Color Inconsistency Between Two Calibrated Monitors

March 8, 2009 6:02 AM

Why do two properly calibrated monitors look very different? I have a dual monitor setup -- a Samsung 216BW and a Wacom Cintiq 21UX -- with each monitor running on its own graphics card (Nvidia 8400 and 8800, respectively). I'm calibrating with an X-Rite i1Display2. According to the i1D2, both monitors are properly calibrated -- but the Cintiq is much warmer/redder, and the Samsung is considerably cooler and yellow/greener. What gives?

It occurred to me that the problem could just be that I am using monitors of two different qualities (cheapie Samsung with a TN panel vs. H-IPS panel Cintiq), but shouldn't that only affect the color gamut - not the basic color tones? I'm considering dropping $2K on an NEC or Eizo to replace the Samsung, but it's only worth it if I know I that I can achieve color consistency with my Cintiq and ultimately accurate color overall.

At this point, I just have no idea if I'm getting inaccurate color from the Samsung or the Cintiq -- and, more importantly, I don't know why it's happening. Please help!

posted by roundrock to computers & internet (12 comments total)

Are you running Windows?
posted by mrbarrett.com at 6:33 AM on March 8

Sorry - yes - Windows Vista x64. I
posted by roundrock at 7:20 AM on March 8

Is your goal to merely make them match, or are you calibrating to some third external point of reference, such as a printer?
posted by rokusan at 7:24 AM on March 8

(I ask because color calibration is inherently relative. My own two eyes are different, for crying out loud, one with a much bluer whitepoint and the other with one much more yellow. So without a reference point...)
posted by rokusan at 7:25 AM on March 8

I don't just want color consistency - I could hand tweak the monitors to make them the same (but I shouldn't have to do that after spending $200 on a calibrator); my focus on the inconsistency is just that it obviously indicates that one (or both) of the monitors is inaccurate. My goal is to make them accurate -- so that anyone else viewing photos on a calibrated monitor is seeing what I see and so that I have color consistency between my monitors and my printer + any given pro lab.
posted by roundrock at 7:28 AM on March 8

You did let them warm up before calibrating, right? (30 minutes is best, IMO)
posted by wierdo at 8:38 AM on March 8

Seconding wierdo- something has gone wrong with the calibration process.

Could be-

- monitors change their color response as they warm up?
- monitors change their color response over time generally?
- computer isn't holding your settings as you expect it to?
- calibrator isn't as accurate as it should be? Does it measure actual colors, or just light level?
- some kind of weirdness with the gamma response? Meaning, the monitors might be "correct" at certain levels, but out of whack at others.

What I'd do is reset the monitors to their default settings, inside the monitor's controls. Then, make eyeball adjustments to get them close to one another. See if you can find the specs on the monitors and find out what their "natural" settings ought to be; meaning at what brightness/contrast/color temperature settings they respond most linearly to. Or, just eyeball it- get them close enough to each other without making too many adjustments. More adjustments (to me) means the monitor won't respond as linearly as the signal changes.

(I'm thinking of my television here- it has a "professional" setting for all its levels that is purported to be the most "uncolored" rendering of the material sent to it. It looks sort of dull, but that's because most TV signals are sort of dull. However, on material that's high def and really optimized to look good (CSI:Miami, certain sports), it really performs beautifully.)

Then, use the control panels of the graphics cards to adjust the colors- I wouldn't use any operating system or application level color corrections, it should be unnecessary and would just add a layer of obfuscation. The graphics card settings (should) be more "real". Meaning, when the computer asks for green, the card sends the appropriate signal to the display device to render green appropriately.

And test your settings at other levels besides the pure test patterns. Black should be black, white should be white, and 50% gray should have no color cast. That's probably easy to get set right. Try 75% gray, 25% gray, etc.

(Failing that, you're probably screwed. You might be seeing differences that others wouldn't ever see. Or the monitors themselves are just too different to get right. They use little fluorescent tubes to create the light, and then RGB filters to adjust the hue. If the inherent color temps of the tubes, or the hues of the RGB filters in the pixels are different enough from one another, you'll never get them to match at all possible combinations. I would bet $3 that getting them close enough will be more than accurate given the differences between their RGB method and the CMYK methods of printing...)
posted by gjc at 9:06 AM on March 8

Is color management on by default in Windows? That is, is the system actually referencing the icc profiles you generated from the calibration process?
(sorry if this sounds stupid...I'm a Mac guy and color management always seemed a lot easier to, well, manage, thanks to ColorSync.)

You might want to head over to the b4print forums and pose your question. Color management is their bread-and-butter.
posted by Thorzdad at 9:41 AM on March 8

Yes - the monitors were warmed up.

Yes - the system is referencing the ICC profiles (after a battle to make them do so!!) via an extra application downloaded from the X-Rite site (Windows was dropping the profile every time it came back from sleep or rebooted).

I did start off with the default monitor profiles prior to calibration (both monitors are relatively new), and I'm really not thrilled with the idea of doing everything by hand in the graphics card control panel - I purchased the i1D2 so that I wouldn't have to eyeball my color, and I know that a lot of pros use this hardware with great results, so I really don't want to abandon it unless that my only option.

Thanks for the suggestions so far ... really hoping to get this working.

I'm headed over to the b4print forums now!
posted by roundrock at 10:04 AM on March 8

I've only ever done dual-monitor calibration under XP, but when I tried, the video subsystem only allowed one profile to be loaded at a time, regardless of what the calibration screen was saying. Both monitors were on one video board, so that might not apply in this case, but it just as equally might. Do you get different results depending on the order in which you calibrate the monitors?
posted by Caviar at 11:33 AM on March 8

Windows does not handle color management well when it comes to dual monitors (I don't know if Mac is better), which is why I got a second graphics card -- there's one profile for each card, and that does work fine (i.e., when I apply a new ICC profile to one monitor, only the colors on that monitor change -- the other monitor sticks to its own profile).

The people at b4print seem to think that it is indeed simply an issue of TN panel vs H/S-IPS. May be a 6-bit vs 8-bit color issue. Apparently with two such different technologies, color-wise, no amount of calibration will make them look the same.

So that's where I am. Guessing I'll be dropping a few $K...

That said, other theories on why this situation are still appreciated! :)
posted by roundrock at 2:03 PM on March 8

Mac is better. I have four monitors, and they're all pretty close despite being different kinds of panels (calibrated with a Spyder).
posted by Caviar at 4:56 PM on March 8