Ad
Ad
Ad
Pages: « 1 [2]   Bottom of Page
Print
Author Topic: Overall reddish cast on Eizo CG243W  (Read 10188 times)
Jonathan Wienke
Sr. Member
****
Offline Offline

Posts: 5759



WWW
« Reply #20 on: January 23, 2010, 09:45:45 PM »
ReplyReply

Quote from: ddk
Not really the case with Macs!

PC/Mac doesn't matter. If your video card has dual outputs but only a single LUT, (which many do), then that LUT is used for both monitors regardless of any other settings--it's a hardware limitation. Dual cards guarantees that each monitor has its own LUT and can be separately profiled.
Logged

ddk
Sr. Member
****
Offline Offline

Posts: 274


WWW
« Reply #21 on: January 24, 2010, 12:01:06 AM »
ReplyReply

Quote from: Jonathan Wienke
PC/Mac doesn't matter. If your video card has dual outputs but only a single LUT, (which many do), then that LUT is used for both monitors regardless of any other settings--it's a hardware limitation. Dual cards guarantees that each monitor has its own LUT and can be separately profiled.

Maybe, but I've been using dual monitors on different macs with single cards for years and never had any issues.
Logged

david
-----------------------
www.pbase.com/ddk
Jonathan Wienke
Sr. Member
****
Offline Offline

Posts: 5759



WWW
« Reply #22 on: January 24, 2010, 06:54:00 AM »
ReplyReply

Quote from: ddk
Maybe, but I've been using dual monitors on different macs with single cards for years and never had any issues.

You've simply been lucky enough to have used dual-LUT cards.
Logged

Tim Lookingbill
Sr. Member
****
Offline Offline

Posts: 1201



WWW
« Reply #23 on: January 24, 2010, 08:03:10 AM »
ReplyReply

Quote from: Jonathan Wienke
PC/Mac doesn't matter. If your video card has dual outputs but only a single LUT, (which many do), then that LUT is used for both monitors regardless of any other settings--it's a hardware limitation. Dual cards guarantees that each monitor has its own LUT and can be separately profiled.

This 2005 Colorsync Mailing List discussion reply by Peter Karp indicates you're sort of right:

http://lists.apple.com/archives/colorsync-...b/msg00062.html

But frankly there's a lot of confusing terminologies in discussions like these to describe internal components of a display profile that I wish would be made more clear. For instance the term "LUTs" often get interchanged with "VCGT", "TRC's", "Table based as opposed to matrix based version of a profile (both have different file sizes)" and "Gamma Curves". That Colorsync discussion doesn't make it any more clear what's really getting downloaded to the video card either.

From my understanding a Mac profile has a "VCGT"-(Video Card Gamma Tag) which contains ONLY the separate RGB curves that work out all the little kinks in a display's measured response compared to a forced target response. If the display's response is already close to target (i.e. 6500K, 2.2 gamma, XXX cd/m2 luminance and neutral in all 255 gray levels) which most displays already are these VCGT RGB curves will almost resemble a straight line. These curves are what make each display change slightly in color cast and contrast globally outside of color managed apps whenever you click on different profiles in the Display Preference panel.

I can confirm on my 2004 G5 iMac with an external Dell 2209WA each display's corresponding "VCGT" curves within their given profiles are occupying the same video card. To confirm this I have both the sRGB standard profile chosen for both displays which doesn't have a "VCGT" clearing the correction from the video card and each display show differences in color cast and contrast but only slightly. VCGT is not a TRC curve which is just a simple 2.2 gamma curve used only by CM apps in case an image is tagged with a working space containing a gamma curve other than 2.2 like the 1.8 gamma of ProPhotoRGB.

What I'm not clear on is the term "Table" based profile as opposed to "Matrix" based. From examining the Samsung embedded profile's table section on this in Colorsync Utility it appears to be a more complicated set of individual matrix transform formula's.

