Green cast after making a 3D LUT in Resolve

Home Forums Help and Support Green cast after making a 3D LUT in Resolve

Viewing 13 posts - 1 through 13 (of 13 total)
  • Author
    Posts
  • #32583

    Googloiss
    Participant
    • Offline

    Hi, I’ve made a calibration and a 3D LUT using Resolve/Calman -> Decklink Mini Monitor 4K -> Mini Converter SDI to HDMI 6G, and I loaded this LUT into the mini converter.

    After grading a video, the result on other screens is obviously a bit towards magenta, so I must have compensated for a green cast on my calibrated monitor (LG 31MU97), why is this happening? Some wrong displaycal setting? I don’t know what to think.

    Thank you!

    #32584

    Vincent
    Participant
    • Offline

    If it has a green cast (amongst all the failures present in such display) it should be visible in desktop.
    If it has noit such cast, you must have done something wrong when creating LUT3D. Use same ICC as you used to create LUT3D but set colroiemtric relative in rendering intent, that will keep desktop white 255 without correcting it.

    #32594

    Googloiss
    Participant
    • Offline

    Hi Vincent!

    The LUT3D I made was already with relative colorimetric… And the verification tests with displaycal were ok.

    I can’t check the green cast in desktop because I’m in decklink input mode, so it’s nothing to do with OS or my graphic card.

    I was thinking about my correction, which has always been LCD GB-r-LED IPS (Dell U2413), can this be the problem?

    #32601

    Vincent
    Participant
    • Offline

    Hi Vincent!

    The LUT3D I made was already with relative colorimetric… And the verification tests with displaycal were ok.

    I can’t check the green cast in desktop because I’m in decklink input mode, so it’s nothing to do with OS or my graphic card.

    I was thinking about my correction, which has always been LCD GB-r-LED IPS (Dell U2413), can this be the problem?

    This is the proper correction.

    Just put a video sample Rec709 MP4 with a 100% white saturation patch on Resolve and measure with ArgyllCMS spotread in command line. Check if it is white or it has the color cast you say. Same with other greys with X% IRE.

    (you must use -X path_to_CCSS.ccss parameter in spotread)

    • This reply was modified 2 weeks, 5 days ago by Vincent.
    #32603

    Googloiss
    Participant
    • Offline

    I’ve never used those features in displaycal, I have to open options > advanced > set additional commandline arguments, and write “-X path_to_CCSS.ccss” in the spotread box? And what do I have to do next?

    About the correction, I meant that maybe I have to choose one of these: “LCD GB-r-LED IPS (Dell U2413)”, “LCD GB-r-LED/RG Phosphor LED family (AUO B156HW01 V.4, Dell U2413)” or “LCD RG Phosphor LED family (AUO B156HW01 V.4 in Lenovo W520/W530)”, from what I understand they are all similar to each other, and I’ve read someone has had a greenish tint with one of these but not with the others.

    #32604

    Vincent
    Participant
    • Offline

    Command line means command line. Run press a key when requested and it will take a measurement. See measured coords and where are they placed relative to your target.

    If there is a visual mismatch between a  perfect measured D65 and your eyes and you can measure with a high res spectrophotometer, use visual whitepoint editor, as you can see in many threads

    #32605

    Googloiss
    Participant
    • Offline

    Usually to check the white point I simply start the interactive display adjustment and see what it tells me, without actually starting the calibration then. And doing so, my white point is correct, no green cast.

    #32608

    Googloiss
    Participant
    • Offline

    I want to make an update, I’ve checked multiple corrections through the interactive display adjustment and I did some more research on my monitor’s backlight type, finding this: https://www.displayspecifications.com/en/model/d1ea6c6, site mentioned in this forum as a reference: https://hub.displaycal.net/forums/topic/useful-resources-5dcf6cf9/

    Apparently my monitor’s backlight is W-LED, so I chose “LCD White LED IPS” correction (the only one that mention non-oled LG panels in parentheses). And in the interactive display adjustment it turned out to be less greenish (and overall more white to my eyes) than GB-r-LED correction, that is what I was expecting, this is where the green cast could come from…

    Am I missing something?

    #32609

    Vincent
    Participant
    • Offline

    It is not WLED, 100% sure since it is not sRGB only. Also it is “very” old to use new WLED PFS/KFS backlight (HP Z24x correction in DisplayCAL, if you want to try a WLED it needs to be that one, but your monitor is too old, first models with an hybrid mix those new phsohor and old phoshor were UP2516D/UP1716D as you can see by plotting a user made CCSS for those models).

    It is usesless to waste time in “interactive display” window. Put a 100% saturation white on Resolve and measure it with spotread, that will validate the whole chain regarding whitepoint: Resolve, LUT3D, decklink and monitor

    • This reply was modified 2 weeks, 5 days ago by Vincent.
    • This reply was modified 2 weeks, 5 days ago by Vincent.
    #32612

    Googloiss
    Participant
    • Offline

    I don’t know the steps I have to do in order to do the measurement you suggest, can you guide me through it?

    #32613

    Vincent
    Participant
    • Offline

    https://www.argyllcms.com/doc/spotread.html

    Open a terminal/console. Go to ArgyllCMS’ bin folder, or add that bin folder to $PATH.
    Then put a 100% white saturation video in Resolve. You can get Rec709 patches in MP4 in AVSForum.
    Place colorimeter in the center over that Rec709 100% saturation white patch.

    Then:

    spotread -x -X path_to_GB_LED.ccss

    Where path_to_GB_LED.ccss is the path of teh CCSS correction, wherever you have stored it.
    When requested press any key and it will take a measurement. It will be written in CIE xyY coords (-x param) so it will be easier to you to place in in a CIE xy 2D plot. Check its whiteness.
    To quit press Q or ESC (I do notremember, it will instruct you which key is needed) or take another measurement with any other key.

    CalMAN should have an equivalent free measure command in its UI, or HCFR if you can use a windows laptop (or VM).

    • This reply was modified 2 weeks, 5 days ago by Vincent.
    #32615

    Googloiss
    Participant
    • Offline

    I’m afraid we’re way beyond my capabilities, are we even still talking about displaycal software? I’m unable to decipher these instructions… Already starting from the first step, open a terminal/console? What terminal?

    • This reply was modified 2 weeks, 5 days ago by Googloiss.
    #32617

    Googloiss
    Participant
    • Offline

    Anyway, I’ve run a verification report unchecking device link profile, the measured whitepoint is x 0.3145 y 0.3302, that is what report interactive display adjustment too… Other than that I don’t know how to measure the on-screen whitepoint.

Viewing 13 posts - 1 through 13 (of 13 total)

You must be logged in to reply to this topic.

Log in or Register

Display Calibration and Characterization powered by ArgyllCMS