Displaycal/argyllcms overcorrecting 2% grayscale on OLED

Home Forums General Discussion Displaycal/argyllcms overcorrecting 2% grayscale on OLED

Viewing 15 posts - 1 through 15 (of 16 total)
  • Author
    Posts
  • #24321

    A.ces
    Participant
    • Offline

    It seems that displaycal/argyllcms is overcorrecting the 2% gamma, , leading to the i1d3 probe to no longer be able to measure the 2% at all, I needed to increase 2.5% luminance in the OLED settings 10 clicks up to make it readable by the probe, and measure 2.39-2.4 is there anything i, or argyllcms/displaycal can do to correct this?

    Also it seems that the RGB balance is basically still perfect at 2% even if I manually adjust the luminance there via the TV settings

    On another note is it possible to get the same gamma curve behaviour, and color temperature on the lower end on the ICC profile as when using a simulation profile, for example verification of ICC profile shows 2-7.5% color temperature going up to around 7200k, or over while simulation profile shows it pretty much measuring flat to around 6500k ish

    #24331

    A.ces
    Participant
    • Offline

    Okay so the calibration for grayscale/gamma looks pretty good when i use use simulation profile as display profile with enable 3d lut enabled

    but why when using a simulation profile but only selecting custom gamma 2.4 do i i get 2% over-corrected leading to the i1d3 not even reading that area in the second file labeled without 3dlut

    on the third file i added measurement using the ICC profile only, is it possible to have the ICC profile also correct/measure the same in the Correlated Color Temperature/white balance area  2-15%  just like the 3DLUT madVR corrected one in the first file does?

    Attachments:
    You must be logged in to view attached files.
    #24335

    Vincent
    Participant
    • Offline

    1st one – apply LUT3D (content colrspace in -> display colorspace out with whatever settings you chose), measure results against rec709 2.4

    2nd one – apply color management to render “other” color values than direct display RGB values so measured color matches simulated profile. It may cause that kind of cliping in black with RGB grey 6 (05 row). IDNK your profile settings. That’s why most photo app have some kind of BPC in its color engines, even if profile has not BPC.
    I mean instead of a LUT3D providing static precalculated rec709 g2.4 to display colorspace transformations it is doing them on the fly.

    3rd one – verify if display profie matches display. Direct RGB values to display, no transformation, measure them compare them to display predicted values.

    #24365

    A.ces
    Participant
    • Offline

    1st one – apply LUT3D (content colrspace in -> display colorspace out with whatever settings you chose), measure results against rec709 2.4

    2nd one – apply color management to render “other” color values than direct display RGB values so measured color matches simulated profile. It may cause that kind of cliping in black with RGB grey 6 (05 row). IDNK your profile settings. That’s why most photo app have some kind of BPC in its color engines, even if profile has not BPC.
    I mean instead of a LUT3D providing static precalculated rec709 g2.4 to display colorspace transformations it is doing them on the fly.

    3rd one – verify if display profie matches display. Direct RGB values to display, no transformation, measure them compare them to display predicted values.

    The display profile is the correct one, the values when i manually measure RGB 5 5 5/2% gray , or any other value in the 2-13% range  with spotread with the ICC profile only active measures exactly the same, while for some reason when the 3dlut is used instead that portion get’s corrected making it measure around 6500k instead of 10000-13000k range like the ICC profile does, are there any settings i should change when making the icc profile to make it do the same correction?

    #24366

    Vincent
    Participant
    • Offline

    I’ve explained in 3rd part what does profile verification. It’s verifiying if profile and display match… “whatever color coordinates grey 20, 40 , 50 or 60 have, it is stored in profile”, so send grey 20,40 or 50, measure it and if it matches what profile predicts… it’s ok.

    That info was used in 1st part (LUT3D creation) to make grey neutral “from whatever is stored in profile”. Same where simulatiing a color managed app in 2nd.

    #24367

    A.ces
    Participant
    • Offline

    I’ve explained in 3rd part what does profile verification. It’s verifiying if profile and display match… “whatever color coordinates grey 20, 40 , 50 or 60 have, it is stored in profile”, so send grey 20,40 or 50, measure it and if it matches what profile predicts… it’s ok.

    That info was used in 1st part (LUT3D creation) to make grey neutral “from whatever is stored in profile”. Same where simulatiing a color managed app in 2nd.

    Ah ok thanks i probably misunderstood what you meant a little, on other thing i noticed is that the MadTPG generator uses 240 240 240 as the 100% whitepoint compared to argyll which uses 100% is this correct behavior?

    #24369

    A.ces
    Participant
    • Offline

    Actually seems that madtpg is using 100% white as the whitepoint i think but for some reason when doing verification with madtpg i get 6510k with values near D65 coordinates, while 100% white measures 6480k  if i spotread 100% white in windows, and in argyll so now  i have no idea which is accurate…

    #24380

    Vincent
    Participant
    • Offline

    CCT is irrelevant and incomplete. If you plan to compare colors use full coordinates.

