LG C8 Lut

Home Forums General Discussion LG C8 Lut

Viewing 15 posts - 256 through 270 (of 279 total)
  • Author
    Posts
  • #29289

    Josh Bendavid
    Participant
    • Offline

    You can validate the colors with DisplayCal against BT.2020 Gamma 2.2 in calibration mode, or against a synthetic profile with BT.2020 SMPTE 2084 outside of calibration mode, but the results are difficult to interpret because the validation test patterns will cover the full BT.2020 space, and a large fraction of points will be clipped (even if the TV is optimally calibrated).  At least you can compare the results between different scenarios.

    I opened a separate thread about this a little while ago here

    Generate testchart for DCI-P3 within BT.2020?

    #29292

    j82k
    Participant
    • Offline

    What I meant is if its possible to measure very specific colors? Like if I go with a color picker over the picture and lets say RGB 44 51 64 looks really weird and wrong in that scene. Is it possible to just validate manually picked colors?

    #29293

    Josh Bendavid
    Participant
    • Offline

    You can manually set the rgb values to be validated using the testchart editor in DisplayCal.  If you’re playing back video from your computer you can probably take a screenshot and then use the colour picker in GIMP or another photo editor.

    (I’m not sure this is a very systematic way to calibrate or verify though.)

    #29295

    chros
    Participant
    • Offline

    Here is the new 1DLUT with corrected EOTF but I really don’t think this will match your panel

    Cheers, I know just wanted to try it out, I modified the first 5 entries in it (reduced the values), no cyan/magenta issue, but the overall image is still brighter, probably I have to do it myself 🙂

    switching back and forth between presets and what I mostly see is that the synthetic 3DLUT desaturates cyan and in some parts that makes it a bit brighter. If I just watched it without comparing I wouldn’t say anything in that sample is elevated. Maybe it looks different on the B8

    I watched a full movie with it (ims) and the good thing is that I haven’t noticed anything outstanding (color issues).
    Sometimes I switched between the 2 presets: colors sometimes are different (obviously), but I lost a bit brightness and overall pop, image gets less vivid.

    I want to try out other synthetic settings first and compare it with my current one, maybe it will act differently.

    Don’t we know what Calman does with LighningLut (when measure only 5 colors, RGBWB) with HDR10? What color gamuts are used, rendering intent,  does it measure the colors in calibration mode?
    We would need to do the very same 🙂

    • This reply was modified 3 years ago by chros.
    #29297

    Josh Bendavid
    Participant
    • Offline

    Calman lightning LUT for HDR10 should be roughly equivalent to  curves+Matrix profiling in DisplayCal (but displaycal will still need to take more than 5 measurements to estimate the greyscale curves for the profile).   Measurements should be taken in calibration mode.

    Then the 3d LUT generation would correspond to BT.2020 gamma 2.2 (or unchanged).

    I don’t know how calman handles the clipping/rendering intent equivalent.  For a matrix calibration (in either Calman or DisplayCal) the colour primaries will be measured at peak brightness, which means that the matrix is only really valid at peak brightness, and any rendering intent which is applied is very likely to mess things up at lower brightness levels where primaries are more saturated, by applying stronger than necessary clipping and/or gamut compression.

    Given the fact that DisplayCal/Argyll cannot exploit the fact that content is limited to the DCI-P3 subset of the BT.2020 space when generating a 3d LUT for gamma 2.2, all of the appearance/perceptual rendering intents will always apply stronger-than-necessary gamut compression.

    So at least from first principles, one should try to do an XYZLUT profile in calibration mode,  and generate a 3d lut for BT.2020 gamma 2.2 with e.g. “absolute colorimetric with whitepoint scaling” for the rendering intent.

    This should work either with the DisplayCal calibration to generate the .Cal file (if it was working) or  with a manual greyscale calibration.

    #29300

    j82k
    Participant
    • Offline

    What should be chosen here when skipping calibration? (grayscale already calibrated manually)

    Attachments:
    You must be logged in to view attached files.
    #29302

    Josh Bendavid
    Participant
    • Offline

    If you already have uploaded the 1D LUT to the TV (and have uploaded a unity 3d lut), you should  check both boxes.

    #29311

    chros
    Participant
    • Offline

    Calman lightning LUT for HDR10 … I don’t know how calman handles the clipping/rendering intent equivalent.

    Well, we don’t have to think about it anymore (unless I screwed up something), yesterday I played with Calman using 2 cases:
    – technicolor: unity 1dlut + unity 1dlut (full ddc reset)
    – cinema: factory 1dlut + unity 3dlut: so I only created a matrix 3dlut on top of the factory 1dlut

    The result is worse (!) than with the built-in factory luts! Both have worse cyan/magenta issue (autocal 1dlut also brightens the image, as others stated). The color checker post calibration result is also bad for both. Interesting bit is that my first ims 3dlut was in place in technicolor (along with factory 1dlut) and the color checker pre calibration result was really good (for that 30 something colors)! 🙂
    That means Calman also applies absolute rendering intent.

