U2410: Grey gradients have pinkish stripes after calibration.

Have you tried to calibrate:

- without "black point compensation" (i get lower dE error without)
- use the "as measured" whitepoint
- with gamma as close to your U2410 native gamma as possible (check with Tools->Report on uncalibrated display device)

To save time when trying different settings try low quality.
 
profiles are good and with high color precision. Using then on my LCD introduce slight banding as usual on 8bit. On CRT gradients are smooth as can be.

so there is nothing to do except using VGA or changing to professional card with 10bit DP connection... or changing monitor for one with hardware calibration...
 
I might have found a solution to your problem but only a test will tell. Remove the "Black point correction" which controls how much color correction is added to the blacks (darkest greys). To disable "Black point correction" you:

1. Options -> Show advanced calibration options
2. Untick Black point correction

I am curious to hear the result :D
 
so there is nothing to do except using VGA or changing to professional card with 10bit DP connection... or changing monitor for one with hardware calibration...

So, basically, I'm out of options. VGA doesn't help. Changing a card is not an option, I only recently upgraded to GTX 680 (and paid a hefty $720 equivalent in euros for it). So, what's left? Sell the monitor and buy what, a $1000+ NEC? I don't know any real alternatives in the same price range as U2410. Maybe except ZR2440W, but that monitor also doesn't have hardware calibration support.
 
I might have found a solution to your problem but only a test will tell. Remove the "Black point correction" which controls how much color correction is added to the blacks (darkest greys). To disable "Black point correction" you:

1. Options -> Show advanced calibration options
2. Untick Black point correction

I am curious to hear the result :D


Since the problem affects the whole grey ramp, not only dark greys, I doubt it would help. But I'll test it, nevertheless. Thanks for your input!
 
"Black point correction"

OK, I dug up some info on Black Point Compensation (I think that's what you meant) and it appears to be Adobe's proprietary addition to ICC profiles, which has only one purpose: to map printed black (which is usually something like R10 G10 B10 or more) to pure RGB black (R0 G0 B0). It is always on for perceptual rendering (the one I normally use), even if you uncheck it in the options.

Bottom line is that BPC doesn't try to "correct" black or dark greys, mixing them with some other colors or something like that. All it does is replace "shifted" black with pure RGB black wherever applicable.
 
with gamma as close to your U2410 native gamma as possible

You are correct, the closer the gamma is to the monitor's native gamma, the less banding occurs. One time I calibrated ignoring the gamma setting altogether, and I got ideal gradients, with zero banding and zero coloring artifacts.

The problem is, the native gamma curve on U2410 is WAY off in Standard Preset (which provides the best colors). It's closer to 2.2 in Adobe RGB and sRGB, but not ideal.

Basically, as a tradeoff, I could calibrate either aRGB or sRGB modes ignoring the gamma setting entirely and in that case I would get impeccable grey ramp at the expense of a less accurate tonal response. This is an option worth to think about.
 
I read this thread on Luminous Landscape http://www.luminous-landscape.com/forum/index.php?topic=63719.0 and saw post 5:

For those not familiar with Argyll, the default configuration is with -K set to 0. The -K variable controls how much color correction is applied to the blacks. Most software packages do apply any color correction to the blacks and this is certainly the case with the Spyder 4 software, Color Eyes Display Pro, BasICColor Display, etc. Spectraview II does provide an option that allows to to calibrate blacks but the default is set to off (and I'm not sure but this may no longer be supported with the PA series monitors). The point here is that ultimately, this won't play a huge roll in the real world but it does demonstrate the limitations of the Spyder 4. Getting back on track, when -K1.0 is specified, the Spyder 4's weakness in the shadows becomes apparent as the resulting profiles contain lots of color casts in the shadows where as the i1 Display Pro performs quite admirably. Again, most software packages DO NOT do this so in most cases you're not going to hit this limitation.

The -K switch in Argyll is "Black point correction" in dispcalGUI and i think it is on although it should b off in the latest dispcalGUI version. You can also give version 0.9.6.6 of dispcalGUI a try which should have "Black point correction" off by default on LCD displays.
 
@stefanp67

Thanks for the dig up, it may prove useful, indeed.

Just to make sure we're on the same page here. Do you mean this by the "black point correction"? (screenshot below)

bpc.png


If this is it, then it should be no problem switching off from the GUI, right? Why would I need some older version for doing that?
 
No it's the one above the calibration quality named "Black point correction". But you can untick "Black point compensation" also.
 
OK, I calibrated with "black point correction" set to zero. There is a notable difference in color and greys look positively more neutral now. Alas, the faint pinkish stripes are still distinguishable. Maybe I should try turning the "black point compensation" off too and see if it makes any difference.
 
