LG C8 Lut

Home Forums General Discussion LG C8 Lut

Viewing 15 posts - 106 through 120 (of 213 total)
  • Author
    Posts
  • #21446

    chros
    Participant
    • Offline

    The other good news is that the color gamut matrices are supported on the 2018 sets as well.

    I’ve added additional functions to set these set_bt709_3by3_gamut_data(picMode, data) and set_bt2020_3by3_gamut_data(picMode, data), where data is a 3×3 numpy array with dtype=np.float32. and of course the default is the identity matrix.  Most likely the dolby vision modes are using the bt709 matrix as for the 3d luts.

    (Is there some file input which would be convenient to implement for these matrices?)

    How come the DV would only use bt.709 data? It doesn’t make any sense.

    I forgot to add: it would be nice to implement that matrix calibration method. CalMAN always uploads a 3DLUT (even when it’s done by running the numbers though a matrix to fill the cube). I am interested to see how a 1DLUT calibration + a real matrix behaves compared to a 1DLUT calibration + a 3DLUT filled from a matrix in HDR10 mode where the CalMAN 1D+3D result has some magenta issues.

    I’ve never heard of this (only matrix), is it initially only supposed to support the 2019 models?

    New HDR10 DisplayCal settings for factory 1dlut:
    – no more fog-layer
    – image is not darkened

    I couldn’t figure out how to measure just 5 patches (WBRGB, as Calman does), but this result is way better than the previous was.

    Still HDR10 profiling (old firmware, no black crush/dithering fix):
    – I noticed that the above testchart (gamma+matrix) resulted in only ~75% DCI-P3 coverage
    – Janos told me, that we should use XYZ instead (without matrix)

    1. factory 1dlut + user matrix filled 3dlut (XYZ testchart)
      – I managed to create a WBRGB testchart by modifying the Testchart for colorimater correction: duplicate the W line, then delete the values from the duplicated W line -> it will result in black (the entry for it in ti1 file: 0 0 0 1 1 1)
      – deselected “Embed calibration curve …”
      – deselected “Apply calibration (vcgt)” on 3dlut page
      – with this DisplayCal could create a 3dlut: resulted in ~96% DCI-P3 coverage

      Unfortunately I’m not sure I like the result: the image is not that vidid as with an uncalibrated mode (no fog layer, that’s good).

    2. unity 1dlut + user 3dlut (XYZ testchart)
      – I used Janos 68 patches testchart for this
      – selected “Embed calibration curve …”
      – selected “Apply calibration (vcgt)” on 3dlut page
      – rest of the settings are the same as above: resulted in the same coverage

      The result: the whole image image is much darker than the uncalibrated mode and it crushes some dark shadows as well (confirmed with Mechanik HDR10 black clipping patterns).
      Maybe I can set a lower gamma (e.g. 2.1, 2.0) on the 3dlut page and see the result.

    I also checked how the peak brightness has been changed in the 2 profiled preset from the original 690 nits (1% window): both resulted in ~670 nits, so there’s “only” a ~20 nits degradation.

    #21447

    János Tóth F.
    Participant
    • Offline

    >How come the DV would only use bt.709 data? It doesn’t make any sense.

    In DV mode, the DV chip shall be allowed to handle all the 3D color mapping (dynamic metadata based tone mapping and color space conversion), so any such processing should be bypassed in the TV’s processor. It looks like it was more convenient for LG to leave these steps (like the 3DLUT processing) in the chain and use (most probably) neutral factory data. (May they also considered the possibility that they will need to correct some unexpected  color errors on lower levels later on with firmware updates through the 3DLUT.) When the 3DLUT does nothing it doesn’t really matter how you label it’s slot. In any case, if we decide to try and use that DV 3DLUT then the target color space should theoretically match the measured primaries (neither Re709, nor Rec2020 and not even DCI-P3), so that the DV chip can work as intended (I personally opt for the alternative and use Rec2020 target but only to keep the HDR10 and DV modes in closer sync).

    >I’ve never heard of this (only matrix), is it initially only supposed to support the 2019 models?

