Eye One Display 2--worse after proflling

[diglloyd]

I understand color management pretty well I think, having long ago learned from Bill Atkinson.

I gave up on calibrating my Apple 30" LCD about two years ago using the EyeOne hardware because its calibration caused banding on the screen worse than using the built-in "dumb" calibration in the Display control panel.

Now I have the latest EyeOne Display 2 hardware puck (just bought it), and after calibration a 16-bit gray ramp created using the gradient tool shows noticeable discoloration between gray steps. Shouldn't hardware calibration make things *better*? With the "dumb" profile created using the monitor calibration in the Display control panel, the overall white point is off, but all the steps are consistent with each other, not showing discoloration between steps.

This is on an Apple 30" LCD less than a year old.

Thanks for any help you can offer.

[Martin_Doudoroff]

Distressing. What about brightness? Does that play any role, whatsoever?

[diglloyd]

Originally Posted by Martin_Doudoroff

Distressing. What about brightness? Does that play any role, whatsoever?

I don't yet know. I set things up, made the profile once, then had to go out for the rest of the day. I can make small adjustments to brightness, but it's got to work the way it is really.

[BobSmith]

You can see banding issues if you're trying to target a condition that's quite a bit different from the display's native performance. What whitepoint and gamma are you targeting? Try building the profile using settings in EyeOne Match where the whitepoint is left unchanged or at its native setting. If that doesn't work, do the same with gamma. I profile an older 20" Cinema Display at a whitepoint of 6500 and 2.2 gamma with an EyeOne spectro and it works just fine. I also have some older 17 inch and 15 inch Apple LCDs. I can't recall which but I've found that some of those have to be profiled at a different whitepoint. The native performace is closer to 5000 on one of the models.

This might also be a problem with a defective sensor as well.

Bob Smith

[Martin_Doudoroff]

I just calibrated and profiled my 23" Cinema Display with Eye One (6500k, 2.2 gamma, 120 lum), then went into PS and ran some gradient tests such as Lloyd describes. I'm consistently seeing artifacts—I wouldn't call them banding, per se, but a discoloration or texture—that are most noticeable between RGB(90,90,90) and RGB(40,40,40) and between RGB(255,255,255) and RGB(240,240,240).

As far as I can tell, color space has little impact. The artifacts are roughly the same through this display profile whether I'm working sRGB or ProPhoto RGB, suggesting that the gradient artifacts aren't a mathematical error produced by all the color space transformations.

I've seen this sort of thing before and always assumed it was a question of the physical limitations of the device. I'm no hardware expert, but I've been under the understanding that these LCD monitors are all really only about 8-bits per pixel, and non-linear, to boot, so banding would be inevitable in any "clean" gradient. Even if my monitor cannot reproduce a technically perfect gradient, I've always found it to be highly acceptable in terms of general purpose color fidelity when profiled with Eye One.

Now, if Lloyd is finding he's getting better gradients by using a dumb profile in System Preferences, I would want to ask whether he gets better overall color consistency with that or the Eye One/Two profile (real-world color subjects, rather than mathematical constructs).

[diglloyd]

Originally Posted by BobSmith

You can see banding issues if you're trying to target a condition that's quite a bit different from the display's native performance. What whitepoint and gamma are you targeting? Try building the profile using settings in EyeOne Match where the whitepoint is left unchanged or at its native setting. If that doesn't work, do the same with gamma. I profile an older 20" Cinema Display at a whitepoint of 6500 and 2.2 gamma with an EyeOne spectro and it works just fine. I also have some older 17 inch and 15 inch Apple LCDs. I can't recall which but I've found that some of those have to be profiled at a different whitepoint. The native performace is closer to 5000 on one of the models.

Thanks Bob. I am using native white point (6500K) and native gamma (2.2). Ambient light was 4 lux.

I suppose a sensor could be bad, but the older one I had produced similar problems. I'll photograph the screen and show the difference with the "dumb" profile and the one from EyeOne Display 2.

[diglloyd]

Originally Posted by Martin_Doudoroff

I just calibrated and profiled my 23" Cinema Display with Eye One (6500k, 2.2 gamma, 120 lum), then went into PS and ran some gradient tests such as Lloyd describes. I'm consistently seeing artifacts—I wouldn't call them banding, per se, but a discoloration or texture—that are most noticeable between RGB(90,90,90) and RGB(40,40,40) and between RGB(255,255,255) and RGB(240,240,240).

As far as I can tell, color space has little impact. The artifacts are roughly the same through this display profile whether I'm working sRGB or ProPhoto RGB, suggesting that the gradient artifacts aren't a mathematical error produced by all the color space transformations.

I've seen this sort of thing before and always assumed it was a question of the physical limitations of the device. I'm no hardware expert, but I've been under the understanding that these LCD monitors are all really only about 8-bits per pixel, and non-linear, to boot, so banding would be inevitable in any "clean" gradient. Even if my monitor cannot reproduce a technically perfect gradient, I've always found it to be highly acceptable in terms of general purpose color fidelity when profiled with Eye One.

Now, if Lloyd is finding he's getting better gradients by using a dumb profile in System Preferences, I would want to ask whether he gets better overall color consistency with that or the Eye One/Two profile (real-world color subjects, rather than mathematical constructs).

Thanks Martin. Banding isn't the right word, I agree. It's the (normal) stepping of the gradient; apparently the video card presumably sends only 8 bits per color to the screen, so one sees 256 steps from white to black. I suppose that is all one can expect; lack of stepping would indicate something worse or something better (a better-than-8-bit screen).

The effect I'm seeing is discoloration and "noise" on each step. It appears to me that the "dumb" profile uses a gamma curve that's somewhat different, even though the gamma is 2.2 in both cases. The EyeOne Display 2 must be measuring a gamma that's "off the curve" of 2.2 and adjusting it.

I'll photograph the screen and post an image.

Lloyd