Purplish blue and icm file error

Home Forums Help and Support Purplish blue and icm file error

Viewing 6 posts - 1 through 6 (of 6 total)
  • Author
    Posts
  • #1236

    connect22015 SourceForge
    Member
    • Offline

    Hello.

    I have two displays, a Samsung Syncmaster 223BW (TN Panel) and an Eizo L768 (PVA Panel).

    I’ve calibrated both using the Spyder 3 and Spyder 3 Elite software and the colors were OK except that I had a slight red tint in the dark grays on the Eizo and a slight yellow/green tint in the dark grays on the Samsung.

    I used them like this for some time, but now I’ve found out about dispcalGUI and decided to give it a try.

    I’ve calibrated both screens using dispcalGUI 3.0.4.3 and the driver for Spyder 3 that comes with it and Argyll V1.8.2 win64. I am using Windows 7 Pro x64. Now the grays are perfect neutral, but I have another bigger problem. Blues have a strong purple tint, this problem affects the Samsung with TN panel more, but it is also present, although not as noticeable, on the Eizo. All other colors are OK.

    This problem only affects color managed applications (Photoshop, Illustrator, IrfanView). If I make a background in Photoshop that is R 0 G 0 B 255 it will look purplish in Photoshop, but perfect blue when I use it as a Windows background. I also noticed that if I create my Photoshop document using ProPhoto RGB the blues are OK in Photoshop.

    The color picker panel in Photoshop also shows banding and a strong purple tint in the blue zone and the transitions are not smooth. It almost looks like solid color instead of a smooth gradient.

    I’ve checked the icm files using the “check measurement file” and coincidentally or not they both show an error exactly for the blue color. After I’ve clicked OK and dispcalGUI rebuilt the profiles, the problem was diminished, but still very noticeable.

    When I started the calibration I’ve adjusted the RGB sliders and brightness/contrast from the display’s menu to 71cd/m² and 6500K and then I restarted the calibration using As measured for whitepoint and white level.

    (I know that 71cd/m² is low, but I work in a room with low light and if I raise the white level above 75cd/m² I feel like my eyes are burning. I also used the same 71cd/m² setting before when I calibrated using the Spyder software and I had no color problems.)

    The rest of the settings were:
    White level and black level drift: Checked
    Gamma 2.2 Relative (also set to 2.2 in the display’s menu)
    Black point correction 0%
    Calibration speed: Low
    Profile quality: high
    Profile type: Curves + Matrix
    Black point compensation: Checked
    Test Chart: Extended for Matrix profiles

    I also tried recalibrating with some of the options unchecked (Black point compensation) or using Single Curve + Matrix, but the blues still look purplish.

    This is a download link to the original profiles that show the error when checked, and to the measurement reports.

    https://drive.google.com/folderview?id=0B3TrPuVK33teb1RzZFJuMmQ1aTA&usp=sharing

    I would really appreciate if someone could come up with a solution to fix this.

    Thank you, and sorry for the long post.

    #1237

    Florian Höch
    Administrator
    • Offline

    Hi,

    nothing wrong with the profiles as far as I can tell. The Samsung seems to have a quite cyan-ish blue primary, so the “purple tint” you’re seeing could be because you’re used to the native cyan-ish tone (I’m assuming sRGB material being viewed, where the blue is quite a bit deeper and out of gamut for this display).

    You usually don’t need to use drift compensation for colorimeters on TFT displays, it won’t hurt but drives up measurement times unnecessarily.

    #1238

    connect22015 SourceForge
    Member
    • Offline

    Hi, and thank you very much for your answer.

    Yes, the R 0 G 0 B 255 background image that I used for testing is sRGB. But I didn’t have this problem before with the old profile.

    Now, when I view this 255 blue test background image in a color managed viewer the image looks half blue (top half) half very purplish blue (bottom half). I know that a TN panel looks different when viewed at different angles, but it’s like the new profile amplified this and reduced the angle needed for color shifting.

    This only happens with blue and only in color managed applications, otherwise I have to change the up / down viewing angle a lot to notice important color shifts.

    Is there any setting in dispcalGUI that I can try to minimize this extreme color shifting effect? I really like how the grays look now and everything else.

    Thanks again.

    • This reply was modified on 2015-10-27 00:21:37 by connect22015.
    #1239

    Florian Höch
    Administrator
    • Offline

    Yes, the R 0 G 0 B 255 background image that I used for testing is sRGB. But I didn’t have this problem before with the old profile.

    The old profile may have been less accurrate, or more accurate, or there was display or instrument drift, or the old profile was created using a different measurement mode or colorimeter correction, or a combination of all these.

    Now, when I view this 255 blue test background image in a color managed viewer the image looks half blue (top half) half very purplish blue (bottom half). I know that a TN panel looks different when viewed at different angles […]

    That sounds unusual, even for a TN panel, and I’m not sure if that can be attributed to the ICC profile at all. A solid color will be affected uniformly by any ICC profile in color managed applications.

    Is there any setting in dispcalGUI that I can try to minimize this extreme color shifting effect?

    No, although you could create a LUT profile (use the “Office & Web” preset) instead of the shaper + matrix one which will guarantee better accuracy in case that’s the issue.

    #1240

    connect22015 SourceForge
    Member
    • Offline

    After some online searching I’ve found someone with exactly the same problem. His solution was to set something called Chromatic Adaptation to None. He was using a software called basICColor display.

    I don’t know what that option does or if it will help or not, but I would like to try it. I looked in the dispcalGUI menus and advanced options, but I didn’t find anything related. Maybe it’s under a different name?

    #1241

    Florian Höch
    Administrator
    • Offline

    You can turn the chromatic adaptation into a ‘null’ operation by setting the whitepoint to 5000K on the “Calibration” tab and re-do the calibration/profile.

    If the underlying issue is indeed a panel related one, it may also help to move the measurement window to the bottom half of the screen.

Viewing 6 posts - 1 through 6 (of 6 total)

You must be logged in to reply to this topic.

Log in or Register

Display Calibration and Characterization powered by ArgyllCMS