Calibration: workflow + results thoughts

Home Forums Help and Support Calibration: workflow + results thoughts

Viewing 15 posts - 1 through 15 (of 32 total)
  • Author
    Posts
  • #2663

    user124853
    Participant
    • Offline

    Hello there,
    I am trying to calibrate & profile a monitor @6500K using Colormunki Display Device and DisplayCAL at white level and gamma value as measured ( ~2.4). I also have to mention that I’ve done this with/without white/black level drift compensation.

    As far as I’ve discussed here,  I’ve found that assuming that we have an 8bit LCD which does not have gamma correction at all (or does not allow the gamma to be set enough closely to taget) and also does not use editable internal 1D LUT, If you calibrate your LCD for gamma 2.2 the DisplayCAL will make your LCD loose colour depth because it does not offer any interactive gamma adjustment. (please correct me, that’s why I am using Gamma as measured value)

    In these conditions, my concern is related to the verification report results (attached).

    19:55:06,226 Brightness error = -0.096061 cd/m^2 (is 90.161656, should be 90.257717)
    19:55:06,226 White point error = 0.184658 deltaE
    19:55:06,230 Maximum neutral error (@ 0.673696) = 8.987361 deltaE
    19:55:06,230 Average neutral error = 5.002081 deltaE
    19:55:06,232 Number of measurements taken = 100

    In order to have the whole picture, using a gamma value at 2.2, the results are:

    15:37:13,859 Brightness error = -0.348277 cd/m^2 (is 100.478861, should be 100.827138)
    15:37:13,860 White point error = 0.115436 deltaE
    15:37:13,862 Maximum neutral error (@ 0.673696) = 0.956004 deltaE
    15:37:13,863 Average neutral error = 0.491909 deltaE
    15:37:13,864 Number of measurements taken = 100

    Am I doing something wrong, is this a bad approach?
    Thanks,

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

    Calibrite Display SL on Amazon  
    Disclosure: As an Amazon Associate I earn from qualifying purchases.

    #2665

    Euri Pinhollow
    Participant
    • Offline

    Log file contains the following: “19:48:20,085 Target advertised gamma = 2.400000”. It makes me think that DisplayCal checks against gamma 2,4 while profile is not made for this gamma. It would be nice if you uploaded the log file of another verification. I have just ran verification and I cannot find a log file to confirm it. Please screencap your verification settings.

    #2666

    user124853
    Participant
    • Offline

    On the other hand, the old (@2.2 gamma) vs. the new (@as measured gamma) html reports, using sRGB simulation profile / Tone curve Rec.1886 looks pretty close.

    Attachments:
    You must be logged in to view attached files.
    #2669

    user124853
    Participant
    • Offline

    Hello Pinhollow, I’ll do this again. Thank you for your presence 🙂

    #2670

    user124853
    Participant
    • Offline

    I confirm, please find attached file. At 2.2 gamma for example, DisplayCal checks against gamma 2.2. If this matter (but I suppose not), I didn’t check in calibration tab gamma at 2.4 or something, but As measured.

    • This reply was modified 7 years, 11 months ago by user124853.
    Attachments:
    You must be logged in to view attached files.
    #2673

    Victor Wolansky
    Participant
    • Offline

    How far is the color temperature of the Uncalibrated monitor? The furthest it is from what you want, the more quality will lose and the more banding you will have as a result of shifting too much in a only 8 bits LUT. I have a laptop that is from factory like 8900 K and take it to 6500 is possible but it creates lots of banding and not a great result no matter what I do.

    #2674

    user124853
    Participant
    • Offline

    Hello there Victor, thank you for your intervention. Please find attached screenshot (measured @6500K monito preset). I suppose that it’s pretty far away from 6500K.

    Attachments:
    You must be logged in to view attached files.
    #2676

    Euri Pinhollow
    Participant
    • Offline

    I still would like to see your verification settings and second log file containing small errors for a profile with 1D LUT.

    What I wanted to check is whether I can achieve similar error magnitude with checking “use simulation profile as target profile” – this will of course give big errors – and will get those lines in log file.

    I would also like to see screenshot of profile creation settings. I need all of these to give a comprehensive answer.

    #2679

    user124853
    Participant
    • Offline

    Sure, please have a look Above (I’ve attached the Verify calibration / Report on calibrated display device report). If you are reffering at html report, also above. All these tests are made on the 2.2gamma profile using 1D LUT. These are the settings. Hope this helps.

    • This reply was modified 7 years, 11 months ago by user124853.
    • This reply was modified 7 years, 11 months ago by user124853.
    #2682

    Euri Pinhollow
    Participant
    • Offline

    To be short: if you uncheck “simulation profile” in verification settings I expect you to achieve errors close to tolerable.

    I still cannot produce a LOG file along with HTML report but I have done gray test chart verification right now with fresh profile without 1D LUT and it gave me aximum deltaE of 2.43.

    #2684

    Euri Pinhollow
    Participant
    • Offline

    Still cannot see verification settings, they are most important.

    #2685

    user124853
    Participant
    • Offline

    Very true, here are the reports without simulation profile checked (old @2.2gamma with 1D LUT and new at gamma as measured without 1D LUT)

    These settings was applied for Verification in both cases (@2.2 with 1D LUT and gamma as measured, with or without Simulation profile checked sRGB/ Rec.1886)

    • This reply was modified 7 years, 11 months ago by user124853.
    Attachments:
    You must be logged in to view attached files.
    #2689

    user124853
    Participant
    • Offline

    Also, the Verification settings could be found in html reports under these fields if this helps.

    Testchart:
    Simulation profile:
    Gamma mapping:

    #2690

    Euri Pinhollow
    Participant
    • Offline

    Now let me explain what you see.

    Test charts basically are  raw data.

    1. If you select no simulation profile, raw data is assumed to be from colour profile which is being verified and measurements of raw data fed to display are compared against colours which they represent in profile being verified.
    2.  If you select a simulation profile, raw data is assumed to be from simulation profile and is converted to colour profile/space being verified, the result of this conversion is fed to display and measurements are compared to colours which raw data represent in simulation profile.
    3.  If you select a simulation profile AND gamma, you are effectively verifying a profile with gamma replaced by selected gamma like in (2).
    4.  If you select a simulation profile as target profile you are verifying how a device complies to this profile.

    Case (1) is a benchmark for created profiles and is guaranteed to pass until LCD drift shifts the colour.

    Case (2) is a benchmark for testing how well a simulation profile fits inside verified profile. If profile being verified is strictly larger than simulation profile, the test is guaranteed to pass if (1) is passed.

    Case (3) is a benchmark for testing (2) and compliance  to selected gamma simultaneously. It will fail if you do not create 1D LUT and your LCD is not close to selected gamma already.

    Case (4) is a benchmark for testing how well a display conforms to some profile. This is only useful for testing displays which advertise conformance to some profile or if display has internal LUTs which may make it profile-compliant.

    For seeing colour accurately within device capabilities, you only need to pass (1), everything else is surplus.

    #2692

    Euri Pinhollow
    Participant
    • Offline

    Hello there Victor, thank you for your intervention. Please find attached screenshot (measured @6500K monito preset). I suppose that it’s pretty far away from 6500K.

    Yeap, pretty away from that. Your measurement report shows that you have dealt with it successfully. If you have done it via LCD settings (which laptops do not usually have), you are not loosing colour depth.

Viewing 15 posts - 1 through 15 (of 32 total)

You must be logged in to reply to this topic.

Log in or Register

Display Calibration and Characterization powered by ArgyllCMS