The issue here is figuring out if Hening's red cast in select colors is caused by the VCGT or bad matrices written into the profile causing color managed apps to screw with hue/sat or a combination of both.

He first needs to confirm how each profile is being assigned to each display on his setup which he hasn't done.
« Last Edit: January 24, 2010, 08:12:32 AM by tlooknbill » Logged
Hening Bettermann
Sr. Member
****
Offline Offline

Posts: 570


WWW
« Reply #24 on: January 24, 2010, 10:01:51 AM »
ReplyReply

Thank you all for your concern!

ddk, what you write is a great consolation! I have so far avoided to use the Eizo for trivia (=non-photography) to keep  the mileage down on this precious instrument, but I'll change that now...

Concerning the dual profile and 1 videocard: When I open the display prefs, the profile i chose is selected for either monitor. I can flip through all the (canned) profiles on either monitor and see corresponding changes on that monitor only. E.g if I choose the Samsung profile for the Eizo, the Desktop light gray on the Eizo changes towards greenish. Wouldn't this indicate that the card is capable of handling 2 profile output?
(It's a NVIDIA GeForce GT 120).

Then why is the Samsung profile embedded in all my screenshots ? Could it have something to do with the fact that they are screenshots, i.e. have gone through Grab (the screenshot utility on the Mac)? Bruce Lindblooms synthetical CC, which is a .tif and the basis of all the screenshots, has Lab embedded as the color space, according to the Mac's Get Info command. Could it be that Grab embeds the profile of the "primary"  monitor, i.e. the one that holds the menu bar? It is very likely that this was the Samsung when I took the shots.

Yes exactly! When I open the synthetic CC in Preview and save the image, Get Info still shows Lab as the color space. When I take a screenshot of the Preview image with Grab on the Eizo, with the Eizo holding the menu bar, and save the image, Get Info shows "Color space: RGB, Color profile: CG243W" (The factory profile, which is currently selected). This does not change if I move the open image to the Samsung when I take the shot. - One mystery solved!

If it is just the Samsung profile which is damaged, that's no problem. IF the Eizo really looses the red cast before all too long, I don't care for color accuracy on the Samsung.
In the meantime, I may try to make new profiles for the Eizo with the native White Point, which seems to give the most neutral white.

Concerning profiling software: The downloaded trial of Color Eyes does not run on my Mac, it stops with error #1 or -1. The support has advised deleting the prefs and repairing permissions, which I have done to no avail. Anyway, it seems that is not the Eizo profiling software which is the problem. Also I am befriending Argyll CMS despite the Unix surface.

What is left?

1- I am shocked to read that sRGB is my System color space. I have never chosen that. Where is this shown, and how can I change it?

2-   What is wrong with muddy Photoshop?

3-   Even if my video card can handle 2 profiles: In the end, I want a high end LUT profile for the Eizo. So if such profiles are too large for the NVIDIA, I am willing to invest in a card that can handle one. Could you recommend a card?

I wrote the above before I had seen post #24 or studied the link to the ColorSync discussion, which I probably will have difficulties to follow, so I better kick this out now.

Best regards - Hening.
« Last Edit: January 24, 2010, 10:11:40 AM by Hening Bettermann » Logged

Tim Lookingbill
Sr. Member
****
Offline Offline

Posts: 1201



WWW
« Reply #25 on: January 24, 2010, 07:04:35 PM »
ReplyReply

From what you indicated everything checks out with the way the system is handling two profiles on one video card. You just didn't supply us with the Eizo screenshots containing the custom profile that makes the preview of the synthetic CC reddish. The previous screenshots with the embedded Samsung looked and measured OK.

The only thing left is to recalibrate and build a profile that is a simple matrix (not a large LUT/table based) profile to see if that changes anything.

If your current video card is a third party upgrade installed version as opposed to the one that came with your system when you bought it, you probably should update the driver if there's one available.