I had the same problem with my Eizo S2433W monitor,
after calibrating the monitor I got pinkish stripes on gray tones.

I solved by changing my monitor panel. It was defective.
 
Defective panel sounds plausible since out of all monitors i have calibrated with my Spyder4 (>10 different models) i have never gotten pinkish stripes on any of them.
 
I solved by changing my monitor panel. It was defective.

Dell is not Eizo. I can't even imagine explaining the nature of this defect to the clueless folks at Dell's support. In the best case scenario, I'll get a refurbished crap with far worse problems than pinkish stripes on calibrated grey ramp.
 
Dell is not Eizo. I can't even imagine explaining the nature of this defect to the clueless folks at Dell's support. In the best case scenario, I'll get a refurbished crap with far worse problems than pinkish stripes on calibrated grey ramp.

this is also true. in any case the pink stripes depends on the inability of the panel to display good grey tones.
my panel displayed 92% of grey tones and the calibration process tried to adjust the problem creating the pinkish effect. now that I changed my panel all is ok.
 
this is also true. in any case the pink stripes depends on the inability of the panel to display good grey tones.
my panel displayed 92% of grey tones and the calibration process tried to adjust the problem creating the pinkish effect. now that I changed my panel all is ok.

One thing I don't understand: If my panel is unable to reproduce good (read: accurate) grey tones, how come it does it in native, uncalibrated modes? Logically, if a panel can't display greys normally, it shouldn't be able to do it in ANY mode or setting. Am I missing a point here?

You say your panel displayed 92% of grey tones. How exactly did you measure it? Please elaborate -- maybe I could make the same measurements on my panel and see what coverage it provides.
 
One thing I don't understand: If my panel is unable to reproduce good (read: accurate) grey tones, how come it does it in native, uncalibrated modes? Logically, if a panel can't display greys normally, it shouldn't be able to do it in ANY mode or setting. Am I missing a point here?
.

Where is the big Denis, he could explain you this better than me. I miss him :D
When calibrating your monitor loss some color gamut because it restrict its ability to display colors in order to have a more precise color reproduction of the more important colors.

You say your panel displayed 92% of grey tones. How exactly did you measure it? Please elaborate -- maybe I could make the same measurements on my panel and see what coverage it provides.

After every calibration I use the UDACT test to certify my calibration, this is a good tool
to see that all is good.
http://dl.dropbox.com/u/17948666/Report20120506 21-55.pdf
 
I have downloaded the UDACT software. It seems to be a great tool, indeed, but unfortunately it doesn't allow any actual measurements until it's registered. I've sent a registration request for a limited license. Let's see what happens.
 
I have downloaded the UDACT software. It seems to be a great tool, indeed, but unfortunately it doesn't allow any actual measurements until it's registered. I've sent a registration request for a limited license. Let's see what happens.

keep us informed.
 
OK, here's the UDACT test results:

Uncalibrated:

Tone values = 99.6%

Calibrated:

Tone values = 91.2%

According to UDACT, "a display for the printing industry should show at least 95% of the incoming tone values." So it's safe to say that the test failed.

But I still don't understand how the fact that the video card (which, according to the same UDACT, is responsible for those tone values after calibration) gives lower tone values could serve as a proof that the display has some physical defect? I really could use some explanation here, preferably in layman's terms.
 
OK, here's the UDACT test results:

Uncalibrated:

Tone values = 99.6%

Calibrated:

Tone values = 91.2%

According to UDACT, "a display for the printing industry should show at least 95% of the incoming tone values." So it's safe to say that the test failed.

But I still don't understand how the fact that the video card (which, according to the same UDACT, is responsible for those tone values after calibration) gives lower tone values could serve as a proof that the display has some physical defect? I really could use some explanation here, preferably in layman's terms.

You finded the problem.
The calibration process needs to "adjust" the grey tones too much resulting in a pinkish color.
Are you sure that your calibration intent is correct?
Your temperature is 6310, not really good, can't you achieve a better value nearer to 6500K with the RGB calibration process?

Why the luminance is so high? 207.6 Cd/m2 is really high, why don't you try to achieve the 120cd/m2?
 
You finded the problem.
The calibration process needs to "adjust" the grey tones too much resulting in a pinkish color.
Are you sure that your calibration intent is correct?
Your temperature is 6310, not really good, can't you achieve a better value nearer to 6500K with the RGB calibration process?

Why the luminance is so high? 207.6 Cd/m2 is really high, why don't you try to achieve the 120cd/m2?

The intent is perceptual. I think it's the best intent for LUT-based profiles, but I'm no expert.

Some of my calibrations come very close to 6500K. It can be achieved.

