Monitor Calibration--Again

[rdenney]

I know I've read about 200 articles on this, some very authoritative, some not, and some that sound one way and really are the other. But nothing is really giving me the insight I need to troubleshoot a problem I seem to be having.

I currently have a Dell U2410 as my primary Photoshop display. It has an IPS panel, and was not cheap (for a consumer monitor) even as a refurb. It provides a wide gamut, though--wider than many other displays.

Here's a review that is typical of what led me to choose this monitor:

http://www.flatpanelshd.com/review.php?subaction=showfull&id=1259151406

I cannot afford the dedicated high-end displays.

The other monitor is a typical HP 24" consumer display (one of the two that came with my computer) that uses a TN panel. Even though I have calibrated it, I don't use it for color-managed work because it's color shifts with viewing angle.

My color management system is a Gretag-Macbeth Eye-One Display 2, using X-Rite's most current software. (X-Rite acquired this range of products from Gretag-Macbeth, or they bought them out altogether, or they changed names--whatever). I bought this system maybe five or six years ago and the Dell is the third monitor I've used it for. (It replaced a previous puck and software package that used a serial interface on the puck which new computers do not support happily.)

But I'm losing faith in this system. I make my photographs look good on the Dell display, from within Photoshop, working in ProPhoto color space, which has a wide gamut. When I'm happy, I convert it to the sRGB colorspace as part of making a downscaled image for web display. The image in sRGB looks identical to the image in ProPhoto after the conversion, as should be expected in a color-managed workflow.

I do not normally expect my web-displayed photos to have matched color on unmanaged displays, or poorly calibrated displays, used by viewers. But I'm finding that the images take on some strange color artifacts, particularly color crossovers tending red or magenta in the shadows, or magenta bands through skies. This is the case for all films scanned using either my Nikon 8000 film scanner, or my Epson flatbed. Other people's photos don't have these artifacts, so I'm beginning to think that my calibration is off on my Photoshop workflow. I've ignored it until now, figuring that displays out in the wild represent too much of a moving target, but I see a trend and am trying to figure out why. This is despite my observation that the Dell monitor has extended red gamut that causes saturated reds to really be accentuated on the display.

I used the X-Rite software in "detailed" mode to calibrate the monitor and then to create profiles, and I referenced the Dell monitor profile in Photoshop and in Vuescan. I have different profiles for the two monitors. My video card has two LUTs, one for each monitor, and the Dell monitor uses the DVI interface.

I'm using Photoshop CS5, but I had the same issues with CS4.

My working theory at this point is that the monitor has extended gamut that goes beyond the sensitivity range of the Eye-One Display 2 puck. Even with a separate profile for each display, the HP display is noticeably more magenta than the Dell display. Perhaps that display is actually more accurate, or at least more representative of the displays most people have that come with a generalized sRGB calibration.

Am I missing some point? Is there a known limitation on the range of older colorimeters when used with current wide-gamut displays?

Search on my posts in the images threads for recent examples of what I'm talking about. There was a comment made by Bryan in the "safe haven for tiny formats" thread about the redness of the shadows in the images I posted there. I don't see it on my monitor at home. I do see it on my display here at work, though it's more magenta than red. Here is the link to my post of the first of that batch of images:

http://www.largeformatphotography.info/forum/showthread.php?43423-safe-haven-for-tiny-formats&p=948610&viewfull=1#post948610

Rick "never really satisfied with color management" Denney

[bdkphoto]

I've had the same issues, with color and toned Black & White and using a high end monitor with hardware/software calibration. With color images , especially portraits or images with red/yellow there is a pronounced red build when saved for the web, and with the toned B&W there was banding in the sky/light areas.

Best I can figure, after a lot of consulting with colleagues and web designers --is it is a compression issue with jpg's. I was never able to resolve the "toned" print banding issues to get that to look good, so I went greyscale to solve that problem, and with color work, I simply dial the reds way down before posting and/or test the image online and color correct from there. I know several others who have created an photoshop action to color correct for web a a batch process, but I just tweak the images as needed.