Quote
1- I am shocked to read that sRGB is my System color space. I have never chosen that. Where is this shown, and how can I change it?

Don't know where you got that idea. You may have got this confused with my mentioning before you recalibrate, to load the sRGB2.1 profile to get an idea how close your display's native response before calibration to the ideal target of 6500K, 2.2 gamma with no multi-colored tints throughout a grayramp really looks. Loading the sRGB2.1(not Apple's sRGB) profile will clear the "Gamma Tables"/"VCGT"/"Calibration Curves" (see how confusing-there's already three names for this one profile component) which is what your calibration software is going to do anyway before it measures the response of your display.

It's not necessary to load the sRGB2.1 profile. Sorry for confusing you.

Quote
3-   Even if my video card can handle 2 profiles: In the end, I want a high end LUT profile for the Eizo. So if such profiles are too large for the NVIDIA, I am willing to invest in a card that can handle one. Could you recommend a card?

This isn't a question about whether your video card can handle a large LUT profile. Your video card is just fine.

To make this simple just examine your custom high end LUT Eizo profile in Colorsync Utility by double clicking on it which will bring up a separate panel. Scroll down and click on the 'vcgt' line entry and look at each RGB "Gamma Table" curve. It should be a straight line or close to it. It shouldn't look like the Samsung's as shown in my previous post.
« Last Edit: January 24, 2010, 07:05:04 PM by tlooknbill » Logged
Hening Bettermann
Sr. Member
****
Offline Offline

Posts: 570


WWW
« Reply #26 on: January 26, 2010, 11:20:02 AM »
ReplyReply

Bad news.

>You just didn't supply us with the Eizo screenshots containing the custom profile that makes the preview of the synthetic CC reddish.

I had meant to do this, but can now see, that #3 and 4 were useless, since they had the Samsung profile embedded.

So here are new screenshots with different profiles. All have the correct profile embedded, i.e. the shots of the Eizo were taken when the Eizo was the monitor holding the menu bar, and accordingly on the Samsung (it requires a restart of the system to make Grab recognize the shift and embed the correct profile).

All LUT profiles. Color Navigator does not seem to have an option for a matrix profile, and after some hours of fiddling with Argyll, I take a break from that. In the end, I want a LUT profile anyway.

In the vcgt panel, all profiles have 100% straight lines in all 3 colors.

Here goes:
1- Preview on Eizo, gamma 2.2, WP target 6500K
2- Preview on Eizo, L*, WP target 6500K
3- Preview on Eizo, L* with the native WP of the monitor. (~7000K)


4- Photoshop on Eizo, gamma 2.2, WP target 6500 K
5- Photoshop on Eizo, L*, WP target 6500K
6- Photoshop on Eizo, L*, with the native WP of the monitor. (~7000K)


7- Preview on Samsung, with the SyncMaster default profile.
8- Photoshop on Samsung, with the SyncMaster default profile.
9- Preview on Eizo, default profile
10-Photoshop on Eizo, default profile

Conclusion:

My first observation, that led me to start the topic, was invalid because my comparison was based on a damaged profile on the Samsung side.
Now, with the SyncMaster default profile, compared to all Eizo profiles, my picture when comparing the screenshots is this:
All look good (that is: close to the physical chart in Solux light) overall when displayed on the Samsung. Photoshop has the edge in that it avoids the oversaturation of red in patch 2/3. Patch 2/1 is very slightly reddish in all. All fail to render the purple in patch 2/2.
When displayed on the Eizo, all look so muddy overall that the rest is not worth while.

It makes no difference if I choose a profile with 100 rather than 80 cd/m^2 max luminance (no screenshots of this). Even with the default SyncMaster profile on the Samsung, there is still a good brightness match of Desktop light gray, both visually and by Lab numbers. (Samsung: 87.045/0/0, Eizo: 86.981/0/0). The Eizo looks more reddish, but the Digital Color Meter obviously can not see it.

