Overall reddish cast on Eizo CG243W

[Hening Bettermann]

Hi!

So I have bought this wide gamut monitor and calibrated it with the Eizo software and the DTP94 puck. I have also just made a profile for my 5D2 using the ColorCheckerSG and Argyll_CMS. On my old monitor, a Samsung 193P, the profile gives a very good visual match between the screen image of the 24 patch ColorChecker, displayed in Raw Developer, and the physical chart, illuminated by one of the 2 Solux® lamps that were used to shoot it. But on the Eizo, there is an overall reddish cast, particularly visible in patches row1/column2, 2/1, 2/3 and 2/6.

How can I achieve more correct colors on the Eizo? My main point is not the match between the monitors, but between the Eizo and the chart. If it was the Samsung which was off, I would not care - I had planned to use it for the tools, and home office use. - The Samsung was calibrated with EyeOne software and the same DTP94  year(s) ago.

I have read WillH's advice in thread
http://luminous-landscape.com/forum/index....or*++red*++cast
but this refers to NEC software. I can not see these options in the Color Navigator.

This is on Mac Pro Nehalem 2.66 GHz running 10.6.2.

Hoping for your help - again... Hening.

[tho_mas]

So I have bought this wide gamut monitor and calibrated it with the Eizo software and the DTP94 puck. ...

what are your targets for profiling (white point, luminance, gamma/trc)?

[Hening Bettermann]

what are your targets for profiling (white point, luminance, gamma/trc)?

Hi tho_mas, white point is 6500, luminance 100 cd, gamma 2.2.

[tho_mas]

Hi tho_mas, white point is 6500, luminance 100 cd, gamma 2.2.

hmh, I think this should be within the realms of possibility ...
Hard to say without seeing the screen.
What if you edit the white point in Color Navigator manually to match a neutral white on your color checker under the Solux lamps visually?
Color Navigator->Detail->Adjust manually.
After the manual adjustment Color Navigator measures again to correct the profile; so you can only adjust manually with the DTP94 connected.
I'd note the resulting white point - so the XY values.
Afterwards I would create a new target with 100cd/m2, gamma 2.2 and the respective XY values as white point.
This is what I do and it works perfectly (though with a CG241W).
And set the priority to "gray balance" not to "contrast".

[Hening Bettermann]

Thank you tho_mas, I'll try this ASAP.

[ErikKaffehr]

Hi,

What about synthetic CC card? Try downloading this:

http://www.brucelindbloom.com/downloads/Co...culator.tif.zip and open in Photoshop.

Compare with your CC card. Do you see the same problem?

Best regards
Erik

Hi!

So I have bought this wide gamut monitor and calibrated it with the Eizo software and the DTP94 puck. I have also just made a profile for my 5D2 using the ColorCheckerSG and Argyll_CMS. On my old monitor, a Samsung 193P, the profile gives a very good visual match between the screen image of the 24 patch ColorChecker, displayed in Raw Developer, and the physical chart, illuminated by one of the 2 Solux® lamps that were used to shoot it. But on the Eizo, there is an overall reddish cast, particularly visible in patches row1/column2, 2/1, 2/3 and 2/6.

How can I achieve more correct colors on the Eizo? My main point is not the match between the monitors, but between the Eizo and the chart. If it was the Samsung which was off, I would not care - I had planned to use it for the tools, and home office use. - The Samsung was calibrated with EyeOne software and the same DTP94  year(s) ago.

I have read WillH's advice in thread
http://luminous-landscape.com/forum/index....or*++red*++cast
but this refers to NEC software. I can not see these options in the Color Navigator.

This is on Mac Pro Nehalem 2.66 GHz running 10.6.2.

Hoping for your help - again... Hening.

[Hening Bettermann]

Hi!

Erik, thank you for the tip with the synthetical CC. With it, the experience is slightly different: on the Samsung, the match with the chart is good (and better than with the camera image); on the Eizo, the look is overall muddy, but  with a good match to the chart with the Solux® lamp turned off or dimmed; that is in diffuse (cloudy) wintery noon daylight from a window 3 meters away and snow outside. In this comparison, the chart image does not look "more reddish"; but the mud makes comparison invalid.

The difference between the monitors is also very clearly seen with the Desktop set to solid light gray, as I have on both monitors.

