XYZLUT Profile - Problem with

Anonymous
2013-11-16
2014-01-22

  • Anonymous
    2013-11-16

    Hi,

    when creating a Curves+Matrix Profile for my display and using it, I get normal colors in dark areas, everything looks good. When using a XYZLUT+Matrix Profile instead everything seems ok but dark areas of pictures now look strange, like there are details visible that shouldn't be visible and the whole dark parts look to bright and compressed.
    I tried it with IrfanView (CM activated) as well as Photoshop CC and dark areas look wrong in both programs.

    Is there anything special to consider regarding XYZLUTs?

    Btw, my settings for calibration&profiling are:
    5800K, 130cd/m^2, Gamma 2.2

    Any help would be appreciated.

     
  • Florian Hoech
    Florian Hoech
    2013-11-16

    Hi, what settings did you use for creating the XYZLUT profile (especially testchart)? Can you attach both profiles so I can have a look?

     

  • Anonymous
    2013-11-16

    Hi,
    I was using the the predefined very large testchart. Please see the attachment for complete settings.

    I'm wondering right now if I'm doing something wrong, since when creating a 3D LUT for madVR I get "washed out" dark scenes as well.

    Just to clarify that: My blacks are totally ok, they are displayed as complete blacks, it's just the "near black" colors that are somwewhat wrong.

     
    Attachments
  • Florian Hoech
    Florian Hoech
    2013-11-16

    Looking at the profiles, the tone reproduction curves are almost the same. The calibration curves show a curious difference though, probably due to either display or instrument drift (or both). How long was the display powered on before calibration/profiling? Was the i1 attached to the screen during that time?

     

  • Anonymous
    2013-11-17

    I always power the display on at least 1 hour before calibration/profiling, so that shouldn't be a problem. As the measurement device is a i1 Pro 2 I think that shouldn't have a drift at all, no matter at which temperature it is operated.
    It's possible though since I made these 2 profiles about 12 hours apart, that I changed some settings in my display in between. However, I thought that profiling should make the images look nearly the same even on two different displays, so changing some little things in the display settings isn't supposed to make such a difference...

    I'm going to do the calibration & profiling now with the same settings (1xXYZLUT, 1xCurve+Matrix), immediately one after another, to have a better comparison of the two.
    I'll report back as soon as I'm done with it.

     
    • Florian Hoech
      Florian Hoech
      2013-11-17

      It's possible though since I made these 2 profiles about 12 hours apart, that I changed some settings in my display in between. However, I thought that profiling should make the images look nearly the same even on two different displays, so changing some little things in the display settings isn't supposed to make such a difference...

      If you change display settings after profiling, you invalidate any profiles that are already made.

       

  • Anonymous
    2013-11-17

    Ok, so here are the two profiles, made immediately after each other with no changes at all. Both were made with the same settings as shown in the picture in my 2nd post, as testchart for the curves+matrix-profile I have selected the "very-large testchart for matrix profiles".

    The results are the same as reported. While the curves+matrix profile shows reasonably good shades of dark colors, the lut profile produces areas of the same grey.

    I've attached two example pictures taken with a camera from my monitor to show the problem.
    The new profiles are also attached.

     
  • Florian Hoech
    Florian Hoech
    2013-11-17

    I always power the display on at least 1 hour before calibration/profiling, so that shouldn't be a problem. As the measurement device is a i1 Pro 2 I think that shouldn't have a drift at all, no matter at which temperature it is operated.

    You can compare the calibration curves of the two profiles and see that there seems to be a fair amount of drift (also manifesting itself in a luminance difference of about 2 cd/m2). So it isn't that stable.

    While the curves+matrix profile shows reasonably good shades of dark colors, the lut profile produces areas of the same grey.

    Sorry, I can't reproduce this using the latest two profiles. Results are almost identical between them, see attached images (source space: sRGB, intent: relative colorimetric).

     
    Last edit: Florian Hoech 2013-11-17

  • Anonymous
    2013-12-20

    I'm having the exact same problem. I have been experimenting with black point compensation, offset and everything else that has to do with blacks, but with limited success. This is with D6500 LUT + swapped matrix with the largest test chart, optimized for sRGB D6500. But even compared to curves + matrix, the colormunki display software produces deeper blacks, instead of the solid grey I get with dispcalGUI. With dispcalGUI rgb(1,1,1) is (too) easily distinguishable from rgb(0,0,0). Which is a shame, because the dispcalGUI results are otherwise much better (more neutral grays, for instance).

    Question: What kind of influence does/can the test chart have on this behavior? I didn't try to just calibrate, without profiling yet.

     

  • Anonymous
    2013-12-20

    Update: with curves + matrix, black compensation turned off, D6500, large test chart, I ended up with blacks that are almost identical to the ones I got from the Colormunki software. Colors are way more off when compared to LUT, though. Still, I think this is the most pleasing profile I got out of the software. Any tips that might enhance this further? Single curve, for instance? Since every calibration takes a few hours, I'm not going to try everything.

     

  • Anonymous
    2013-12-20

    You can clearly see the difference in the darkest tones between LUT and matrix if you look at the attached curves.

     
  • Florian Hoech
    Florian Hoech
    2013-12-20

    One is using BPC, the other isn't, that's not good for a comparison.

     

  • Anonymous
    2013-12-20

    Both are without black point compensation. What makes you think bpc is on? I'm using the "-" sign for turned off and "+" for turned on. Maybe that's what's causing confusion.

     
    • Florian Hoech
      Florian Hoech
      2013-12-20

      Can you attach the two profiles please?

       

  • Anonymous
    2013-12-20

    It might be contributing, but it's not causing the issue. Anyway, I'm happy with my current matrix profile, as it's much better than the one from the Colormunki software (which has an unacceptable green color cast). But the LUT profile is still more neutral overall and I will switch instantly whenever this issue ceases to exist.

    I've attached two photos, just to show how apparent the difference is.

     
  • I tried, but I had black point compensation turned on and it was on medium. I will try your suggestion, thanks!

    (Not anonymous anymore, but I'm the same person.)

     
  • Florian Hoech
    Florian Hoech
    2013-12-21

    What you could try is setting black point (hue) correction to a high value (e.g. around 100%). This will lighten up black, but should lead to the tone response curves meeting at zero in the profile.