    Indeed. I thought this was added in 2019 but read @josh-2 ‘s posts… (although I am not sure whether he checked if this is actually functional on either ’18 or ’19 models, or just a placeholder for planned future / disabled debugging feature).

    New HDR10 DisplayCal settings for factory 1dlut:
    – no more fog-layer
    – image is not darkened

    I couldn’t figure out how to measure just 5 patches (WBRGB, as Calman does), but this result is way better than the previous was.

    I repeat: 5 patches just won’t cut it with colprof (the ArgyllCMS tool behind DisplayCAL). It has no clue about your gamma with no points to fit curves on (black and white gives you a line, not a curve and the native response is not a line).

    Well… may be you should try a quick 1DLUT calibration then to see if the smallet 17^3 3DLUT is the issue with uncalibrated 1D response: run dispcal (D65 white, gamma 2.2 target, default high speed mode, drop to Very High is it would take more than ~10 minutes), upload the 1DLUT, do the profiling and upload the 3DLUT.

    Also note that Game mode (I believe you are experimenting in Game mode — I did so myself because it’s faster to measure due to it’s lower latency and didn’t have much value in factory state for me anyways) had significantly reduced (about 100 nits lower) peak white luminance in both HDR10 and DV modes on my C8 (compared to Cinema modes).

    #21450

    Josh Bendavid
    Participant
    • Offline

    I did confirm that the additional matrices are active on the 2018 sets (by swapping the matrix rows and enjoying the messed up colors 🙂

    #21451

    chros
    Participant
    • Offline

    Also note that Game mode (I believe you are experimenting in Game mode)

    No, I ended up using:
    – technicolor for: factory 1dlut + matrix filled user 3dlut
    – cinema for: unity 1dlut + “full” user 3dlut
    – set up cinema home for: as cinema was before experimenting (so I can easily switch between the 3)

    I repeat: 5 patches just won’t cut it with colprof (the ArgyllCMS tool behind DisplayCAL). It has no clue about your gamma with no points to fit curves on (black and white gives you a line, not a curve and the native response is not a line).

    I understand, but I used this method with factory 1dlut (technicolor preset). Do you say that it’s still wrong?

    Well… may be you should try a quick 1DLUT calibration then to see if the smallet 17^3 3DLUT is the issue with uncalibrated 1D response: run dispcal (D65 white, gamma 2.2 target, default high speed mode, drop to Very High is it would take more than ~10 minutes), upload the 1DLUT, do the profiling and upload the 3DLUT.

    I don’t think it should do anything with the lut size: it’s like the end result was with higher gamma (Cinema preset: unity 1dlut + “full” user 3dlut).
    Maybe I’ll wait until you or others will be able to try HDR10 profiling out with Josh’s util 🙂

    • This reply was modified 8 months, 1 week ago by chros.
    #21456

    Josh Bendavid
    Participant
    • Offline

    In DV mode, the DV chip shall be allowed to handle all the 3D color mapping (dynamic metadata based tone mapping and color space conversion), so any such processing should be bypassed in the TV’s processor. It looks like it was more convenient for LG to leave these steps (like the 3DLUT processing) in the chain and use (most probably) neutral factory data. (May they also considered the possibility that they will need to correct some unexpected  color errors on lower levels later on with firmware updates through the 3DLUT.) When the 3DLUT does nothing it doesn’t really matter how you label it’s slot. In any case, if we decide to try and use that DV 3DLUT then the target color space should theoretically match the measured primaries (neither Re709, nor Rec2020 and not even DCI-P3), so that the DV chip can work as intended (I personally opt for the alternative and use Rec2020 target but only to keep the HDR10 and DV modes in closer sync).

    Anyone have a pointer to to how the TLMS2RGB matrix should be constructed given the desired set of target primaries?

    #21458

    chros
    Participant
    • Offline
    1. factory 1dlut + user matrix filled 3dlut (XYZ testchart)
      – I managed to create a WBRGB testchart by modifying the Testchart for colorimater correction: duplicate the W line, then delete the values from the duplicated W line -> it will result in black (the entry for it in ti1 file: 0 0 0 1 1 1)
      – deselected “Embed calibration curve …”
      – deselected “Apply calibration (vcgt)” on 3dlut page
      – with this DisplayCal could create a 3dlut: resulted in ~96% DCI-P3 coverageUnfortunately I’m not sure I like the result: the image is not that vidid as with an uncalibrated mode (no fog layer, that’s good).