The Samsung was calibrated to max brightness, which was 122 cd. On the Eizo, the muddy look of the CC does not change at all when I choose the 120 (or 80) cd profile which I also made.

What can I conclude from this so far? Obviously, to use the synthetic CC for comparing monitors rather than the camera image. Beyond that? How come that the CC looks more muddy on the Eizo?

I have a feeling that it might be a good idea to do the visual white point adjustment of the monitor in darkness.

Thank you both for your interest. - Hening.

[ErikKaffehr]

Hi,

One thing you may consider is which color space you use. For a time I had an iMac 24" and an Eizo FlexScan SX2461W side by side. The iMac 24" was much brighter. Now, my Eizo was using Adobe RGB while the iMac was using sRGB. I also made the Eizo darker as the iMac was a bit to bright to match my prints, and couldn't be made darker. Both were calibrated using my ColorMunki, clearly shaking my belief in calibration. I think I could match my iMac on the Eizo by using sRGB and increasing brightness.

Coming back to the Macbeth (sorry I mean X-rite) color charts. The problem is that when you take a picture of a CC-card the picture will have a tone curve applied to enhance contrast. It's nice for the eye but makes comparison harder. That's the reason I recommend a synthetic target.

Best regards
Erik

Hi!

Erik, thank you for the tip with the synthetical CC. With it, the experience is slightly different: on the Samsung, the match with the chart is good (and better than with the camera image); on the Eizo, the look is overall muddy, but  with a good match to the chart with the Solux® lamp turned off or dimmed; that is in diffuse (cloudy) wintery noon daylight from a window 3 meters away and snow outside. In this comparison, the chart image does not look "more reddish"; but the mud makes comparison invalid.

The difference between the monitors is also very clearly seen with the Desktop set to solid light gray, as I have on both monitors.

The Samsung was calibrated to max brightness, which was 122 cd. On the Eizo, the muddy look of the CC does not change at all when I choose the 120 (or 80) cd profile which I also made.

What can I conclude from this so far? Obviously, to use the synthetic CC for comparing monitors rather than the camera image. Beyond that? How come that the CC looks more muddy on the Eizo?

I have a feeling that it might be a good idea to do the visual white point adjustment of the monitor in darkness.

Thank you both for your interest. - Hening.

[Hening Bettermann]

I have to add to the above, that the monitor gray looks *lighter* on the Eizo with 100 cd than on the Samsung with nominal 120. I assume that aging is the reason.

Now it is dark in my part of the woods, and I make new discoveries.
The Desktop set to dark gray on both monitors.
4 copies of the synthetic CC opened in PS and Preview respectively on the 2 monitors and compared to the physical CC illuminated by 1 Solux lamp, in the otherwise dark room, show this:

PS CS3:the Eizo image looks more muddy (than the chart and than the Samsung image) in most patches, in particular in patch 3/2 (green), 3/3 (red) and 3/5 (magenta). Patches 1/6 (bluish green) and 2/5 (yellow green) look pale (under-saturated).

Preview: The Samsung image is the one that comes closest to the chart, of all the 4. The Eizo edition is slightly more reddish, but not muddy at all. The reddish cast is in particular visible in patch 2/1 (orange) and 2/3 (moderate red). And also in patch 1/2 (light skin); here, it makes the Preview/Eizo image the best of the quartet.

The PS/Eizo image is way out muddy and impossible. All the other 3 are closer to each other and the chart, Preview/Samsung being the best, but all 3 would be acceptable.

I find it very difficult to judge the gray scale in comparison to the chart.

So when adjusting the White Point manually, I tried to match the monitor (light) gray of the Samsung. That was impossible. Then I tried to match the CC image with the chart, looking at the color patches more than the gray scale. I found it very very difficult. At last, I found  that I got the best result by setting all: R,  G and B to 100%! The profile created with this new target has a color temperature of 7008 K!  And both the monitor gray and the CC images are just as before. Is this software just pretending?

Remaining questions:
-1-Should I try a different software? The Samsung was calibrated with ColorEyes. I would have to buy a new license, since my version does not run under Snow Leopard.
-2-Why is the PS image so muddy on the Eizo?

Sigh! - Hening.