    #24389

    A.ces
    Participant
    • Offline

    CCT is irrelevant and incomplete. If you plan to compare colors use full coordinates.

    I was looking at the visual daylight temperature values which corresponds to D65, I probably should have mentioned that, anyway it still seems like when using MadTPG that it uses something different than 100% white when it reports the D65 coordinates, compared to the D65 coordinates I get when I use spotread.exe with the correct setting, and CCSS file measuring a 100% white patch in windows.

    #24523

    A.ces
    Participant
    • Offline

    I’ve explained in 3rd part what does profile verification. It’s verifiying if profile and display match… “whatever color coordinates grey 20, 40 , 50 or 60 have, it is stored in profile”, so send grey 20,40 or 50, measure it and if it matches what profile predicts… it’s ok.

    That info was used in 1st part (LUT3D creation) to make grey neutral “from whatever is stored in profile”. Same where simulatiing a color managed app in 2nd.

    Do you know if it’s possible for the VCGT 1DLUT portion of the icc profile to also have BPC  like when creating a madVR Rec 709 3DLUT to not clip/overcorrect, i tried the option labeled BPC in the profiling tab but it doesn’t  actually do anything for the gamma curve made using the ICC VCGT it still overcorrects to around 2.5, should’t argyll have a feature/sanity check for this?

    #24529

    Vincent
    Participant
    • Offline

    ArgyllCMS is not doing what you claim. The clipping happens in 2nd one and it is not related to VCGT or calibration at all.
    It is related to simulating how a color managed app will work, that’s what 2nd one does. Most of these color managed apps have some kind of BPC in their engines to avoid this even working with colorimetric intents.

    There is no issue.

    #24531

    A.ces
    Participant
    • Offline

    ArgyllCMS is not doing what you claim. The clipping happens in 2nd one and it is not related to VCGT or calibration at all.
    It is related to simulating how a color managed app will work, that’s what 2nd one does. Most of these color managed apps have some kind of BPC in their engines to avoid this even working with colorimetric intents.

    There is no issue.

    i thought when using  these settings  it only verifies  the 1Dlut portion of the profile, edit: there’s something weird going on  the lower end seems to have changed it was clipping yesterday but today  it measures 2.41….

    • This reply was modified 3 weeks, 3 days ago by A.ces.
    Attachments:
    You must be logged in to view attached files.
    #24535

    Vincent
    Participant
    • Offline

    1st one – apply LUT3D (content colrspace in -> display colorspace out with whatever settings you chose), measure results against rec709 2.4

    2nd one – apply color management to render “other” color values than direct display RGB values so measured color matches simulated profile. It may cause that kind of cliping in black with RGB grey 6 (05 row). IDNK your profile settings. That’s why most photo app have some kind of BPC in its color engines, even if profile has not BPC.
    I mean instead of a LUT3D providing static precalculated rec709 g2.4 to display colorspace transformations it is doing them on the fly.

    3rd one – verify if display profie matches display. Direct RGB values to display, no transformation, measure them compare them to display predicted values.

    Clipping on grey 6 happens on 2nd one.  Your capture is 3rd one.

    • This reply was modified 3 weeks, 3 days ago by Vincent.
    #24542

    A.ces
    Participant
    • Offline

    1st one – apply LUT3D (content colrspace in -> display colorspace out with whatever settings you chose), measure results against rec709 2.4
    2nd one – apply color management to render “other” color values than direct display RGB values so measured color matches simulated profile. It may cause that kind of cliping in black with RGB grey 6 (05 row). IDNK your profile settings. That’s why most photo app have some kind of BPC in its color engines, even if profile has not BPC.<br>
    I mean instead of a LUT3D providing static precalculated rec709 g2.4 to display colorspace transformations it is doing them on the fly.
    3rd one – verify if display profie matches display. Direct RGB values to display, no transformation, measure them compare them to display predicted values.

    Clipping on grey 6 happens on 2nd one.  Your capture is 3rd one.

    I see but this time when I did a profile yesterday I got clipping using those setting in the image also, though today after a pc restart, and sign out/sign-in it isn’t clipping and measures around 2.4, which I just find weird, is there some kind of bug in windows again regarding vcgt? I noticed that a newer Nvidia driver had broken gradients, and a big fat discoloured line when viewing a grayscale pattern, maybe I should just use a LTS release LOL.

    #24546

    Vincent
    Participant
    • Offline

    is there some kind of bug in windows again regarding vcgt? I noticed that a newer Nvidia driver had broken gradients, and a big fat discoloured line when viewing a grayscale pattern, maybe I should just use a LTS release LOL.

    IDNK, I have not see it, but I have an AMD although sometimes like just after driver update Win LUT loader kicks in, thus banding until restart.

Viewing 15 posts - 1 through 15 (of 16 total)

You must be logged in to reply to this topic.

Log in or Register

Display Calibration and Characterization powered by ArgyllCMS