Do I have to return the screen??

:-(  

Good light! - Hening.

[attachment=19749:1_Ei_Pr_...00K.tiff.jpg]
[attachment=19750:2_Ei_Pr_...00K.tiff.jpg]
[attachment=19751:3_Ei_Pr_...00K.tiff.jpg]
[attachment=19752:4_Ei_PS_...00K.tiff.jpg]
[attachment=19753:5_Ei_PS_...00K.tiff.jpg]
[attachment=19754:6_Ei_PS_...00K.tiff.jpg]
[attachment=19755:7_Sa_Pr_...ult.tiff.jpg]
[attachment=19756:8_Sa_PS_...ult.tiff.jpg]
[attachment=19757:9_Ei_Pr_...ult.tiff.jpg]
[attachment=19758:10_Ei_PS...ult.tiff.jpg]
---------------------------------------------------------------------------
Logged

Tim Lookingbill
Sr. Member
****
Offline Offline

Posts: 1201



WWW
« Reply #27 on: January 26, 2010, 05:33:48 PM »
ReplyReply

All those screenshots look identical viewed in Preview on my system. They all look correct. DigtitalColor Meter LAB readouts are within specs and there is no red cast. However, for some reason they won't open in Photoshop because I get an error warning stating "Can't complete your request because of an unknown or invalid JPEG marker type is found".

Also both custom profiles embedded in those screenshots ARE simple matrix based, but with NULL vcgt meaning this type of curve is not correcting for any global visual inaccuracies in your display's contrast and neutrality. You're seeing your display's raw native response.

It doesn't matter anyway because this is just getting too complicated troubleshooting over the web.

The important point you made is you say your Eizo Photoshop previews of the synthetic CC is muddy but not in Preview which suggests Photoshop is not using the custom Eizo profile. Go into Photoshop's Color Settings scroll up within the RGB Working Space menu and check to see if "Monitor RGB..." shows the name of your custom profile.

If it doesn't, then you have other unknown software issues you have to deal with which are too difficult to solve here.

Also don't expect a DSLR capture of the CC chart to look exact shot under Solux lights without considerable editing because this light's halogen underpinnings throws off camera's WB sensors. Stick with the synthetic CC chart to make critical assessments of your display calibration/profile.

Sorry I can't help you any further.
« Last Edit: January 26, 2010, 05:34:15 PM by tlooknbill » Logged
Hening Bettermann
Sr. Member
****
Offline Offline

Posts: 570


WWW
« Reply #28 on: January 27, 2010, 12:09:58 PM »
ReplyReply

Thank you for your extensive reply!

> [the screenshots] won't open in Photoshop because I get an error warning stating "Can't complete your request because of an unknown or invalid JPEG marker type is found".

They will open if you remove the fake .jpeg extension so the .tiff is revealed. (The LL software requires jpegs, but *converting* to jpeg even with the lowest quality setting will *IN*crease file size from ~70 to ~100 kB when done in PS, which I knew from an earlier instance. First now, when I check this, I see that converting in Preview will *de*crease the file size to ~30 kB. Also, in this case, converting with one of the involved apps would have added yet another layer of confusion which I could not look through).

>Also both custom profiles embedded in those screenshots ARE simple matrix based, but with NULL vcgt meaning this type of curve is not correcting for any global visual inaccuracies in your display's contrast and neutrality. You're seeing your display's raw native response.

Does this mean these profiles do nothing at all?? Would mean that the CN software is just faking? I don't understand. I think you wrote earlier that these curves "should be a straight line or close to it." ??

>The important point you made is you say your Eizo Photoshop previews of the synthetic CC is muddy but not in Preview which suggests Photoshop is not using the custom Eizo profile.

No what I wrote was

>All look good [...] overall when displayed on the Samsung. [...] When displayed on the Eizo, all look so muddy overall that the rest is not worth while.

