Dear Community,

I have a very specific issue regarding the LUT creation: I did a 3^21 calibration profiling run for my PA32UDCM in SDR the other day and hit awesome results for Rec709 with both 2.4 and 2.2 EOTF of DeltaE under 0.4 with a 1000 patch confirmation.

I have a very particular client who wants to match photo and video and was thinking to try out a sRGB calibration. Since the profile was only 2 days old, I thought to just use it to create the LUT as the EOTF/color should be very similar to Rec709. However the confirmation came back with only 84Nits peak Luminance instead of 100. Naturally I thought it would be an sRGB problem, but after further testing and even recreating a Rec709/2.4 LUT, same result here (also hitting 84Nits peak) when before everything was fine.

My suspicion is that it is the Limited Range of 16-235 creating an issue when not doing the LUT creation in the same session as the profiling, but after relaunching. I did set the Limited range of course in the profiling again. All settings are identical to the calibration run, but results are different. It even marks the profile as Limited Range in the LUT creation window, but seems to do it for the full range.

Am I missing something? Would be great to know for the next calibration run.

Best,
Philipp

The only different between Rec709 and sRGB as the target colour space is the EOTF of Rec709 is 2.4, and for sRGB is 2.2.
(So as you both already you have done sRGB...)
But, if the input signal changes from Legal TV to Full data range, that will cause issues totally outside of the LUT generation.

Regardless, you appear to have issues not related Rec709 vs. sRGB and LUT generation.
That is something you will have to isolate for yourself, as remote guessing will not help.
But you can post the profiles, including the verification profiles, and the LUT for assessment.
(Zip them up and upload to the forum, or send via email.)

Steve

I have exported the new Rec709/2.4 LUT and the profile I ran for the display. And ran a quick test for it to illustrate the issue.

I also re-measured the Rec709/G2.2 that is still installed in another slot from that calibration run (with excellent results). The older measurement screenshot (9th of April) is from that same day with the old G2.4 LUT (which I have not saved and got lost as a working LUT).

LUT creation was done the same way as this new Rec709/G2.4 with max luminance set to 100.

I can not figure out where this is going wrong. Measurements seem to be going correct, as the old calibration is still great. It's just the new one that is wrong.

Settings on the ASUS are the same, uploaded the Unity Bypass first and then uploaded the new LUT.

Any chance you see the issue?

Attached file: Archiv.zip

Sorry, you need to provide the pre- calibration profile, the LUT for Rec709 and for sRGB, and the verification profiles with the LUTs active too.

Edit: The LUT has been generated to do a peak luma reduction.
The pre-calibration shows the white point is slightly too green, so the LUT will do a reduction for that, and on top of that you have set a Max limit.
As the pre-calibration is only 105 nits you are forcing the LUT Generation to go lower.

Second Edit: You had 'Limit Range' enabled within Settings for the pre-calibration Profile!
That should NOT be used for any 'relative' calibration (unless you really know what you are doing, and are trying to do something very specific).
It is for Absolute calibration, such as ST2084.

Third Edit: you need to re-do the profile, as the Limit Range setting was used when profiling, so the profile data is incorrectly scaled.

Steve

I might be sounding totally dumb here, but the profiling run of the 3^17 is the one done with the unity bypass active and hence the pre-calibration run (only in monitor adjustments present)? Then I create the calibration LUT for the target, upload it into the monitor and measure again to confirm the calibration?

File Naming:
-SDR 3^17.bcs is the pre-calibration profile
-SDR G2.4 L100 04-14.3dl is the LUT created as seen in the screenshot

See my post edits above.
But yes, any verification is made with the LUT active to 'check' the calibration accuracy.
The workflow is defined in the 3D LUT Calibration user guide.

Steve

Thanks for checking all of this out Steve! Really appreciate the help.

Since my client monitor was calibrated for limited range, I wanted the two to match. The calibration with the limited range option went very smoothly and worked like a charm (as seen in the active results from the Rec709/G2.2 measurement that is still active). I did also create the LUTs the same way with the limited range profile. Measurements for both Rec709 G2.2/G2.4 came in awesome after the calibration and with limited range active. Without the max luminance of 100, the peak luma remained at 105 after the calibration, so I had to adjust it to max at 100. After that it matched the EOTF perfectly.

The issue only seems to appear when starting a new instance (like restarting the application) and trying to create new LUTs from the limited range profile. It seems that color space doesn't create the correct limited range LUTs, but for full range (when it did create the correct limited ones beore)? Very confusing to say the least.

All in all, I will definately do full range for both monitors again to avoid these issues going forward.

Using the the Limit Range function will have invalidated the profile.
You will need to re-run a calibration.
You likely got an 'acceptable' result as the monitor is fairly linear.
But it will not be a 'correct' result, if Limit Range was indeed active when the original profile was made.

Steve

Well, you may be pleased to know that you have found a bug!

The bug has not bee seen before, as your workflow of using Limit Range with relative calibration is wrong, and I doubt any other users (and certainly not us) have tried it like this.

So yes, when you first made the LUT the Limit Range setting was ignored in LUT Generation, BUT THE PROFILE IS STILL WRONG.
It appeared to work due to the display being fairly linear.

But when you tried to make a second LUT, after closing ColourSpace, etc., the Limit Range setting was correctly used, and the LUT reduced by approx. 9%.

Steve