that works in ShaderToy, upon request of simple filters like brighten-10% or pixel-merging or pixel-filter/sorting-task, etc)

That's how I kind of PhotoShopped it anyway. I also posted the generic algorithm (just today) in the other github.

And it probably will improve further if I did some spatial-correction by supersampling it via R-(blank)-G-B in quarters that is spatially-correct, rather than R-G-B in thirds that is slightly spatial-incorrect (the way I did it in a quick and dirty way). Sampling W would have been ideal, but the W is a hardware-controlled subpixel as a weighting between R+G+B with no software control, so the algorithm can only control the subpixels that software does have control over.


Feature regression isn't about nonstandard subpixel structure --

It's about RGB versus BGR.

BEFORE: In some past versions of Windows, you could configure multimonitor -- e.g. RGB on laptop/monitor and BGR on TV.
AFTER: Right now you can't. It's either RGB or BGR globally for all displays.

Subpixel structure, either standard or nonstandard, should be a per-display configuration setting.

I've edited the github to clarify this potential misunderstanding / misinterpretation;
Click to expand...
What are you talking about, hyst run Cleartype app and adjust monitors one after another.
 
vick1000 said:
What are you talking about, hyst run Cleartype app and adjust monitors one after another.
Click to expand...
It's a false behavior at least for Windows 10. The select-monitor doesn't work. The problem really shows up if you have a television set that uses the upside-down B-G-R.

Please see the research (by the author of Better ClearType Tuner) about this feature regression (see Conclusion)
https://github.com/bp2008/BetterClearTypeTuner/wiki/ClearType-Investigations

Also, tons of reddit complaints:
https://www.google.com/search?q=ClearType+Tuner+multimonitor

Perhaps this was fixed for Windows 11, merits a re-test.
 
Last edited:
Found out that this is a popular item by many users at Microsoft Feedback Hub.

For anyone who wants to join the ClearType OLED upvote party at Microsoft Feedback Hub:
Also should be fixed too:
  • https://aka.ms/AAkjw7z -- "ClearType setting for different subpixel layouts for different screens (for second monitor rotated to portrait/vertical position)" (6 upvote)
  • https://aka.ms/AAkcebe -- "ClearType text tuner incorrectly warns about non-native resolution"
 
Last edited:
Hello,

I have corresponded with the author of MacType, and good news.

MacType is customizable to be compatible with Samsung QD-OLED triangular-subpixel structure.

For ClearType lovers who "upgraded" to an a QD-OLED, please test this...

If you have a Dell Alienware AW3423DWF or another QD-OLED, please test this image:

233757764-d157e337-45fb-40da-9de8-3686d310c47d.png


Please specify if 1, 2, 1b, 2b, 1c, 2c, 1d, 2d looks the best on your QD-OLED display.

I'm going to obtain some prototype LG OLED MacType profiles for R-G-B, please stand by. It won't fix the yellow fringing, but will make text clearer for ClearType lovers, who wants to enable that feature in MacType too.
 
Last edited:
You must log in or register to reply here.
Back
Top