    Here are the steps I did for “factory 1dlut + user matrix filled 3dlut (XYZ testchart)” (Technicolor preset):

    aiopylgtvcommand 192.168.1.78 start_calibration hdr_technicolorExpert
    aiopylgtvcommand 192.168.1.78 upload_3d_lut_bt2020 hdr_technicolorExpert
    aiopylgtvcommand 192.168.1.78 upload_3d_lut_bt2020_from_file hdr_technicolorExpert "b8bt2020g22p17t4p5.cube"
    aiopylgtvcommand 192.168.1.78 end_calibration hdr_technicolorExpert

    Profiling was done after the 2nd command (after uploading 3dlut unity), and here are steps:
    – you can download the WBRGB patch set I used (copy into “c:\Program Files (x86)\DisplayCAL\ti1\” directory)
    – I don’t think I like the results 🙂

    1. 2. unity 1dlut + user 3dlut (XYZ testchart)
      – I used Janos 68 patches testchart for this
      – selected “Embed calibration curve …”
      – selected “Apply calibration (vcgt)” on 3dlut page
      – rest of the settings are the same as above: resulted in the same coverageThe result: the whole image image is much darker than the uncalibrated mode and it crushes some dark shadows as well (confirmed with Mechanik HDR10 black clipping patterns).
      Maybe I can set a lower gamma (e.g. 2.1, 2.0) on the 3dlut page and see the result.

    Here are the steps I did for “unity 1dlut + user 3dlut (XYZ testchart)” (Cinema preset):

    aiopylgtvcommand 192.168.1.78 start_calibration hdr_cinema
    aiopylgtvcommand 192.168.1.78 upload_1d_lut hdr_cinema
    aiopylgtvcommand 192.168.1.78 upload_3d_lut_bt2020 hdr_cinema
    aiopylgtvcommand 192.168.1.78 upload_3d_lut_bt2020_from_file hdr_cinema "b8bt2020g213p17no1dt1.cube"
    aiopylgtvcommand 192.168.1.78 end_calibration hdr_cinema
    

    Profiling was done after the 3rd command (after uploading 1dlut and 3dlut unity), and here are steps:
    – note that (as I mentioned) I had to lower gamma (to 2.13 on 3dlut page) to get around the same brightness overall, but it’s still darker in dark scenes
    – I don’t think I like the results either 🙂

    3. What I noticed during experimenting:
    – there’s no factory 3dlut by default?
    — Technicolor preset has factory 1dlut, I uploaded a unity 3dlut to it and it looked the same as Ciname Home preset (that we can’t modify)
    – the unity 1dlut results in different color temperature (even when user 3dlut is present):
    — open up the menu and take a look at the color of fonts: it’s much warmer with factory 1dlut (and menu set to warm2)

    • This reply was modified 8 months, 1 week ago by chros.
    #21460

    Josh Bendavid
    Participant
    • Offline

    My understanding of the behaviour is that the 3d lut does not affect the tv menu/GUI (and I think also the GUI in smart tv apps)

    #21468

    chros
    Participant
    • Offline

    My understanding of the behaviour is that the 3d lut does not affect the tv menu/GUI (and I think also the GUI in smart tv apps)

    Thanks, probably that’s the case, but what was strange for me that unity 1dlut disables (?) the color temperature completely. (I don’t mind this at all, just mentioned it.)