I see that in the meantime you (Erik) have written a new answer, which however I do not understand: The color spaces appear in the same menu as the profiles, I can only choose one. ? Otherwise, I use ProPhoto whereever I have the choice. - I have used the synthetic target, as you may see in my "log" above.

[Tim Lookingbill]

This appears to be either a corrupt calibration profile or the numbers derived from the colorimeter's measuring of the display and written within the profile's matrix formula are inaccurate. Henning's initial description of the appearance of the X-rite hues in certain patches seems to suggest this, but it could be caused by the tone curve default suggested by Erik.

Also some versions/brands of calibration packages have colorimeter filters/software driver that may not be tuned to give an accurate measurement of one particular display inducing it to come up with a messed up set of matrix formulas that affects only some colors.

To get an idea if the matrices are messed up if on a Mac that allows the DigitalColor Meter to be set to CIELab readouts, you'll be able to test if it's a matrix issue either on the synthetic or the original shot by taking LAB readings from DCM.

I actually did this to get the source Raw image used to create the DNG profile for my Pentax K100D DSLR to actually deliver the Lab readings included with my X-rite CCchart to see if the LAB readings give an accurate representation on a display. After some extreme edits in ACR using these Lab readings to guide me, I got a very exact visual match to the original as demonstrated in the screenshots below.

Unfortunately I can't apply those LAB based preset edits derived from the Lab corrected CCchart to other Raw images without even further additional edits. Just applying the DNG profile to ACR's default was enough but keeping in mind how contrast can affect hue/saturation so critically to guide my edits visually.
[attachment=19572:CCchartA...Gprofile.jpg]
[attachment=19573:CCchartC...rLabEdit.jpg]

[ErikKaffehr]

Hi,

The Eizo has different presets for color spaces. I don't know how to set because I have a much simpler model. But it does support both sRGB and nearly Adobe RGB. Choosing sRGB mode may be an option. You obviously need to reprofile once settings are changed. The setting I talk about is on the monitor.

Regarding sRGB vs. Adobe RGB, it's said that there are advantages to both. I'm not advocating either but it could be helpful to have the same color space when comparing two monitors.

Regarding ColorEyes I'd suggest that you could try a demo license. I think it's usable for a few days. I think that the Color Eyes gives some dE values? That could be helpful.

Best regards
Erik

I have to add to the above, that the monitor gray looks *lighter* on the Eizo with 100 cd than on the Samsung with nominal 120. I assume that aging is the reason.

Now it is dark in my part of the woods, and I make new discoveries.
The Desktop set to dark gray on both monitors.
4 copies of the synthetic CC opened in PS and Preview respectively on the 2 monitors and compared to the physical CC illuminated by 1 Solux lamp, in the otherwise dark room, show this:

PS CS3:the Eizo image looks more muddy (than the chart and than the Samsung image) in most patches, in particular in patch 3/2 (green), 3/3 (red) and 3/5 (magenta). Patches 1/6 (bluish green) and 2/5 (yellow green) look pale (under-saturated).

Preview: The Samsung image is the one that comes closest to the chart, of all the 4. The Eizo edition is slightly more reddish, but not muddy at all. The reddish cast is in particular visible in patch 2/1 (orange) and 2/3 (moderate red). And also in patch 1/2 (light skin); here, it makes the Preview/Eizo image the best of the quartet.

The PS/Eizo image is way out muddy and impossible. All the other 3 are closer to each other and the chart, Preview/Samsung being the best, but all 3 would be acceptable.

I find it very difficult to judge the gray scale in comparison to the chart.

So when adjusting the White Point manually, I tried to match the monitor (light) gray of the Samsung. That was impossible. Then I tried to match the CC image with the chart, looking at the color patches more than the gray scale. I found it very very difficult. At last, I found  that I got the best result by setting all: R,  G and B to 100%! The profile created with this new target has a color temperature of 7008 K!  And both the monitor gray and the CC images are just as before. I this software just pretending?

Remaining questions:
-1-Should I try a different software? The Samsung was calibrated with ColorEyes. I would have to buy a new license, since my version does not run under Snow Leopard.
-2-Why is the PS image so muddy on the Eizo?

Sigh! - Hening.

