What is wrong with the 3D lut generation?

Home Forums Help and Support What is wrong with the 3D lut generation?

Viewing 6 posts - 1 through 6 (of 6 total)
  • Author
    Posts
  • #27000

    Roman Strijbos
    Participant
    • Offline

    I’m sorry I posted in the wrong ‘general’ forum just over an hour a go. This post is the correct one.

    I’m trying to calibrate my EIZO CG2420 but I can’t get the greys right. I followed this guide to the letter but i can’t reproduce the same results Macleod92 is getting.

    I tried all the settings in the 3DLUT generation tab. Firstly the results were VERY off, almost if there was a glitch in the 3Dluts. Very big color banding in the blacks. When I downgraded to Argyll 1.8.3 (as I read somewhere on this forum) the results got better but still not where I would like them to be. I’m not very technical in all this but why isn’t displaycal able to adjust the greys with whitepoint scaling? The result I got using colornavigator as a base is OK but it’s far from ideal and I have the feeling that a proper 3DLUT could do so much more for my screen.

    Any thoughts or solutions? What could be the problem?

    The colornavigator calibration verification looks ok, but still not super good. I would rather not rely on this calibration for the rest of my profiling in displaycal.. I changed the colornavigator whitepoint to the correct whitepoint of displaycal btw.

    Ill attach the two measurement reports.

    Attachments:
    You must be logged in to view attached files.
    #27004

    Roman Strijbos
    Participant
    • Offline

    Now that I ran a calibration with very few patches (175) I find that my accuracy is increasing. Very weird! Any thoughts?

    #27005

    MW
    Participant
    • Offline

    Not sure about your issue but the 175 patch set uses a different implementation. It’s optimized for PC monitors that’s why it’s expected to give better results on your Eizo.

    #27010

    Vincent
    Participant
    • Offline

    -CN calibration seems to be at native gamut and looks flawless. dE in near black are caused by CN ICC profile which stores “infinite contrast perfect black” to avoid some rounding errors and issues in color manged apps and of course your display has not infinite contrast. Just look on expected vs measured L*. All other things are perfect including grey range.

    -Regarding Resolve LUT3D all seems OK. If you are getting issues on grey “visually” on the videos or gradients you work with since base HW calibration in your cal2 is so good, you can try to “simplify” things.
    > intent : relative colorimetric => “my white is perfect”
    > profile : XYZLUT + BPC => “my black is perfect although I want full gamut volume correction in a detailed way”
    Just do not try to simulate rec1886 in an IPS display. Check visually results and check actual errors (near black errors may rise)

    • This reply was modified 1 month, 2 weeks ago by Vincent.
    #27044

    Roman Strijbos
    Participant
    • Offline

    Thanks guys,

    I tried again with BPC on, the result gets worse.. And I already had the ‘relative colorimetric’ setting on. Still: more patches generates a more glitchy 3DLUT with weird (visible) artifacts in the greys.

    What could cause this?

    #27055

    Vincent
    Participant
    • Offline

    Thanks guys,

    I tried again with BPC on, the result gets worse.. And I already had the ‘relative colorimetric’ setting on. Still: more patches generates a more glitchy 3DLUT with weird (visible) artifacts in the greys.

    What could cause this?

    LUT3D is the result of a transformation between 2 colorspaces (profiles).
    Try to verify if destination profile (“destination” in LUT3D maker, the one created by displaycal) actually matches your display in HW cal mode, specially on grayscale.

    As temporal solution use CN to get a Rec709 g.24 HW cal, verify it in displaycal as if simulation profile is displayprofile, no LUT3D or device link involved, just if HW cal matches what you choose as target.

    • This reply was modified 1 month, 2 weeks ago by Vincent.
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