Since it is an overall muddy look even more than the red cast in some patches, I made a new profile with a max luminance of 120 cd, but it makes now visual difference whatsoever.

>Go into Photoshop's Color Settings scroll up within the RGB Working Space menu and check to see if "Monitor RGB..." shows the name of your custom profile.

Thank you for this tip! That menu shows Generic RGB profile, + the profile of the monitor which holds the menu bar. So this seems ok.

> Also don't expect a DSLR capture of the CC chart to look exact shot under Solux lights without considerable editing because this light's halogen underpinnings throws off camera's WB sensors. Stick with the synthetic CC chart to make critical assessments of your display calibration/profile.

You may misunderstand me. I am not comparing the screen display of the synthetic CC to screen displays of camera images of the CC, but to the physical chart on my desktop lit by a Solux lamp. When comparing 2 monitors, I think I need a "third opinion".

Is there any better light source for a reasonable price? I chose the Solux according to Bruce Frasers book on Color Management.

> Sorry I can't help you any further.

Thank you very much for your help so far!

-----------------------------------------------------------------------
New discovery:
When I drag my monitor closer to the balcony door and compare the screen image to the CC chart viewed in diffuse daylight under an overcast sky, then the Eizo image is close to the chart! In this light, the chart *does* look muddy. The Samsung looks clearer and thus prettier, AND this is closer to the chart lit by a Solux lamp. The Eizo's native white point of 7 000 K (which I have chosen for the actual profile) is probably close to todays overcast daylight. But what if I view the chart in clear sunshine in the late afternoon? (My typical shooting light). The other way round: What does the synthetical chart look like when I make a monitor profile with a target white point of 5 000 K and compare that to the physical chart? -> Just as muddy.

So what am I to believe??  








Logged

Jonathan Wienke
Sr. Member
****
Offline Offline

Posts: 5759



WWW
« Reply #29 on: January 27, 2010, 03:18:46 PM »
ReplyReply

Quote from: Hening Bettermann
So what am I to believe??  

That the "color cast" of the monitor is dependent on the ambient light in which the monitor is viewed. Just like ambient light is important when trying to match the monitor to the print. Or the Color Checker.
Logged

Mikko S
Newbie
*
Offline Offline

Posts: 5


« Reply #30 on: March 22, 2010, 04:31:53 AM »
ReplyReply

Hening

Did you ever got your Eizo calibrated correctly? I am asking this as I was planning to buy the same monitor but not willing to spend 2000 Euros just to face the same problems as with 500 Euros HP monitor. From my experience the the wide gamut ISP panels seem to have this problem. I recently had exactly the same problem with HP 2475 with ISP panel (attached to 24" iMac as a secondary monitor), far too red overall cast, I tried all possible calibrators (I have them all) and spent weeks with it. Finally I tried ColorEeys Display Pro calibration software and that finally solved the problem.
Logged
Hening Bettermann
Sr. Member
****
Offline Offline

Posts: 570


WWW
« Reply #31 on: March 22, 2010, 08:48:14 PM »
ReplyReply

(accidental dup cleared, HB)
« Last Edit: March 23, 2010, 08:50:34 PM by Hening Bettermann » Logged

Hening Bettermann
Sr. Member
****
Offline Offline

Posts: 570


WWW
« Reply #32 on: March 22, 2010, 08:55:38 PM »
ReplyReply

Hi Mikko

I thought I had answered your PM, but now I can not find it. I'll try again here. -

I am now perfectly happy with the Eizo, and I hardly remember the problems I had. I think it all was a failure on my side, comparing to a Samsung whose profile turned out to be corrupted. Also, part of it may have been as ddk wrote, the reddish cast went away by itself after a little while. - The profiling software that comes with the monitor works just fine.

Good light! - Hening.
Logged

Pages: « 1 [2]   Top of Page
Print
Jump to:  

Ad
Ad
Ad