I see that in the meantime you (Erik) have written a new answer, which however I do not understand: The color spaces appear in the same menu as the profiles, I can only choose one. ? Otherwise, I use ProPhoto whereever I have the choice. - I have used the synthetic target, as you may see in my "log" above.

[Hening Bettermann]

Hi again, and thank you all for your concern.

@ Erik: Now I see, the Color Mode button allows me to choose between a handful of color spaces including sRGB, plus Custom plus 3 Calibration sets. However, I am reluctant to use sRGB. I have spent about 2,000 EURO to get a wider gamut, and I want to use it. My main concern is not the match between the monitors, but the match between the best monitor and the chart.

The Eizo software, too, can save a validation report with DeltaE values. For the original profile, these were: (dE2000):average 0.2, max 0.6, dE White Point 0.3
For the profile I created with the new target: average 0.2, max 0.7, dE White Point 0.2.
So my manual White Point is slightly better by the numbers, but I can not see the difference. Also a White Point of 7000 K worries me.

@ tlooknbill:

> Also some versions/brands of calibration packages have colorimeter filters/software driver that may not be tuned to give an accurate measurement of one particular display inducing it to come up with a messed up set of matrix formulas that affects only some colors.

Hm, this was the Color Navigator software supplied with the monitor, and a supported puck. Of course, the puck might be damaged...??

I can not quite look through how the Lab values come in. When trying to compare them to reference values given on Bruce Lindbloom's site, I don't know which reference illuminant to choose and which RGB model. Anyway, here are the values for the 1. row of the synthetic CC, + the neutral gray patch. Both opened in PSCS3 and Preview, respectively. Can you see anything on these values? And why is the PS image so muddy?
------------
Eizo screen, Bruce Lindblooms synthetic CC target, L*a*b* readings from Apple Digital Color Meter
CC patch number - L - a - b  

image opened in PSCS3, working color space ProPhoto

1   33.019   10.316   14.719
2   62.745   13.340   18.203
3   46.722   -4.062   -21.023
4   38,099   -8,859   18,711
5   51,778   6,930      -24,883
6   67,871   -23,098   0,953
22   47,375   0,277      1,387


image opened in Preview

1   37,923   14,371   16,523
2   66,448   16,617   18,449
3   50,304   -4,383   -21,383
4   43,343   -13,555   21,438
5   55,486   9,488      -24,234
6   70,723   -31,645   0,605
22   51,427   0,273      1,344
---------------------------------------------

Hening.

[ErikKaffehr]

Hi!

The Delta E values you measure seem to be pretty good. The dE values you see are very good. I read a rule of thumb is that differences below dE 5.0 are hard to see. The difference between the LAB values in "preview" and CS3 is something I don't understand. I might look into this a bit more in a couple of days but I don't have the energy right now. It's a pity that folks like the Digital dog don't chime in (or would it be bark in? :-) on this discussion.

One thing you may consider. Eizo monitors are said to be well calibrated from factory, could you just use factory presets and profiles from Eizo and see what it looks like?

Best regards
Erik

Hi again, and thank you all for your concern.

@ Erik: Now I see, the Color Mode button allows me to choose between a handful of color spaces including sRGB, plus Custom plus 3 Calibration sets. However, I am reluctant to use sRGB. I have spent about 2,000 EURO to get a wider gamut, and I want to use it. My main concern is not the match between the monitors, but the match between the best monitor and the chart.

The Eizo software, too, can save a validation report with DeltaE values. For the original profile, these were: (dE2000):average 0.2, max 0.6, dE White Point 0.3
For the profile I created with the new target: average 0.2, max 0.7, dE White Point 0.2.
So my manual White Point is slightly better by the numbers, but I can not see the difference. Also a White Point of 7000 K worries me.

@ tlooknbill:

> Also some versions/brands of calibration packages have colorimeter filters/software driver that may not be tuned to give an accurate measurement of one particular display inducing it to come up with a messed up set of matrix formulas that affects only some colors.

Hm, this was the Color Navigator software supplied with the monitor, and a supported puck. Of course, the puck might be damaged...??

