Pages: [1]   Go Down

Author Topic: bizarre eye-one match behavior  (Read 8987 times)

sbay

  • Full Member
  • ***
  • Offline Offline
  • Posts: 225
    • http://stephenbayphotography.com/
bizarre eye-one match behavior
« on: June 06, 2010, 06:11:48 pm »

Today I decided to recalibrate my apple cinema display (23") with eye-one match (3.6.2) and I'm getting very odd behavior. I used D6500 and 2.2 gamma -- the first thing I noticed was that eye-one was telling me that my luminance was way too low. After adjusting brightness to read 120 CD (by turning brightness almost to the max) and finishing the calibration, the screen went very blue. Furthermore, the input-output mappings were way off (normally the RGB lines are very close to each other with a slope of one) -- now the blue was much steeper than green/red. Here is a screen shot of the eye-one summary:

http://bayimages.net/eye-one-d6500-120.png

I thought that maybe there was a software issue, so I updated to 3.6.3 but the issue still remains. However, if I calibrate using native white point everything seems fine: the colors on the screen seem more in line with my previous settings, the RGB mappings are 1 to 1, and the brightness level of the display is back to normal levels. The only odd behavior is that at the end of the calibration process eye-one is telling me that my current white point is 5000k.

http://bayimages.net/eye-one-native-120.png

I've also tried calibrating the monitor using the mac os control panel (where you match the gray apple to a lined gray background). Using the control panel, I get a native white point of 6479. And the colors are very similar to the eye-one profile with target white point set to native with a final reported value of 5000.

I'm guessing that eye-one match is simply reporting the wrong final white point, but I'm not 100% sure. Would appreciated any feedback if others have encountered this problem before.

Thanks,

Stephen

ChasP505

  • Full Member
  • ***
  • Offline Offline
  • Posts: 114
bizarre eye-one match behavior
« Reply #1 on: June 06, 2010, 08:29:35 pm »

So you made no changes to your normal monitor settings or system and just attempted to recalibrate using the settings and method that previously worked fine?
Logged
Chas P.

sbay

  • Full Member
  • ***
  • Offline Offline
  • Posts: 225
    • http://stephenbayphotography.com/
bizarre eye-one match behavior
« Reply #2 on: June 06, 2010, 09:22:33 pm »

Quote from: ChasP505
So you made no changes to your normal monitor settings or system and just attempted to recalibrate using the settings and method that previously worked fine?

Before today, I had been calibrating to D6500 and 120 brightness (the last time was a few months ago). Today, the recalibration with the same settings did not work and left a very blue cast to the screen.

Stephen
« Last Edit: June 06, 2010, 09:23:40 pm by sbay »
Logged

Pat Herold

  • Full Member
  • ***
  • Offline Offline
  • Posts: 161
bizarre eye-one match behavior
« Reply #3 on: June 07, 2010, 02:34:22 pm »

Is this a Snow Leopard ICC version 4 profile issue?

http://www.xrite.co.uk/ph_product_overview...;SupportID=5083
Logged
-Patrick Herold
  Tech Support,  chromix.com

sbay

  • Full Member
  • ***
  • Offline Offline
  • Posts: 225
    • http://stephenbayphotography.com/
bizarre eye-one match behavior
« Reply #4 on: June 07, 2010, 10:06:56 pm »

Quote from: pherold
Is this a Snow Leopard ICC version 4 profile issue?

Shouldn't be -- I am using the v2 profiles.

sbay

  • Full Member
  • ***
  • Offline Offline
  • Posts: 225
    • http://stephenbayphotography.com/
bizarre eye-one match behavior
« Reply #5 on: June 09, 2010, 08:33:36 pm »

I don't seem to be getting many responses, so let me rephrase.

My apple cinema display should have a native whitepoint of about 6500k (6479K according to the visual calibrator in the control panel). Since the native whitepoint is so close to 6500k I expect that I would get similar profiles when I ask Match to target 6500k or the native whitepoint. However, the profiles are turning out to be wildly different.

The target=6500K profile leaves a strong blue cast. In contrast, the target=native generates a more reasonable profile (to my eye) but Match reports a color temperature of 5000K. However, I don't think the display is that warm and the value contradicts what the visual calibrator reports.

Stephen

shewhorn

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 537
    • http://
bizarre eye-one match behavior
« Reply #6 on: June 10, 2010, 03:15:16 am »

Quote from: sbay
I don't seem to be getting many responses, so let me rephrase.

My apple cinema display should have a native whitepoint of about 6500k (6479K according to the visual calibrator in the control panel). Since the native whitepoint is so close to 6500k I expect that I would get similar profiles when I ask Match to target 6500k or the native whitepoint. However, the profiles are turning out to be wildly different.