    I watched half a movie yesterday with “unity 1dlut + user 3dlut (XYZ testchart)” (Cinema preset) and sometimes I switched between the unprofiled Cinema Home preset, and I noticed (in addition to the above notes):
    – profiled Cinema preset removes some veil/fog from the image 🙂

    • This reply was modified 8 months, 1 week ago by chros.
    #21479

    János Tóth F.
    Participant
    • Offline

    3. What I noticed during experimenting:
    – there’s no factory 3dlut by default?
    — Technicolor preset has factory 1dlut, I uploaded a unity 3dlut to it and it looked the same as Ciname Home preset (that we can’t modify)
    – the unity 1dlut results in different color temperature (even when user 3dlut is present):
    — open up the menu and take a look at the color of fonts: it’s much warmer with factory 1dlut (and menu set to warm2)

    May be factory modes work with neutral 3DLUT and use a custom 3×3 matrix instead.

    Even if the factory 1DLUT is fully neutral (I don’t think it is, at least not since the “macro block” fix), touching it by calibration commands switches to a neutral white point preset (the factory cool/normal/warm white point settings found in the EZ-ADJUST menu become inactive and you get to start the 1DLUT calibration with  a fully clean slate). So, the white point will always change when you (first) touch the 1DLUT (unless you had set every Gain vaues to 192 in the EZ-ADJUST menu beforehand). It would make sense if accurate factory modes normally used a neutral 1DLUT and it was only used to manipulate fancy picture modes (like DV Cinema Home which is brighter than DV Cinema) and/or doing things like the first “macro block fix” (crushing the dark grays to hide flashing errors).

    The 3DLUT (neither factory, nor custom) shouldn’t normally change the white balance if the 1DLUT was calibrated (or change the white point if the 1DLUT wasn’t re-programmed but the white point was set in  EZ-ADJUST or normal user menu Gain controls).

    The OSD is unaffected by the 3DLUT but affected by the 1DLUT or EZ-ADJUST settings. I think this is why we can’t have a test pattern generator from the WebOS Store for 2018 models (not like the 2019 iTPG is good for DV validation but it’s an excuse).

    #21493

    chros
    Participant
    • Offline

    May be factory modes work with neutral 3DLUT and use a custom 3×3 matrix instead.

    Hm, interesting thought. I’m getting run out of presets then to experiment with 🙂

    The 3DLUT (neither factory, nor custom) shouldn’t normally change the white balance if the 1DLUT was calibrated (or change the white point if the 1DLUT wasn’t re-programmed but the white point was set in  EZ-ADJUST or normal user menu Gain controls).

    Yes, It doesn’t change it, if the proper WP was used in the 3dlut.

    The OSD is unaffected by the 3DLUT but affected by the 1DLUT or EZ-ADJUST settings.

    Cheers, it makes sense.

    Even if the factory 1DLUT is fully neutral (I don’t think it is,

    It can’t be neutral: otherwise I wouldn’t have different results between Cinema Home and Cinema presets.

    touching it by calibration commands switches to a neutral white point preset (the factory cool/normal/warm white point settings found in the EZ-ADJUST menu become inactive

    That’s also interesting, I haven’t thought of that before: why would the (global!) EZ-ADJUST menu become inactive if we only (!) modify a single (!) preset? 🙂

    It would make sense if accurate factory modes normally used a neutral 1DLUT and it was only used to manipulate fancy picture modes

    But that’s definitely not the case: Cinema Home and Technicolor (factury 1dlut + unity 3dlut)  presets look the same.

    • This reply was modified 8 months, 1 week ago by chros.
    #21495

    stama
    Participant
    • Offline

    The 1D LUT controls the whitepoint setting, indeed. When you set the 1D LUT, you are setting the R, G and B ramps, which are moving the coordinates of the whitepoint inside the gamut. And if you also set a unity 3D LUT, then you have the native primaries of the panel (supposedly almost the DCI-P3 primaries), and now the whitepoint is the native panel whitepoint located around ~10000K.

    My experience trying to calibrate the 1D LUT in HDR with Calman is that it’s unreliable. The panel is glowing after some white patches are displayed on the screen (visible in a completely dark room). So, when you try to display a pure black patch in such conditions, you are not going to get no light emitted from the panel, and I imagine the other patches are compromised as well. And Calman’s Autocal does start with bright gray patches, and moves from there towards the darker end of the gray ramp, so by the time you are next to the really dark patches, the panel is compromised. ABL is also engaged by then from time to time. Since this seems to be an instability of the OLED panel, I don’t know how you can achieve better results. Maybe that’s why everybody recommends to just set the 3D LUT for HDR, and let the gray ramp alone.