I can not quite look through how the Lab values come in. When trying to compare them to reference values given on Bruce Lindbloom's site, I don't know which reference illuminant to choose and which RGB model. Anyway, here are the values for the 1. row of the synthetic CC, + the neutral gray patch. Both opened in PSCS3 and Preview, respectively. Can you see anything on these values? And why is the PS image so muddy?
------------
Eizo screen, Bruce Lindblooms synthetic CC target, L*a*b* readings from Apple Digital Color Meter
CC patch number - L - a - b  

image opened in PSCS3, working color space ProPhoto

1   33.019   10.316   14.719
2   62.745   13.340   18.203
3   46.722   -4.062   -21.023
4   38,099   -8,859   18,711
5   51,778   6,930      -24,883
6   67,871   -23,098   0,953
22   47,375   0,277      1,387


image opened in Preview

1   37,923   14,371   16,523
2   66,448   16,617   18,449
3   50,304   -4,383   -21,383
4   43,343   -13,555   21,438
5   55,486   9,488      -24,234
6   70,723   -31,645   0,605
22   51,427   0,273      1,344
---------------------------------------------

Hening.

[Hening Bettermann]

Hi again.

Using the Eizo default profile gives the same picture, except that the images are slighlty lighter overall; the difference between the Preview and the PS rendering however is largely the same. As before, the PS image comes closer to the physical chart as I dim the Solux lamp.

It seems to me that there are 2 different problems here: One is the slightly reddish cast on the Eizo; the second is the dim look of the PS image on the Eizo, which differs significantly both from the PS look on the Samsung and the Preview look on the Eizo.

Could the reason be that it is a part of the not-full support of CS3 for Snow Leopard, that PS fails to convert from its own working space to the monitor profile, and that this makes more of a difference on the wide gamut monitor? - No, the pattern is the same if I set the PS working space to Adobe or even sRGB.

I tried ColorEyes yesterday night, but it got stuck. Maybe I'll try again tonight, but since my profile seems to be so good by the numbers, I have little hope that it will make the difference.

Good light! - Hening.

[Tim Lookingbill]

The Lab numbers you posted from both Preview and CS3 are quite a bit off compared to the numbers I get in Preview, CS3 and DigitalColor Meter on my chart which all match up on my 2004 G5 iMac in OS 10.4.11 calibrated with the original i1 Display and EyeOne Match 3.6.2, not i1 Display 2. My numbers are based on an averaging of several CC charts created by Babelcolor that can be found in this pdf:

http://www.babelcolor.com/download/RGB%20C...olorChecker.pdf

This page:
 
http://forums.imatest.com/index.php?topic=709.0

shows how other owner's CC chart appear which are quite close to what I got.

Frankly that's not the real issue anyway. The fact as Erik pointed out you are getting two different Lab readouts between Preview and CS3 which are both color managed is an indicator the profile is either corrupt or inaccurate or something isn't right with your setup.

Right now I'ld suggest you not use your Solux lit chart to troubleshoot this problem and stick with a synthetic preview made from known Lab numbers you can verify between the two apps in order to eliminate variables between your camera, DNG profiling and the display calibration in order to get this sorted out.

It might help if you could post a screenshot of this red that shows up in the CC chart on the Eizo mentioned in your initial post and a screenshot viewed with the Samsung.

[Hening Bettermann]

Hi again!

In order to make these screen shots more comparable, I made a new profile for the Eizo, trying to match the desktop light gray of the Samsung.  I set the luminance to 80 cd and came pretty close:

Digital Color Meter Lab reading of Desktop Light Gray:
Samsung    87,059 - 0 - 0
Eizo         86,981 - 0 - 0

Also, I chose L* rather than gamma 2.2. The Samsung was calibrated with ColorEyes, and ColorEyes recommends L*, and even if I have no record of it, I believe I followed the recommendation.

BTW I can see a slight green-to-red gradient in the gray on the Eizo, but Digital Color Meter can not measure it.

The Samsung profile had delta E average 0.37, max 0.68. The Eizo has average (dE2000) 0.2, max 0.5, dE White Point 0.4

The screenshots are from Bruce Lindbloom's synthetical ColorChecker target, opened in Preview and PSCS3, and displayed on either monitor.

Comparing the Preview images, I find a red cast on the Eizo, see patches 1/2, 2/1, 2/3, 2/6. Much grosser however is the dull appearance of the PS Eizo image compared to the other 3.