[Doug Fisher]

>> Is there a known limitation on the range of older colorimeters when used with current wide-gamut displays?<<

It is true that some older calibration pucks don't have filtering that is well suited for wide gamut monitors. You might call X-rite and ask them about your particular series/release of the i1 Display 2 in order to confirm things. The should be able to tell you based on the serial and/or revision number of your puck. I have found their support people to be better than average and quite helpful.

Doug

[rdenney]

>> Is there a known limitation on the range of older colorimeters when used with current wide-gamut displays?<<

It is true that some older calibration pucks don't have filtering that is well suited for wide gamut monitors. You might call X-rite and ask them about your particular series/release of the i1 Display 2 in order to confirm things. The should be able to tell you based on the serial and/or revision number of your puck. I have found their support people to be better than average and quite helpful.

Doug

I'll do that. I've become so accustomed to getting no support on anything, and it angers me so much when I get ignored, that I don't even try. And I did buy this quite a long time ago. But I'll give them a call. That would at least confirm or deny that one theory.

Rick "appreciative" Denney

[rdenney]

After some more exploration and some discussion, it's become clear that there are indeed limitations in gamut with the i1display2 device. That devices was originally developed by Gretag-Macbeth, and the one I have came in a Gretag box. Some OEM versions of it were filtered for wider (or different) gamuts, but mine wasn't.

So, it looks as though I'll be investing in a new monitor calibration and profiling system soon. I'm attracted to the i1Display Pro, which is an X-Rite design and which definitely accommodates wide-gamut displays. The software and device combination is also apparently MUCH faster than the old iMatch/i1Display2 combination (and the Colormunki combination, which isn't much cheaper in any case, or the Spyder3). And the prices have come down quite a bit.

I thought I knew about color management, but the rapid transition to wide-gamut displays, optimized for gaming or watching TV and movies rather than viewing photos, has changed everything. I'm really going to have to start over.

At least the X-Rite system is a lot cheaper than yet another new display.

Rick "also looking into Colorneg" Denney

[Kirk Gittings]

FWIW, I have a Eye-One Display One and a newish Lacie and am thinking of buying a larger new one. I thought the E1 was working fine but I knew it was old technology. So I borrowed a newer 2 and found that mine was off noticeably. I don't think there is anyway to save it and just ordered the ColorMunki Display as there is a rebate on it right now.

I only print B&W and have learned that soft proofing is nowhere near perfect anyway so color calibration for me for printing is not a big issue. So calibration for me is mainly for delivering accurate files to my clients. Unfortunately having visited the offices of many of my clients, I know that few have well calibrated equipment and by far most are pathetic. Sometimes this seems like a waste but at least I am doing my part.

[bob carnie]

Rick
depending on the image, you will see more colors editing in Pro Photo and when you convert to SRGB you may see conflicting colours due to PS deciding for you how a colour will show itself in a smaller space. Perceptial and RC both move colours differently and I am sure the dude from Chromix that I see on this site can enlighten you .


I'll do that. I've become so accustomed to getting no support on anything, and it angers me so much when I get ignored, that I don't even try. And I did buy this quite a long time ago. But I'll give them a call. That would at least confirm or deny that one theory.

Rick "appreciative" Denney

[pherold]

Rick,
X-Rite makes available an "i1Diagnostic" program that will walk your i1D2 through its paces and give you a report on whether it passes or fails. It's sometimes rather generous, in that it might give you a pass when there's still something wrong, but it's worth a shot. You can download it here:
http://www.xrite.com/product_overview.aspx?ID=766&Action=support&SoftwareID=506

If you are going to replace it, the i1Display Pro you're considering is a great device, and as you've noticed there's some sort of rebate going on right now.