    Regarding 0.45: when storing data in 8 bits, recording devices don’t allocate 50% of the range allowed by the 8 bits per color channel to lower light intensities, and 50% to higher light intensities. That’s because the human visual perception is not able to distinguish between small higher light intensities,  while it’s able to distinguish very easily between small lower light intensities. So, the range of numbers allowed by 8 bits are not spread evenly between the lower and higher light intensities. More of the 8 bit numbers are allocated to lower light intensities, and fewer to higher light intensities. This is done by multiplying the recorded luminance with a power function that happens to have the exponent 0.45 (aka = 1/2.2). This is the “Rec. 709 transfer function of video”. The power function with exponent 0.45 is the “encoding gamma” (it’s not a pure power function though, it has a small linear segment near 0). The encoding gamma is not the same thing as the “decoding gamma” which is the transform that a display applies to the encoded luminance values. When we calibrate the TV, we are configuring the luminance transform that is applied by the TV to these encoded values, the “display gamma”. The Rec 709 standard did not define what should be the display gamma, only what is the encoding gamma. Later publications recommended a gamma of 2.4 to be applied by displays (though for a large amount of time 2.2 was what everybody was using, and when CRTs were the display tech these were supposed to have a 2.5 display gamma).

    I don’t know what is the relationship with the flag that can be set on enabling calibration, but thought I might add this explanation.

    Charles Poynton, which is the guy that is usually referenced when speaking about video encoding and color spaces, wrote two books about this, and I found one of his earlier drafts speaking about gamma here.

    #21499

    János Tóth F.
    Participant
    • Offline

    That’s also interesting, I haven’t thought of that before: why would the (global!) EZ-ADJUST menu become inactive if we only (!) modify a single (!) preset? 🙂

    In other worlds, when you touch the 1DLUT of a picture mode, that picture mode effectively switches to an imaginary “Native” EZ-ADJUST white point preset which is the same as if you manually adjusted “warm” (or cool, or medium) to 192-192-192 in the EZ-ADJUST menu because this is the optimal way of 1DLUT calibration: a redundant processing step (the 2 point white balance in EZ-ADJUST) is bypassed and the 1DLUT can start from scratch. Otherwise, depending on the EZ-ADJUST settings and your 1DLUT calibration target, the 1DLUT calibration could be locked out some of the available dynamic range. Say, WARM is 192-120-100 and CalMAN decides (based on sensor data and your target WP) that you need significantly more Green. But you can’t increase Green with the 1DLUT, so you have to decrease both Red and Blue instead. And I guess every single procession step which is not a no-op/bypass might degrades the image quality for some small degree, so it’s better to have as few active steps as possible and that’s why there is no “custom” preset in EZ-ADJUST for starting a 1DLUT calibration with setting the 100% white point with those kinds of controls. But even then, after you set up 100%, you might still face a similar issue at other points: where the 100% 2-point control decreased a component at, say, 90% and now the 1DLUT has to increase it, so they effectively work against each other, potentially causing rounding errors in the process.

    But that’s definitely not the case: Cinema Home and Technicolor (factury 1dlut + unity 3dlut)  presets look the same.

    It can’t be neutral: otherwise I wouldn’t have different results between Cinema Home and Cinema presets

    Sorry, I probably lost track of the state of your test cases. You say X and Y looks the same in an argument that they must have different 1DLUTs? :O

    I suggest you to re-evaluate your current theories based on your current test cases with the information of the WP presets above and why a theoretical neutral factory 1DLUT is different from a user uploaded neutral 1DLUT (due to how the EZ-ADJUST presets are utilized).

    #21561

    chros
    Participant
    • Offline

    My experience trying to calibrate the 1D LUT in HDR with Calman is that it’s unreliable. The panel is glowing after some white patches are displayed on the screen (visible in a completely dark room). So, when you try to display a pure black patch in such conditions, you are not going to get no light emitted from the panel, and I imagine the other patches are compromised as well.