The target=6500K profile leaves a strong blue cast. In contrast, the target=native generates a more reasonable profile (to my eye) but Match reports a color temperature of 5000K. However, I don't think the display is that warm and the value contradicts what the visual calibrator reports.

Stephen

Hi Stephen,

I don't know if this is what's going on in your case but...

Colorimeters are very cheap electronics and it wouldn't surprise me at all if the device to device precision is extremely sloppy. I own an Eye One Display 2, DTP94, Spyder 3, Spectrolino, and i1 Pro. The first 3 are colorimeters, that last two are spectrophotometers. The spectros pretty much agree with one another. The colorimeters... not so much. Between my DTP94 and Spyder 3 there is a about an 1800º delta in color temperature. The Eye One Display 2 (which is actually from the Spectraview kit) is the closets to reality in terms of desired color temp. If I had multiple copies of the Eye One Display 2 however, I'm not sure how consistent it would be from copy to copy, whether it would be 200 degrees or 900 degrees or more I just don't know as I haven't had multiple devices to compare. I have compared two DTP94's side by side once and they were 200ºK apart from one another (and the DTP94 is considered to be one of the most consistent colorimeters).

This is one of the reasons why I scratch my head when people say "calibrate to 6500º K". Well, yes, it's good to have a standard but when the tools being used to measure such results are... well.... flaky, what's the point? If two people have the EXACT same screen and one person is using a DataColor colorimeter, and another is using an X-Rite device, I can almost guarantee you that if both folks calibrated to 6500ºK, that neither of those screen would match one another.

Now... I'm NOT saying to go out an buy a spectrophotometer. They are expensive and they typically don't do as good of a job at measuring the shadows from an emissive device like a TFT. If you can get your hands on one though you could profile your screen with it, take a white point reading with your colorimeter and then you'll know what the offset is and you can recalibrate with whatever that measured value is.

OR... if the monitor has been designed to have a native white point of 6500ºK then instead of setting the target color temp to 6500ºK, set it to "native" instead. This has an additional benefit in that when you profile a TFT to "native" it means that the profiling software no longer has to bend one of the RGB channel curves in order to attain the desired color temp and this means less overall correction which means less posterization (often referred to as "banding") as a result of your monitor profile.

Cheers, Joe
Logged

sbay

  • Full Member
  • ***
  • Offline Offline
  • Posts: 225
    • http://stephenbayphotography.com/
bizarre eye-one match behavior
« Reply #7 on: June 10, 2010, 10:21:44 am »

Thanks for the reply Joe.

I initially thought that the issue was a hardware problem with my puck and that it had gone bad (it is several years old). In which case the solution would be simple -- just get a new one. However, that doesn't explain to me why the two profiles would be so different (if the native whitepoint of the monitor is about 6500K then choosing a target of 6500k or native should give similar results). So then I started thinking it may be caused by a software problem as well.

I'm also corresponding with the Xrite support staff, but so far we haven't been able to figure anything out other than something is wrong and going through the usual process of eliminating other candidate causes (e.g. monitor presets which the ACD doesn't have, universal access, etc.).

i1diagnostics reports that the puck is okay and working fine. However when I look at what it displays and has the puck measure, I don't see how it could tell if the color readings were drifting off specification.

Stephen

shewhorn

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 537
    • http://
bizarre eye-one match behavior
« Reply #8 on: June 10, 2010, 10:54:34 am »

Quote from: sbay
i1diagnostics reports that the puck is okay and working fine. However when I look at what it displays and has the puck measure, I don't see how it could tell if the color readings were drifting off specification.

Stephen

That's the thing, it can't. If I were to bet, I'd bet that your puck is absolutely fine, it just reports (like my DTO94 and Spyder 3) a different color temp for 6500ºK. The spectrophotometers like the Spectrolino and i1 Pro come with a white tile. The tile is actually matched by serial number to the individual device and every time before you take a set of measurements it needs to be calibrated to that white tile so those devices where white point is concerned tend to be a lot more consistent. I'm not sure how much it would be to have the device checked out but perhaps give X-Rite a call.

There is another possibility and it could be that your display is on its way out. Even with a radically different color temp, having to crank it up to its max to get 120 cd/m^2 indicates to me that the CCFL backlight is a little old. Over time CCFL backlights do lose their brightness. There was actually a batch of 23" ACDs that were known for losing half of their brightness in as little as 6 months. If this is a radical, all of a sudden change though I'd be less likely to believe that although if there was no change to your system and you were previously getting 6500ºK I'd point my finger at either the monitor or the puck way before considering software as the problem. If you wanted to test it with different software though you could download a demo copy of Color Eyes Display Pro (demo lasts for 10 days) which will be compatible with your Eye One Display 2.

