I made a tool for applying 3D LUTs to the Windows desktop

Home Forums General Discussion I made a tool for applying 3D LUTs to the Windows desktop

Viewing 9 posts - 271 through 279 (of 279 total)
  • Author
    Posts
  • #38593

    Vincent
    Participant
    • Offline

    But when I compared them on a QD-OLED they both measure very differently. i1 Pro 2 is not accurate enough.

    Where does SPD drift from each other on a QD-OLED (using i1pro2 at 3nm)?

    i1Basic Pro 2 on Amazon  
    Disclosure: As an Amazon Associate I earn from qualifying purchases.

    #38594

    Vincent
    Participant
    • Offline

    Here’s my verification with the method I explained above in SDR Rec.709, 2.4 gamma.

    Display is a Sony A95K

    Probes I used are Jeti 1501 & i1 Display Pro

    I verified in DisplayCal and HCFR. And it’s all spot on accurate.

    Without doing the maths… it looks very close to D65 on CIE 2012 2 degree.

    i1Display Pro on Amazon  
    Disclosure: As an Amazon Associate I earn from qualifying purchases.

    #38596

    EP98
    Participant
    • Offline

    Where does SPD drift from each other on a QD-OLED (using i1pro2 at 3nm)?

    I don’t know. I don’t have the i1 Pro 2 anymore to check.

    Without doing the maths… it looks very close to D65 on CIE 2012 2 degree.

    I’ll have to check later. QD-OLED adjusted to CIE 2012-2 compared to a D65 1931-2 CRT to see if they match WP.

    But I do all calibrations in 1931-2. With a perceptual match. I would only use Alternate CMF’s for white point only if I were to use them.

    When I tried 2012 for full calibration, it percetually looks different from 1931.

    Mostly in blue. Blue takes on a purplish hue. When I compared two oleds. One adjusted to 1931, the other to 2012. Also I compared to a 1931 crt so I can have a good reference. 2012 is just too different for it to be usable.

    #38597

    EP98
    Participant
    • Offline

    Where does SPD drift from each other on a QD-OLED (using i1pro2 at 3nm)?

    When I would switch i1 Pro 2 bewteen 10nm and hi res mode the switch barely made a difference. About 0.01 difference in measurment.

    But the 3nm Jeti was significantly different in measurment. Enough at least to make a perceptible difference.

    The biggest difference was in White Balance. Less so in Color. You can compensate for the less accuarate i1 Pro 2 by calibrating a display it does well on like a smaller gamut display. Then perceptually match to the QD display. To work around its limitations.

    • This reply was modified 1 week, 6 days ago by EP98.
    #38599

    Vincent
    Participant
    • Offline

    Where does SPD drift from each other on a QD-OLED (using i1pro2 at 3nm)?

    When I would switch i1 Pro 2 bewteen 10nm and hi res mode the switch barely made a difference. About 0.01 difference in measurment.

    But the 3nm Jeti was significantly different in measurment. Enough at least to make a perceptible difference.

    The biggest difference was in White Balance. Less so in Color. You can compensate for the less accuarate i1 Pro 2 by calibrating a display it does well on like a smaller gamut display. Then perceptually match to the QD display. To work around its limitations.

    But I’m talking about the technical source of error so it’s easy to identify  “potential” misreadings without a jeti and aim for an slightly modified WP.
    I mean:
    -Do you remember if the “valleys” of SPD were taking more noise in i1Pro2? If bottom of the valley was raised up the integral may add some value that de balance R+G+B
    -Do you remember if there was a misalignment on main wavelegth of the peaks vs the Jeti Specbos? This may hint some error in difraction grating + auto correction, or deteriorated ceramic tile. Testing with i1Diagnostics may bet a hint of an actual malfunctioning i1Pro2
    -Do you remember if the left tail of blue peak fell deeper into short wavelengths, or at least in a less inclined downhill (“slow slope” in blu, on left side) than a QLED/WLED/GBLED/WLEDPFS?

    I mean, reading a QLED/RGBOLED SPD is an easy task because of the relatively slow slopes and thick/broad peaks. If an i1Pro2 was getting a misread it must be related to something like these 3 examples I wrote, not because SPD being difficult to read.

    • This reply was modified 1 week, 6 days ago by Vincent.
    #38601

    EP98
    Participant
    • Offline

    I mean, reading a QLED/RGBOLED SPD is an easy task because of the relatively slow slopes and thick/broad peaks. If an i1Pro2 was getting a misread it must be related to something like these 3 examples I wrote, not because SPD being difficult to read.

    I don’t know or remember. I sold my i1 Pro 2 a while ago so I can’t verify any of the stuff you said except for 1.

    I do know that the ceramic tile is not deterioted. I have comparisons of White Balance with both i1 Pro 2 & Jeti on a CRT. It was mostly to show the buyer that i1 Pro 2 still measures accurately.

    I ended up purging alot of my photos when I was organizing the files on my phone so that’s all I could find.

