LG OLED 3D LUT Profiling for SDR with High Peak Output
Usually LG users perform SDR 3D LUT profiling using CalMAN or LightSpace for 100 nits, since its the peak output levels the SDR movies has been mastered.
As we know the LG's 2018/2019 models don't have ABL for up to ~150 nits peak output, so when you are profiling for 100 nits, there will be no problem during any SDR movie content playback related to frame APL which can enable the ABL limiter.
When you profile for higher nits, 250 nits for example, while during patch measurements when you will use 10% Windows it will not enable the ABL function, it can happen in movie content frames with high APL levels the ABL limiter to be enabled.
Since for having an accurate/reference level picture we don't want to have enabled any local dimming/dynamic function or any ABL limiter; ase these functions can deviate from the calibration it has been performed to any display, its better idea when maximum accuracy required to profile the display to the nits level which will not enable any of these functions, as we talking for LG's, maximum peak is safe to be up to 150 nits.
There some theories that LG's are not stable with high peak output from Tyler and that i1Display PRO is not recommended meter for large profiling due to drift from John:
''The panel is much less stable at 250 nits vs 100 nits. I usually do lightning LUT for Day mode and larger iRP for night mode.
''I do these sets every day and I never have a problem. Best to use ISF Bright with default settings, if you want the brightest picture make sure gamma is set to 0 and you can set the OLED light to 100, then set the screen insertion to 15/5/15, do a 1D LUT then a 3,000 pt 3D LUT and see how the results look. Your deltaE should be below 0.5 for grayscale and the CMS should be in the range of around DeltaE of 1ish.
MY guess is your panel is drifting trying to do a 10,000 pt LUT especially with an i1Display Pro. One thing to remember, although you can do large LUTs with lower cost meters, you really need a fast meter like a Klein to reliably do these large LUTs or set you screen insertion to a very short interval to help keep the panel from drifting.
I asked both of them if they have any data to provide to see how they have find out that panels are drifting at 250 nits before some days
, so its was time to bring some real data to see how these panels actually drift with high brightness and if its possible to perform a successful profiling also.
There two users in AVSForum with LG OLED's which are both using i1Display PRO colorimeter for their 3D LUT profiling and they have performed a normal ~100 nits but a high-brightness profiling with ~250 nits also, so I will use these two data.
For reference, we have the user ''HiFi4Vision'' which have posted his results from using IR Profiling with CalMAN (100 nits and 250 nits) with his i1Display PRO here
, and the user ''ebr9999'' which have performed display characterization using LightSpace with his i1Display PRO (100 nits here
and 277 nits here
User ''HiFi4Vision'' has verified his 100 nits IR Profiling performed by CalMAN using ColorChecker Classic measurements (24 patches only for verification) and he had average 0.6 dE2000 with max 0.9dE2000 with 100 nits IR Profiling, while for his 250 nits IR Profiling he had average 1.1dE2000 and max 2.2dE2000.
The user ''ebr9999'' has verified his 129 nits 3D LUT profiling performed by LightSpace using 10-Point Cube measurements (1000 color patches...which is the perfect way to evaluate a display performance volumetrically) and he had average 0.29dE2000 with max 0.9dE2000 while for his 277 nits profiling he had average of 0.37de2000 and max 1.31 dE2000.
According to Tyler, the OLED panels is not so stable at 250 nits but for that reason CalMAN has added more intelligence to their drift compensation algorithm
, to prevent and count any drifting during the measurements time.
The problem with CalMAN is that the patchset when you are using IR Profiling is dynamically adjusted, in other words, if you perform 5x IR Profilings with CalMAN to the same display with the same pre-cal settings, the patchsets used per each time it will not be exact the same, this is not helping the user to see if different meter or patch generation setting he used, how the are helping or not. For example to find out how different values of patch delay before meter read or how the black frame insertion settings affect the end results.
It will be good for all CalMAN users it if can be added a drift plot chart; as the drift comp is active to all IR profilings and there no way to disable it; this will help the users to understand how different settings affect the panel stability etc.
The LightSpace user for example used the same meter/patch generation settings and the same patchset (so fixed patch order with fixed patch RGB triplets), so its easy looking his drift plot data to understand how the panel is drifting between 129 / 277 nits.
Lets start with Drift Plot of 129 nits:
The axis of luminance deviations is +-3 nits, the RGB lines show the chromaticity changes over the 5000 patch measurements, since LightSpace took one white patch sample after 50 other measurements patches (the user selected per 50, as there capability to set whatever values you like..like 100, 30 etc.)
He is the plot of 277 nits:
If you exclude the initial drop, because the user seems that he hadn't used a pre-roll patchset before running the actual profiling, there no significant changes to affect the profiling so much, such kind of drift deviations generally are small.
Since the CalMAN user had posted his CalMAN files and I had also the LightSpace user files also, I have import them all to LightSpace to evaluate using LightSpace Tools what is happening with Cube Viewer / 1D LUT Viewer and LUT Preview.
Starting from Cube viewer, lets compare using animated PNG files the generated cube's between LightSpace - CalMAN, while both users had i1Display PRO:
Looking the 1D LUT Viewer, as usual, the known problem witb CalMAN, is the reduced peak output visible if you look the right top corner.
The RGB lines look closer with CalMAN as the user had already pre-calibrated the 100% White before performed IR Profiling.
The LightSpace lines have a spacing because panel was more uncalibrated, so the red line for example is lower compared to other 2 lines as display had more red to its 100% White, so red channel reduced more to fix RGB balance.
Now looking the LUT Preview, 2 reference images are loaded and the 2 LUTs are applyed to these images, the will be pop-up the clipping to blue for example:
(The ''SOFS7_SE'' is the LightSpace file, while the ''SDR Expert Bright IR 10000'' is the CalMAN file)
..or the yellow + blue clipping more noticed:
Now lets compare the 4 files, CalMAN 100 nits, LightSpace 129 nits, CalMAN 250 nits and LightSpace 277 nits.
The summary of that test is that i1Display PRO and the LG OLED has no problem to be profiled for 277 nits, with no visible issues to cube data evaluation, something it can be informed by the 1000 points post-verification the user performed.