Cheers, Joe
Logged

sbay

  • Full Member
  • ***
  • Offline Offline
  • Posts: 225
    • http://stephenbayphotography.com/
bizarre eye-one match behavior
« Reply #9 on: June 10, 2010, 11:30:57 am »


The weird thing is that it is only when I use a non native whitepoint target that I have to crank the brightness all the way up. When I use native as the target the brightness is at normal levels to reach 120 cd/m^2 (a little under the halfway point which is consistent with my prior use).

In fact, even if I don't touch the brightness setting on the display, the instant I select the different whitepoint settings in the dropdown menu, the brightness changes tremendously (i.e., switching from native to 6500K results in an immediate dimming of the display).

Thanks for the tip on the Color Eyes demo. I'll check it out.

Stephen

shewhorn

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 537
    • http://
bizarre eye-one match behavior
« Reply #10 on: June 10, 2010, 12:45:35 pm »

Quote from: sbay
In fact, even if I don't touch the brightness setting on the display, the instant I select the different whitepoint settings in the dropdown menu, the brightness changes tremendously (i.e., switching from native to 6500K results in an immediate dimming of the display).

This is perfectly normal. TFTs can not change their color temperature. They are fixed based upon the color of the CCFL backlight. In order to change the color temperature of a TFT you have to dim pixels. One thing a lot of people don't understand is that when your screen is displaying pure white, all pixels are actually OFF. When you're displaying "black" all pixels are on. The reason you pay attention to things like black levels are because the backlight actually bleeds through the pixels (which is why TFT panels have more difficulty achieving a good black level).

So, when you set your calibration software to "native" white point that means that all pixels on the display are off and that it doesn't adjust the color temperature. Instead it measures that white point and then starts correcting color. The resulting video LUT curve is pretty much going to be at a 45º angle going up from left to right. You'll see slight deviations of the R,G,B channels but that's only to correct the colors that are out of whack.

When you tell your calibration software to set the display to a specific color temp, it must do this through the LUT (lookup table) on a TFT. With a CRT what you used to do is adjust the individual output levels of the red, green, and blue guns to change the white point. Since we don't have individual red, green and blue guns* we have to start dimming pixels. This is equivalent to say creating a blank white image in Photoshop, creating a curves layer, switching to the blue channel and then setting the output to say... 200 (this will give "white" a yellow cast). As a result the overall luminance goes down. Your screen probably needs to pull two channels in order to get to the desired color temp that you are specifying resulting in an even greater reduction in luminance.

Some software is better at compensating for the reduction in luminance that occurs as a result of adjusting the white point via the video LUT but... every time you change your profile you do need to recalibration the backlight output in order to compensate for any loss in luminance as a result of profiling to a white point that is not the monitor's native white point.

* Now for the asterisk. Some displays such as the HP Dreamcolor, and Lacie 724 and 730 (which are actually the Samsung XR24 and XR30 respectively) have red, green, and blue LED backlights and with these screens you can actually truly change the white point of the screen without negatively impacting desnity resolution. These screens are however VERY expensive.

Cheers, Joe
Logged

sbay

  • Full Member
  • ***
  • Offline Offline
  • Posts: 225
    • http://stephenbayphotography.com/
bizarre eye-one match behavior
« Reply #11 on: June 29, 2010, 09:33:48 pm »

Just as an update -- I've had a number of discussions with the support staff at Xrite and they were pretty helpful. After sending them the ICC profiles and testing my puck on a few monitors we determined that it was likely to be bad. I got a new i1 display (actually the LT model but they sent me a serial to upgrade it to the display 2) and that solved the calibration problem.

Stephen

shewhorn

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 537
    • http://
bizarre eye-one match behavior
« Reply #12 on: June 29, 2010, 10:40:33 pm »

Quote from: sbay
Just as an update -- I've had a number of discussions with the support staff at Xrite and they were pretty helpful. After sending them the ICC profiles and testing my puck on a few monitors we determined that it was likely to be bad. I got a new i1 display (actually the LT model but they sent me a serial to upgrade it to the display 2) and that solved the calibration problem.

Stephen

Glad you found a solution and good to know. Rare that it happens but everything will eventually fail at some point. What were the steps they took you through to determine that the puck was faulty (or did they have you send it in so they could evaluate it)?

Cheers, Joe
Logged
Pages: [1]   Go Up