I don't calibrate luminance. I adjust brightness manually, because my ambient light often changes throughout the day. Brightness was set to 30 during that particular measurement (this monitor is very bright even on low brightness settings).

Why, do you think the temp and luminance could have some influence on grey accuracy? It may be so, but a slightly warmer temp or high luminance can't really cause pinkish stripes, can they?

I can recalibrate with a more accurate D65 and set luminance to 120cd/m2. Let's see what UDACT says after that.
 
The intent is perceptual. I think it's the best intent for LUT-based profiles, but I'm no expert.

Some of my calibrations come very close to 6500K. It can be achieved.

I don't calibrate luminance. I adjust brightness manually, because my ambient light often changes throughout the day. Brightness was set to 30 during that particular measurement (this monitor is very bright even on low brightness settings).

Why, do you think the temp and luminance could have some influence on grey accuracy? It may be so, but a slightly warmer temp or high luminance can't really cause pinkish stripes, can they?

I can recalibrate with a more accurate D65 and set luminance to 120cd/m2. Let's see what UDACT says after that.

I don't know the software you are using but if you don't try to achieve 6500k and some luminance you like, what are you calibrating?
What do you do in the calibration process if not this?

The pinkish problem is a conseguences of your tone values dropped to.91% after calibration.
You should find a better calibration settings.that can let you loose less tones values
 
I don't know the software you are using but if you don't try to achieve 6500k and some luminance you like, what are you calibrating?
What do you do in the calibration process if not this?

I do try to achieve it. I always adjust RGB values manually and I get minimal white point deviations (around 0.2 dE) during interactive adjustment. What else can I do?

As for the luminance, I don't regard it as an essential part of calibration and normally ignore it. I prefer adjusting my brightness settings manually, according to the ambient light conditions. Am I wrong in doing that?

I do calibration mostly for two things: color accuracy and tonal response close to 2.2.

As for calibration settings, my current settings (LUT + matrix, perceptual intent) should be the best for wide gamut monitors. But I'll try curve-based profiles and see if it helps.
 
I don't know the software you are using but if you don't try to achieve 6500k and some luminance you like, what are you calibrating?
What do you do in the calibration process if not this?
calibration is about:
- proper gamma
- identical temperature on all grey levels
- measuring gamut (and if monitor is good also to correct it)

temperature of 6500K in not that important actually and brightness level is not important at all :eek:

all what is needed is to make sure that all monitors in workflow are as close to eachother as possible. If one is 7000K and 200cd/m2 then others should be identical. Exception can be made if monitor is used just as preview eg. web designer might want to see how his page actually looks on average W-LED TN crap people are using :D
 
Last edited:
What else can I do?

calibration is not about:
- proper gamma
- identical temperature on all grey levels
- measuring gamut (and if monitor is good also to correct it)

What I am saying is that generally a good calibration software ask you what is your calibration intent, generally it ask you for the gamma, the temperature, the luminance, the desidered black levels.

If you say to the software (or if the software assume it) that the calibration intent is
6500K, 2.2 gamma, 120cd/m2, 0cd/m2 black levels, the profiling process will try to be as much as compliant to the initial intent.

This means that if you don't calibrated the monitor well to the initial intent parameter, the software needs to heavily work on icc profile, it needs to do more work, resulting in loosing more tonal values.

If I set the "web defaults" as a calibration intent and than I do not se the luminace to 120cd/m2 manually on my monitor, my profiling software try to lower my brightness using ICC resulting in a crappy calibration. This because the profiler thinks that "web" settings equals to 120cd/m2. I don't know if I explained the concept well but this is one of the reason why you should know your software well before starting a calibration.

Before calibrating the software you need to know your software well, once calibrated the monitor according to the software calibration intent, you'll get the best profiling process available for your instrumentations (monitor colorimeter).

Another important thinks is that the brightness is directly related to color temperature
and black levels.
If you set the brightness to 120cd/m2, than you calibrate and profile the monitor, than you raise the brightness, your previois calibration is broken since your color temperature is changed. Gamma curves changes accordingly to brightness specially on non professional monitors.

Once you have done all as I suggested (that is the normal good way to calibrate a monitor) if the pinkish remains you have some options.
1) Try to higher the desidered black levels to 0.20cd/m2, this helps the profiler to don't try to adjust the unadjustable.
2) Try to calibrate using an sRGB intent and not a 2.2 intent. The sRGB gamma curve is different from the 2.2 one, generally some monitors like it more.
3) When you calibrate the red green blu to achieve the 6500K try to not lower the settings too much. I'll try to explain this better, if you can achieve the 6500K with R=95 G=90 B=99
and you can achieve the same 6500K with R=90 G=80 B=94, choose the first one, always choose the "bigger value" for your calibration intent. This generally produce less tonal loosing in the profiling process, at least from my eperience with all the monitors I calibrated.
 