    Can’t Calman do BFI? (DisplayCal can, you have to enable advanced settings, e.g. with: 0, 20, 0) There are guys who used some insane settings like 6 minutes with LS profiling SDR (!) 🙂

    In other worlds, when you touch the 1DLUT of a picture mode, that picture mode effectively switches to an imaginary “Native” EZ-ADJUST white point preset …

    OK, cheers, I misunderstood the concept, now it’s clear (that’s how I thought beforehand).

    Sorry, I probably lost track of the state of your test cases.

    🙂 So, one more time:

    I ended up using (so I can easily switch between the 3):
    – technicolor: factory 1dlut + matrix filled user 3dlut
    – cinema: unity 1dlut + “full” user 3dlut
    – cinema home: as cinema was before experimenting

    You say X and Y looks the same in an argument that they must have different 1DLUTs?

    🙂 You’ve put together 2 things.

    First:
    – you said: “It would make sense if accurate factory modes normally used a neutral 1DLUT and it was only used to manipulate fancy picture modes”
    – I replied: “But that’s definitely not the case: Cinema Home and Technicolor (factory 1dlut + unity 3dlut)  presets look the same.”
    – I mentioned Cinema Home in the sense of “accurate factory mode” (after setting Color Temp to W50): that means that every preset has a factory 1dlut

    Second:
    – you said: “Even if the factory 1DLUT is fully neutral (I don’t think it is,”
    – I replied: “It can’t be neutral: otherwise I wouldn’t have different results between Cinema Home and Cinema presets.”
    – that means: that you are right, factory 1DLUTs are not neutral

    So, in summary: every preset has a factory 1dlut in use that is not neutral. 🙂

    I suggest you to re-evaluate your current theories …

    Thanks, but there’s no need for that: that’s how I thought initially, so they are correct.

    #21562

    chros
    Participant
    • Offline

    I watched half a movie yesterday with “unity 1dlut + user 3dlut (XYZ testchart)” (Cinema preset) and sometimes I switched between the unprofiled Cinema Home preset, and I noticed (in addition to the above notes):
    – profiled Cinema preset removes some veil/fog from the image 🙂

    I can confirm this, after watching couple of contents with it, in half of the scenes are clearly visible.
    It’s even obvious when I start an SDR content straight after watching an HDR one with Cinema preset: that veil is all over the image.

    I have couple of question about the 2. second method I used (thanks to Janos) for the 2nd test case (Cinema mode) (I uploaded the p68 patchset as well):

    The 1D LUT controls the whitepoint setting, indeed. When you set the 1D LUT, you are setting the R, G and B ramps, which are moving the coordinates of the whitepoint inside the gamut. And if you also set a unity 3D LUT, then you have the native primaries of the panel (supposedly almost the DCI-P3 primaries), and now the whitepoint is the native panel whitepoint located around ~10000K.

    Q1: so, before profiling we upload unity 1dlut and unity 3dlut and that results in native WP. But the settings I used don’t take into account the D65 WP, do they? I mean:
    – WP was set as measured under “Calibration settings”
    – there’s no WP setting on 3dlut tab

    So how come that the final 3dlut is still look close the Warm 2?

    Q2: as I mentioned I had to raise gamma to 2.13 to get similar brightness (although still a bit black crush is present) as with Cinema Home.
    – I set gamma on 3dlut tab
    – but the display still applies tonemapping in calibration mode

    So, why is Gamma setting is the correct one and not PQ on 3dlut tab?
    (Of course if PQ should be used, as Ted mentions it, it’d raise even more questions than it answers 🙂 )


    @Josh
    , have you tried to profile HDR mode?


    @Florian
    , what settings would you use for that?

    • This reply was modified 8 months, 1 week ago by chros.
    #21570

    János Tóth F.
    Participant
    • Offline

    First:
    – you said: “It would make sense if accurate factory modes normally used a neutral 1DLUT and it was only used to manipulate fancy picture modes”
    – I replied: “But that’s definitely not the case: Cinema Home and Technicolor (factory 1dlut + unity 3dlut)  presets look the same.”
    – I mentioned Cinema Home in the sense of “accurate factory mode” (after setting Color Temp to W50): that means that every preset has a factory 1dlut

    Note that Home is accurate in HDR10 but it’s obviously not in DolbyVison (this stands for both the C8 and the C9). I don’t have a Home in SDR on the C9 and I can’t remember if there was one on the C8.