    I also generated 8 different ims 3dluts, couple of results:
    – if I set 10000 for bt2020 and 4000 or 1000 for dic-p3 d65 then the image gets brighter
    – I didn’t notice much difference when I set 1000 for both bt2020 and dic-p3 d65 compared to 10000
    – I managed to get back some pop when I used gamma 2.23 for both bt2020 and dic-p3 d65 (along with 10000) (gamma 2.3 is already too much)
    — Sometimes the image still has veil/fog on top of it, I don’t know how to get rid of that

    I watched 2 episodes of The Expanse yesterday (s04e08 and 9, they have plenty of scenes that trigger the issue) with this new ims gamma 2.23 and so far that the best here: no more cyan/magenta issue (along with the fog sometimes) but the resulted image never looked better overall!

    Given the fact that DisplayCal/Argyll cannot exploit the fact that content is limited to the DCI-P3 subset of the BT.2020 space when generating a 3d LUT for gamma 2.2, all of the appearance/perceptual rendering intents will always apply stronger-than-necessary gamut compression.

    I agree, that’s why I don’t understand why j82k doesn’t see the desaturated colors in his set. So, in summary about rendering intents, and please correct me if I’m wrong:
    – absolute types: clips source colors outside the destination gamut, and that’s when we have the issue
    – relative: similar to absolute (clips), that’s why I saw only a slight change
    – perceptual types: compresses the source gamut into destination, that’s why we/I (?) get desaturated colors, this is where we don’t have issue
    – saturation: same as preserve-saturation (ims) but applies more saturation
    – preserve-saturation (ims): “uses three-dimensional compression and expansion to try and make the source gamut match the destination gamut, and also favours higher saturation over hue or lightness preservation”
    — so this is similar to perceptual but it saturates the compressed colors!

    So this tells me the 2018 models don’t like absolute rendering intent for whatever reason. What are your thoughts?


    @j82k
    , I have sent you a PM on doom9, cheers.

    • This reply was modified 3 years ago by chros.
    #29313

    Josh Bendavid
    Participant
    • Offline

    – if I set 10000 for bt2020 and 4000 or 1000 for dic-p3 d65 then the image gets brighter
    – I didn’t notice much difference when I set 1000 for both bt2020 and dic-p3 d65 compared to 10000

    Sorry not sure exactly which settings you’re referring to.

    Yes I think your summary of the rendering intents is mostly correct (relative vs absolute matters in case the greyscale/whitepoint is not calibrated first)

    I don’t see why clipping should cause problems per se.  So if something is really problematic here maybe it’s on the DisplayCal/Argyll side.  Of course the exact appearance of the clipping will depend very much on the exact colourspace coverage of the display.

    But ok it sounds like we might benefit a lot from having the ability to apply this “content colourspace” setting when calibrating to pure gamma curves (currently only available when generating 3d luts explicitly for SMPTE 2084).

    #29314

    chros
    Participant
    • Offline

    Sorry not sure exactly which settings you’re referring to.

    Create synthetic ICC profile window: White level value.

    I don’t see why clipping should cause problems per se.

    Me neither, but that’s the case with our 2018 models.
    Btw, do you also see the issue on yours? (I don’t remember)

    So if something is really problematic here maybe it’s on the DisplayCal/Argyll side.

    I’m not sure: that is with factory 3dlut and Calman matrix lut as well. I think it’s more like LG 2018 bt2020 gamut processing.

    #30093

    quietvoid
    Participant
    • Offline

    Hi, I just did a 1D LUT for HDR and made changes for it to track PQ better (for me this was brightening 40%+ by around 1.5%).

    Now I’m confused on what to do for the 3D LUT, am I supposed to do a synthetic LUT to restore “proper” colors? Or to do profiling?

    I’m also not sure if verification is possible. Any advice would be helpful.

    • This reply was modified 2 years, 10 months ago by quietvoid.
    #30095

    quietvoid
    Participant
    • Offline

    Sorry for the double post but just updating, apparently I can’t edit more.
    I ended up not needing to adjust the 1D LUT, since it introduced some green banding and the 3D LUT seems to have fixed the PQ tracking.
    Did a BT2020 2.2 synthetic to DCI-P3 2.23, works well.
    Compared to the factory 3D LUT, it’s desaturated. But comparing with Dolby Vision, the new 3D LUT looks very similar.

    However now the blacks are elevated slightly, and I don’t think it’s fixable because it happens right after the ddc_reset.
    So no matter the LUTs uploaded, black pixels aren’t turned off.

    #30194

    chros
    Participant
    • Offline

    So what’s you final conclusion/question? 🙂 (You have written multiple places).

    I created a 1dlut for the factory 3dlut and that’s what I use now: it doesn’t fix the magenta/cyan issue, but removes the fog/veil in darker scenes while also introduce slight black crush 🙂

    #30195

    quietvoid
    Participant
    • Offline

    My question is if there’s a way to restore my true blacks, after uploading a 1D LUT.
    I’ve yet to try uploading the 1D LUT without doing a ddc_reset first.

    #30196

    chros
    Participant
    • Offline

    That’s strange, I don’t get raise blacks on B8 with old firmware 4.10.25 no matter what I do. Yes, try to just upload the 1dlut and don’t do anything else, maybe…

    I have tried different 3dluts, and I tried a lot, but didn’t like any of them. (Uploaded the same 1dlut to Cinema and Technicolor preset, where the latter has the factory 3dlut, and switch between the 2).

Viewing 15 posts - 256 through 270 (of 279 total)

You must be logged in to reply to this topic.

Log in or Register

Display Calibration and Characterization powered by ArgyllCMS