It's less likely, but I want to mention it anyway, that what you're seeing in the shadows is color banding. When a wide gamut monitor is driven by a typical 8-bit video card, your system only has 256 steps in which to define all the colors. The wider the gamut of a monitor, the more "stretched out" the resolution of the color would be, and it is likely that some colors will band together instead of showing perfectly smooth gradients.

[Kirk Gittings]

When a wide gamut monitor is driven by a typical 8-bit video card, your system only has 256 steps in which to define all the colors. The wider the gamut of a monitor, the more "stretched out" the resolution of the color would be, and it is likely that some colors will band together instead of showing perfectly smooth gradients.

Interesting. What video card is best for wide gamut monitors? The only reference to "bits" I am seeing on my video card specs is "64 bit memory interface".

[rdenney]

Rick,
X-Rite makes available an "i1Diagnostic" program that will walk your i1D2 through its paces and give you a report on whether it passes or fails. It's sometimes rather generous, in that it might give you a pass when there's still something wrong, but it's worth a shot. You can download it here:
http://www.xrite.com/product_overview.aspx?ID=766&Action=support&SoftwareID=506

If you are going to replace it, the i1Display Pro you're considering is a great device, and as you've noticed there's some sort of rebate going on right now.

It's less likely, but I want to mention it anyway, that what you're seeing in the shadows is color banding. When a wide gamut monitor is driven by a typical 8-bit video card, your system only has 256 steps in which to define all the colors. The wider the gamut of a monitor, the more "stretched out" the resolution of the color would be, and it is likely that some colors will band together instead of showing perfectly smooth gradients.

Thanks. Yes, I do see some banding on that monitor, but not usually when I convert to sRGB. I mainly use ProPhoto because it gives the scanner software a bigger target to hit, and I get less clipping of wide-ranging negatives and transparencies. The banding doesn't bother me if I know it's just a display artifact in that particular color management and video-card display chain. The point of careful calibration, of course, is to avoid the amount of color response warping that the video card LUT has to perform. But if the calibration sensor isn't seeing the color range being displayed, it's a waste of time.

I do get color crossovers, but that's because my current profile is wrong by a greater margin in the shadows than in midtones and highlights.

I just took one of the images I'd posted in the other thread, and viewed it on a typical newish factory sRGB-calibrated monitor. The magenta was there in abundance. I corrected the image to remove it (see attached), and that required a very significant color balance movement (like 40% in the green-magenta dimension) to achieve the same look I get on my "calibrated and profiled" Dell monitor. I've linked the image as I've put it on my web site, and then attached another low-res downsample from the original using a different process. Instead of converting to sRGB in the "save for web" dialog, I converted the PSD file first to sRGB, displayed it on the unprofiled monitor, and made those big corrections. I then "saved for web" and this time embedded the sRGB profile. (Both will probably be resized by the forum software, but I suspect the difference will still be visible.)

82905

82906

Edit: Seeing them both on my office laptop, I'm wondering if I didn't go a bit too far. Gotta get a monitor profiled so that it will display some ground truth to restore my confidence.

Rick "time for a hardware upgrade" Denney

[rdenney]

Interesting. What video card is best for wide gamut monitors? The only reference to "bits" I am seeing on my video card specs is "64 bit memory interface".

I'm not sure any of them are really displaying only 8-bit color. I surely do often get smooth gradients. But if the monitor is really out of whack, and the profile is having to make big, big moves to get it back in whack, it could cause some posterization. I've seen it, but mostly I think it's more a scanner artifact (with a flatbed) than a monitor thing. The video card in my computer is not designed to be quick, but it is one of the better Radeon display adapters with native dual monitors and a separate LUT for each display. I am quite sure that is NOT my problem, at least for monitors that can be calibrated to be pretty close before profiling. But that calibration uses the same sensor that the profiler uses, and if the sensor can't see the display's gamut, it won't control the extremes. I don't think their software sets gray to be neutral, but rather looks at pure red, green, and blue, plus a sampling of other colors. If the pure red, for example, is significantly brighter than the sensor can see with only red light, it won't adjust the red enough and the result will be a reddish cast on the neutral colors. The profiler will try to correct the middle values, but it can't fix everything.