1) Try to higher the desidered black levels to 0.20cd/m2, this helps the profiler to don't try to adjust the unadjustable.
2) Try to calibrate using an sRGB intent and not a 2.2 intent. The sRGB gamma curve is different from the 2.2 one, generally some monitors like it more.
3) When you calibrate the red green blu to achieve the 6500K try to not lower the settings too much. I'll try to explain this better, if you can achieve the 6500K with R=95 G=90 B=99
and you can achieve the same 6500K with R=90 G=80 B=94, choose the first one, always choose the "bigger value" for your calibration intent. This generally produce less tonal loosing in the profiling process, at least from my eperience with all the monitors I calibrated.

1) Got it.
2) Not sure about sRGB. It's a wide gamut monitor and it's highly unlikely that the sRGB curve is what it "likes more".
3) My RGB values when setting white point are well over 200 (it uses direct values, not gain percentages), so I think everything's OK in that department.

I disagree about luminance though. It's very inconvenient to calibrate luminance as well, as you suggest. Because, due to significant changes in ambient light, I would have to make different profiles for daylight, evening and night and constantly switch between them, which is highly inconvenient. If I calibrate to a target luminance, I'm bound to keep monitor's brightness at the same level as during calibration. There's a reason why calibration software allows ignoring the luminance entirely.

Still, I agree that in many cases high luminance may affect gamma curve and even color temps. I may find a compromise by setting the luminance target higher than 120 cd/m2 -- say, 160 -- and keep it that way. 120 is too dark for me, especially during daytime.
 
Please read this:

Here is my last calibration result for the sRGB preset.
I achieved it with RGB Gains at 255-255-255 in the service menu, 10 bit DP connection, and "absolute" gamma scaling in Dispcal.
The calibration itself took ~23 minutes and the profiling took another ~15 minutes (but you can choose much more iteration points for the profiling :D).

Current calibration response:
Black level = 0.14 cd/m^2
White level = 122.66 cd/m^2
Aprox. gamma = 2.20
Contrast ratio = 873:1
CIE chart from the emulated sRGB gamut

It's the same monitor as mine. As you see, very good results can be achieved by calibrating this monitor, and I'm aiming for it eventually. I just don't know how. Unfortunately, the original poster hasn't been on the forums since February and I can't contact him for more detailed info. Especially, the part about 10-bit connection.
 
I really wonder what he meant by "absolute" gamma scaling. Something tells me it's the key ingredient of his magic potion.

EDIT: Never mind, I figured it out. :) It's absolute gamma setting in dispcalGUI. I just hadn't noticed it before for some reason.
 
Last edited:
1) Got it.

Does it solved the problem?

2) Not sure about sRGB. It's a wide gamut monitor and it's highly unlikely that the sRGB curve is what it "likes more".

My Eizo monitor is a wide gamut one but it loose less tonal values when calibrating it to sRGB :)

I disagree about luminance though. It's very inconvenient to calibrate luminance as well, as you suggest. Because, due to significant changes in ambient light, I would have to make different profiles for daylight, evening and night and constantly switch between them, which is highly inconvenient. If I calibrate to a target luminance, I'm bound to keep monitor's brightness at the same level as during calibration. There's a reason why calibration software allows ignoring the luminance entirely.

Still, I agree that in many cases high luminance may affect gamma curve and even color temps. I may find a compromise by setting the luminance target higher than 120 cd/m2 -- say, 160 -- and keep it that way. 120 is too dark for me, especially during daytime.

you can disagree but this is a fact. :)
when a software allows you ignoring the luminance means that it took the luminance of your monitor as a reference but if you change the luminance after the calibration you entire calibration is broken.

you can choose the luminance before calibration but you can't change it after the calibration process without broking the entire calibration intent.
 
It may be a little offtopic but i tried the UDACT on my calibrated Samsung S24A850DW and got:







Maybe not Eizo ColorEdge / NEC Spectravideo class but tonal values are rather good at 97.9%.
 
Holy crap! I had to turn down the brightness to 0 (zero!) in order to achieve 120 cd/m^2 during interactive adjustment. I haven't done the actual profiling yet, but the colors surely look unpleasantly dim at zero brightness.

I'll do the calibration tomorrow and report the results.
 
Holy crap! I had to turn down the brightness to 0 (zero!) in order to achieve 120 cd/m^2 during interactive adjustment. I haven't done the actual profiling yet, but the colors surely look unpleasantly dim at zero brightness.

I'll do the calibration tomorrow and report the results.

I always saied that this was an overpriced monitor on its lunch :D
 
Back
Top