    So, ‘Home Cinema’ and ‘Cinema’ looking the same (with synced user menu settings) in HDR10 doesn’t mean anything. They both use the same 1DLUT and that 1DLUT could be either neutral or non-neutral. But DolbyVision  ‘Home Cinema’ most probably uses  a non-neutral 1DLUT (I never tried to calibrate this preset but DV Game is similar and that one becomes accurate after a 1DLUT calibration).

    Second:
    – you said: “Even if the factory 1DLUT is fully neutral (I don’t think it is,”
    – I replied: “It can’t be neutral: otherwise I wouldn’t have different results between Cinema Home and Cinema presets.”
    – that means: that you are right, factory 1DLUTs are not neutral

    So, in summary: every preset has a factory 1dlut in use that is not neutral. 🙂

    Sorry but I still don’t see any proof of that. Not here, not elsewhere. I think I know how to dives the answer though. The C9 probably has no TRC alteration in the factory 1DLUT to hide the flashing errors and it doesn’t choose different dithering methods based on ‘calibration flags’ (I can’t be sure of that though, it’s just my guess), so I should see the same tone response between Cinema with factory 1DLUT set to Cool (and Cool set to 192-192-192 in EZ-ADJUST) and Cinema with a nutral 1DLUT uploaded with CalMAN. I will check this when I have time. I don’t think you made a similar comparison (you didn’t alter EZ-ADJUST, at least not to neutral WP settings). -> I hate to ask but it would be best if you did this test, since you have a fully pre-macroblockfix state (I don’t).

    Q1: so, before profiling we upload unity 1dlut and unity 3dlut and that results in native WP. But the settings I used don’t take into account the D65 WP, do they? I mean:
    – WP was set as measured under “Calibration settings”
    – there’s no WP setting on 3dlut tab

    So how come that the final 3dlut is still look close the Warm 2?

    Q2: as I mentioned I had to raise gamma to 2.13 to get similar brightness (although still a bit black crush is present) as with Cinema Home.
    – I set gamma on 3dlut tab
    – but the display still applies tonemapping in calibration mode

    So, why is Gamma setting is the correct one and not PQ on 3dlut tab?
    (Of course if PQ should be used, as Ted mentions it, it’d raise even more questions than it answers 🙂 )

    Ted misunderstood my intentions. He thought I wish to permanently disable the math based tone-mapping and do it with the 3DLUT. I knew that won’t work. And he suggested I should do this in SDR mode where there is no HDR tome mapping (but that has lower peak white).

    The math based HDR10 tone mapping IS disabled in calibration mode. At least it should be. In this case, you calibrate the ~P3 ~10000K ~gamma 2.2 panel response with the 3DLUT to Rec2020, ~6500K and gamma 2.2. The factory defaults probably use a 3×3 matrix (or the 3DLUT) for Rec2020, EZ-ADJUST for ~6500K and the 1DLUT (or probably nothing from this processing group) for gamma 2.2 (the display panel already has a nominal gamma 2.2 response without any LUTs or other SoC side processing).

    Your 3DLUT targets D65 because the Rec2020 profile has a profile white point of D65. That’s why you have to give the NO answer after ticking the “would you like to use the measured white instead” question.

    I have no idea why you end up with incorrect gamma other than your profile doesn’t manage to model the display’s response or DisplayCAL fails to devise the proper corrections between the profiles (or the LG hardware misbehaves: rounding errors and format based drifts and such and such). It’s probably the display profile due to the WRGB magic with the “W boost”. But I am not sure why it seemed to work fine for me. I will try to calibrate the C9 this way tomorrow and show you screenshots with a validation report.

Viewing 15 posts - 106 through 120 (of 213 total)

You must be logged in to reply to this topic.

Log in or Register

Display Calibration and Characterization powered by ArgyllCMS