Rick "who didn't know the trouble being caused by buying a wide-gamut monitor, and who didn't even know it was a wide-gamut monitor when it was bought" Denney

[pherold]

To answer your question about 8-bit video cards I risk giving you more information than you want to know....!
There's been a movement afoot for a couple of years to try to get true 10-bit color from Photoshop to be actually viewable on a monitor. This requires a pipeline that has 10-bit support all the way from Photoshop to the operating system to the video card to the monitor. The monitor for example, would have to be one of the high end NEC's or Eizo's that support this. This article gives a lot of good information - like some of the common confusion about this - what OS's work - what video cards work...
http://www.imagescience.com.au/kb/questions/152/10+Bit+Output+Support

One of the reasons why NEC's and Eizo's are expensive is because they have their own high-bit graphics card built in to the display. This way these wide gamut monitors can have at least 1024 steps with which to define color (instead of 256). This helps, because the monitor can do the "heavy lifting" of adjusting the color and allow the computer's graphics card (which is almost always 8-bit) to not have to do much adjusting. This is why a Dell will tend to get some banding - especially if as Rick says, it has to make big moves which will limit the number of definable points to even less than 256.

[rdenney]

One of the reasons why NEC's and Eizo's are expensive is because they have their own high-bit graphics card built in to the display. This way these wide gamut monitors can have at least 1024 steps with which to define color (instead of 256). This helps, because the monitor can do the "heavy lifting" of adjusting the color and allow the computer's graphics card (which is almost always 8-bit) to not have to do much adjusting. This is why a Dell will tend to get some banding - especially if as Rick says, it has to make big moves which will limit the number of definable points to even less than 256.

So, you could say that the calibration step in software like iMatch is taking the place of all that LUT management inside those high-end monitors. We just achieve it by manipulating the brightness, contrast, and color controls on the monitor. I suppose the question would be whether those controls are software-driven and subject to their own limited data space, which I suspect some are. The objective of the exercise (whether done using the buttons or using a sophisticated monitor-base calibration system) is to minimize the color conversions that have to be made by the computer's video card software. We are working around a data-width bottleneck, same as we do when scanning.

And it's apparent, based on what you've provided, that my response to Kirk was related to whether 8 bits were being well used vs. whether they were being poorly used.

I don't think I can afford the 10-bit setup, even though I have some of the pieces in place. I guess when we target our images for web display, we ought to worry about it, though few on the viewer side will be able to see it anyway. When we target our images for printing, however, it seems to me that prints have less gamut than monitors, and that a bit of banding on the monitor is no issue if we know it won't be on the print.

Most of my output since upgrading to this monitor has been related to Blurb books and to making prints. I'm uploading scanned photos to my web page for the first time in a couple of years, after revamping my web page to make it easier to do, and now I'm running into all these issues.

Rick "appreciating the lessons" Denney

[Kirk Gittings]

Ok my info above was full of s__t.

I actually have an Eye One Display 2 and now I am confused about what I borrowed to check mine out. Anyway my ColorMunki Display came yesterday and I ran it a few minutes ago-truely dramatic difference from the profile created by the borrowed calibrator and clearly more neutral. That is hard to grasp unless you can look at the rendering of two profiles side by side. This program and hardware is simply more sophisticated and in sync with this generation of monitors and IMO well worth the $.

[rdenney]

Ok my info above was full of s__t.

I actually have an Eye One Display 2 and now I am confused about what I borrowed to check mine out. Anyway my ColorMunki Display came yesterday and I ran it a few minutes ago-truely dramatic difference from the profile created by the borrowed calibrator and clearly more neutral. That is hard to grasp unless you can look at the rendering of two profiles side by side. This program and hardware is simply more sophisticated and in sync with this generation of monitors and IMO well worth the $.

Okay, that confirms it for me. Time to whip out the plastic...

...again.

Rick "for whom this will be his third color management system" Denney