search

eoyilmaz / displaycal-py3

DisplayCAL Modernization Project
https://eoyilmaz.github.io/displaycal-py3/
GNU General Public License v3.0
809 stars 60 forks source link

White Point Issue & CDM Issue? #411

Open decayedxistence opened 5 days ago

decayedxistence commented 5 days ago

Two things I've noticed:

1) If I try to manually set the measured cdm from default 120 to 100, DisplayCal will fail the measurement

2) Even if white point measurement set to "as measured" or manually set to 6500K, with the several measurements I did this morning show up as 6400k in the measurement report for both the target and result when it should be 6500K (even as assumed target WP)

I'm using Resolve and the same monitor and in the previous Display Cal v3.9.12, It always measured and calibrated to around 6500K (give or take a few single digits).

eoyilmaz commented 4 days ago

2. Even if white point measurement set to "as measured" or manually set to 6500K, with the several measurements I did this morning show up as 6400k in the measurement report for both the target and result when it should be 6500K (even as assumed target WP)

Could this be related with #392

We are seeing the effect especially in the white level and balance.

decayedxistence commented 3 days ago

Hmm...doesn't sound like it's related. I'm trying to create a 3D LUT for Resolve and not an .icc profile when I notice the issues mentioned above.

eoyilmaz commented 3 days ago

1. If I try to manually set the measured cdm from default 120 to 100, DisplayCal will fail the measurement

this is fixed with 06a58aa78638cdecb0c98c6f408b7f95ae4d6430

eoyilmaz commented 3 days ago

@decayedxistence I have a couple of questions, what display are you profiling & measuring, is it an internal display to a Macbook laptop or is it an Apple XDR display or is it a regular monitor. I'm still thinking that this might be related to #392 if this is an XDR monitor that you are measuring.

decayedxistence commented 2 days ago

@eoyilmaz Using a BenQ EW3270U connected to an Ultra Studio monitor 3g to my macbook pro.

In DisplayCal - selecting Video 3D LUT for Resolve.

eoyilmaz commented 2 days ago

alright, let me try to reproduce this.

decayedxistence commented 2 days ago

Thanks. Appreciate you, @eoyilmaz

Adam-Color commented 1 day ago

@eoyilmaz and @secayedxistence , I have had numerous issues with setting up 3D LUT creation for resolve, as the preset and the methods that worked prior to python3 no longer work. In order to get usable results, you need to use video levels in resolve, full levels all the way through in DisplayCAL, turn off drift compensation, tell it to measure black level as measured, tone curve as measured (tone curve calibration is broken in this workflow) black output offset to 100%, absolute gamma, rendering intent to preserve saturation, and apply calibration (vcgt) on. With these settings, I was able to get a really solid calibration result, with proper white point. Perhaps the preset simply needs to be updated? I've included my calibration reports before and after calibration for reference. Uncalibrated Calibrated

eoyilmaz commented 1 day ago

the preset and the methods that worked prior to python3 no longer work

If your workflow was working under Python 2 and not working in Python 3, perhaps we have a bug.

I didn't generate too many 3D LUTs lately, especially with DisplayCAL 3.9.x, maybe I should look into that after I finish with the current work.

decayedxistence commented 1 day ago

@eoyilmaz and @secayedxistence , I have had numerous issues with setting up 3D LUT creation for resolve, as the preset and the methods that worked prior to python3 no longer work. In order to get usable results, you need to use video levels in resolve, full levels all the way through in DisplayCAL, turn off drift compensation, tell it to measure black level as measured, tone curve as measured (tone curve calibration is broken in this workflow) black output offset to 100%, absolute gamma, rendering intent to preserve saturation, and apply calibration (vcgt) on. With these settings, I was able to get a really solid calibration result, with proper white point. Perhaps the preset simply needs to be updated? I've included my calibration reports before and after calibration for reference. Uncalibrated Calibrated

Hmmm Interesting. In the calibration I did for a 3D LUT in Resolve with V 3.9.13, the only issue I saw was the WP coming in at 6400K no matter what I did but also didn't get the "recommended tolerance passed" but did get the "normal tolerance passed". Will try your method and see what happens!

Adam-Color commented 1 day ago

the preset and the methods that worked prior to python3 no longer work

If your workflow was working under Python 2 and not working in Python 3, perhaps we have a bug.

I didn't generate too many 3D LUTs lately, especially with DisplayCAL 3.9.x, maybe I should look into that after I finish with the current work.

Not workflows that I have tried, but all the workflows used on YouTube for resolve 3D LUT creation were not working when I tried them (they all use the python 2 version). The only way I was able to make it work was using 302-resolve-patches-do-not-update and spending many hours of trial and error. More attention given to 3D LUT workflows will be very much appreciated.

decayedxistence commented 1 day ago

the preset and the methods that worked prior to python3 no longer work

If your workflow was working under Python 2 and not working in Python 3, perhaps we have a bug. I didn't generate too many 3D LUTs lately, especially with DisplayCAL 3.9.x, maybe I should look into that after I finish with the current work.

Not workflows that I have tried, but all the workflows used on YouTube for resolve 3D LUT creation were not working when I tried them (they all use the python 2 version). The only way I was able to make it work was using 302-resolve-patches-do-not-update and spending many hours of trial and error. More attention given to 3D LUT workflows will be very much appreciated.

Try upgrading to the latest DisplayCal version, Python, & Argyll....from there, follow Stuart's guidelines, even though it's an Eizo monitor and he's using a spectro, same rules should apply - https://www.youtube.com/watch?v=KQyuJ-EL3DA&t=940s

See if that works and produces better results for you.

eoyilmaz commented 2 hours ago

The only way I was able to make it work was using 302-resolve-patches-do-not-update and spending many hours of trial and error.

This has been merged already, and the latest version should work in the same way.