Bruce Lindbloom's reference values for his synthetic CC for Illuminant D65,
rounded down to 3 post-comma digits:
row 1 and patch 22, middle gray;
37,257   12,766   14,869
65,959   13,500   17,224
50,585   -1,609   -21,263
43,188   0,433      21,970
55,662   11,1813   -25,010
71,232   -31,861   1,507
51,244   -0,076   0,687

Lab readings of
Preview on Samsung, row 1 and patch 22
37,841   13,996   16,410
65,943   15,648   18,121
50,205   -4,762   -21,539
43,119   -13,723   21,371
55,424   9,430      -24,691
70,771   -31,648   0,043
51,415   0,207      1,195

PS on Samsung
36,945   13,621   15,699
65,943   15,648   18,121
49,673   -5,395   -21,770
42,382   -13,555   20,383
54,841   8,711   -25,008
70,343   -31,891   -0,016
50,921   -0,199   1,039

Preview on Eizo
37,856   14,223   16,383
66,566   16,609   18,820
50,289   -4,430   -21,598
43,326   -13,895   21,723
55,375   9,785   -24,852
70,835   -31,645   0,414
51,431   0,273   1,273

PS on Eizo
35,450   10,102   14,371
64,199   12,328   17,441
49,044   -2,684   -20,504
40,923   -9,305   18,969
54,014   6,363   -23,422
69,551   -21,277   0,984
49,514   0,277   1,277

[attachment=19650:CC_Previ...ung.tiff.jpg]
[attachment=19648:CC_PS_Samsung.tiff.jpg]
[attachment=19649:CC_Previ...izo.tiff.jpg]
[attachment=19647:CC_PS_Eizo.tiff.jpg]
----------------------------------------------------------
Have a nice week-end! - Hening.

[Tim Lookingbill]

I extracted the embedded profile of one of your screenshot images and got a large 720K LUT based file named "193P_dig_max-native.icc". The screenshot of Colorsync Utility shown below shows the "vcgt" tag curves applied to the video card to correct nonlinearity and color cast down the entire grayscale of the native response of your display. These RGB vcgt curves are having to do way too much work for an 8 bit video system.

Also the gamut shown on the right is much smaller than sRGB and I know Eizo/Samsung? wouldn't put out a high quality display with this size gamut. I can only surmise your profile is toast.

My only suggestion is for you to recalibrate and profile by first loading a 21 step grayramp spanning the entire display and set brightness and contrast with the display's buttons to get this grayramp to look as perfect as possilble with the sRGB2.1 profile loaded as your system profile.

Then go through the calibration process by setting gamma 2.2, white point brightness to match the Samsung's and build a simple matrix profile to where the final profile is way under 720K, more like around 10K and use native color temp and just measure the response and build the profile.

That table based profile you made is messed up. You might want to try using some other calibration software that works with your colorimeter.
[attachment=19692:BadEizoProfileLL.jpg]

Strangely everyone of those screenshots you posted has the Samsung 193 profile embedded. Could it be that your Eizo is using the Samsung profile?

[ddk]

Its normal, I had the same issue of the reddish cast with my CG241W when I first bought it, and went crazy to solve the problem, just like you, but it went away within 2-3 weeks of use. I think it has with the burn-in period for a monitor. FYI I also use the ColorEyes software and the DPT94. I get identical results calibrating with either the CE or Color Navigator software.

[Jonathan Wienke]

Strangely everyone of those screenshots you posted has the Samsung 193 profile embedded. Could it be that your Eizo is using the Samsung profile?

If you're running dual monitors, it's best to have separate video card for each monitor, otherwise you can run into a situation where one profile is always being used for both monitors, regardless of settings, profiling methods, etc. That may well be what is happening here. The messed-up non-linear table profile is a strong indication that something along those lines is happening; all of the Eizos I've profiled have had very linear TRC curves.

[ddk]

If you're running dual monitors, it's best to have separate video card for each monitor, otherwise you can run into a situation where one profile is always being used for both monitors, regardless of settings, profiling methods, etc. That may well be what is happening here. The messed-up non-linear table profile is a strong indication that something along those lines is happening; all of the Eizos I've profiled have had very linear TRC curves.

Not really the case with Macs!