    Jeti measured x: 0.3119 / y: 0.3295

    i1 Pro measured x: 0.3107 / y: 0.3286

    Very close to each other. About +-0.015. Which is what jeti’s should measure within each other after recalibration. So my i1 Pro 2 measures as accurately as a refrence probe. So not an issue with ceramic tile.

    The other questions I don’t know. And I don’t remember the exact difference in xy they measured on QD-OLED.

    You know what. You might be able to find it. I uploaded data onto DisplayCal after I profiled my i1 Pro 2 to my i1D3. I did multiple uploads since I profiled it multiple times. You may be able to find it. I also uploaded Jeti files also. You’ll have to look for Sony A95K.

    • This reply was modified 1 week, 5 days ago by EP98.
    Attachments:
    You must be logged in to view attached files.
    #38605

    EP98
    Participant
    • Offline

    I named my ccmx files Sony A95K.

    Sony BVM F170 is also my uploads.

    https://colorimetercorrections.displaycal.net/?get&type=*&manufacturer_id=SNY&instrument=*&html=1

    #38968

    nelldrip
    Participant
    • Offline

    I’ve finally finished implementing and evaluating a method that uses the synthetic profile editor to combine white point changes in the target color space,st2084/rec.2020.
    Many thanks for Vincent and EP98

    In conclusion, this method is preferable because it can eliminate correction with DisplayCal as much as possible, but it cannot be calibrated sufficiently with this alone.

    I don’t like the unnatural discrepancy in grayscale, especially at low luminance.
    In this state, it was absolutely unusable for grading, and 1D correction using HCFR was necessary.
    As Vincent points out, this may be a limitation of the current DisplayCal/Argyll CMS.

    Of course, I used madTPG’s HDR mode for TPG, and the 4400 self-made measurement grid include 255 points of grayscale as accurately as DisplayCal can accept.
    Nevertheless, the error was unacceptable for a only 100-point HCFR grayscale measurement.

    In this respect, the method using the compositing profile editor is a more complicated workflow, so it’s hard to say whether anyone can use it easily.
    It may be easier to output the standard value of D65 once and calibrate the white point mismatch from 1D.

    For reference, I attach the results measured with the LUT output by DisplayCal.
    The saturation map on the chromaticity diagram was quite satisfactory, but the grayscale is very strange.

    My goal is to make it easy for everyone to get the right HDR monitor display.
    Of course, it can also be applied to SDR monitors (limited to sRGB), and if you want an accurate sRGB monitor, you should use this method.
    In order to optimize the display of HDR monitors, it is best to first match the output to the ST2084/2020 standard values, and then apply a luminance map according to the monitor display capability.
    In the past, some monitors had poor saturation maps, in which case you may need to set a suitable rendering intent in the 3D LUT Creator.

    Attachments:
    You must be logged in to view attached files.
    #38972

    nelldrip
    Participant
    • Offline

    Whether the SDR is relative or absolute must be defined as absolute.
    I think it defined as standards.
    A good creator doesn’t do irresponsible things like brighten or darken depending on his mood, for video in SDR he uses 100cd/m2 and for sRGB he uses 80cd/m2.
    Of course, we know that many self-proclaimed creators who don’t know this are making stupid choices.
    As a result, I am deeply dissatisfied with the disastrous situation that irresponsible color management based on the ICC has resulted.
    *Surely, if there is no misunderstanding and it is operated perfectly, I think it is a very good system.

    Of course, this is also a problem in HDR, and as a result of grading with an HDR monitor that is not sufficiently calibrated, we often see glittering telops and unnatural images such as people’s shirts shining.
    This is the fault of a stupid colorist.

    I think the problem with ICC is that we might think that setting the ICC means we can calibrate.
    Unlike 3D LUT by DWM LUT, I doubt whether it is possible to calibrate saturation enough.
    Of course, it may be enough on a sRGB, but in reality, I think it is extremely doubtful whether it is managed and operated properly on current wide gamut displays.
    On this issue, Vincent, I think it’s very important to have someone like you who can give us the information to do the right thing.
    I think SDR PC monitors should be calibrated exactly to sRGB (0.2-80cd/m2) or BT.709 (0.1-100cd/m2, gamma2.2 or 2.4).
    Unfortunately, there are no further display criteria.
    The current confusion is the negligence of the monitor and still image industry in not sufficiently responding and disseminating information to the ICC.

    With HDR, I think customers need to see it properly. It doesn’t have to be exact, but authors should have a calibrated monitor that can display it correctly.
    I would like to devise and provide a method that is sufficiently inexpensive and easy to introduce.

    With the method I proposed this time, I succeeded to calibrate consumer monitor that supports a sufficient luminance range and a wide color gamut, in this case INNOCN’s 27M2V, etc., to same quality as a reference for HDR production.
    Even when compared with the PA32UCX, which also applies white point correction with the latest ASUS ProArt calibration software 4, the display is almost the same, and I am proud that I was able to realize an HDR reference environment that required least thousands of USD ever, for only about 1000 USD.

Viewing 9 posts - 271 through 279 (of 279 total)

You must be logged in to reply to this topic.

Log in or Register

Display Calibration and Characterization powered by ArgyllCMS