Home › Forums › Help and Support › 3.3.1 Maximize Lightness Difference sequence
- This topic has 10 replies, 2 voices, and was last updated 6 years, 10 months ago by Florian Höch.
-
AuthorPosts
-
2017-06-19 at 4:40 #7571
Hello,
I’ve been using this new sequence with the LG B6 Oled, and it has greatly improved the max self-check error from ~6dE to 3dE and under😀.
I’m having a problem with the white balance being off after profiling, though, with increased red. This was with a 2600 patch sequence. I tried a 75 patch run to test, which resulted in a better gray scale. Not great but good enough. After trying the higher patch number again, I got the same error as the first time with elevated red. This was measured subsequently with hcfr. Luckily, I was able to adjust the white balance with the tvs controls and maintain the color accuracy from the profile.
So, overall the new sequence has been an improvement in profiling and creating a lut for the LG B6 oled. Now, if I can figure out the gray scale issue. I have only tried the “maximize lightness difference” option so far.
Thanks!
Charles
2017-06-19 at 4:46 #75722017-06-19 at 5:48 #7576And compressed archives
Attachments:
You must be logged in to view attached files.2017-06-19 at 9:30 #7579Hi,
I’ve been using this new sequence with the LG B6 Oled, and it has greatly improved the max self-check error from ~6dE to 3dE and under.
Thanks for the feedback, that is very useful information. The difference in average dE would be interesting.
I’m having a problem with the white balance being off after profiling, though, with increased red. This was with a 2600 patch sequence. I tried a 75 patch run to test, which resulted in a better gray scale.
The profile’s white balance is fine, but the OLED’s white chromaticity has drifted quite a bit between the two runs, during the run with the smaller patch set the measured white was considerably closer to the target whitepoint, thus the 3D LUT had to correct less. You can set rendering intent to relative colorimetric if you don’t want the 3D LUT to correct the whitepoint.
2017-06-19 at 22:58 #7582Thanks Florian
The average self check errors were good before when using “minimize display response delay”. They’ve stayed consistent, all around 0.4 and RMS around 0.55.
2017-06-20 at 22:45 #7586Just an update,
It looks like I spoke too soon concerning the max dE improvement. with the new patch sequence. I was comparing the profile and LUT that I just did using “maximize lightness difference” with the profile/LUT using “minimize display response delay” that I did a couple of months ago before updating to the current version. That previous profile’s self check errors were .4 average, 6.19 max and .54 RMS.
I just ran two profiles back to back, one using “minimize display response delay”, the other “maximize lightness difference”, both using 1500 patches. This time the profile using min display response delay had a lower max self check error:
average .5
max 1.53
RMS .56.
For the max light difference profile, the self check errors were:
average .41
max 2.47
RMS .51
So, I’m getting bettor errors with both patch sequences. I don’t know if its the new DisplayCAL version or if my display has become more stable.
Also, the two patch sequences in combination with this display are still resulting in different white balance errors, with the “maximize display response delay” sequence having more elevated red. The previous profiles I posted where you mentioned the display white point shift, I believe I did readjust the 2-point hi setting on the tv between the two runs, explaining the difference. With these two back to back profiles, the white point was adjusted and checked to read 0dE with the interactive display adjustment window before each profile run. Like you mentioned, I can use relative colorimetric rendering or just adjust the white balance subsequently. I just thought I’d mention that in case it’s useful in evaluating the new sequence. I’ve attached each profile and a gray scale measurement for each with hcfr.
Sorry I don’t have something more conclusive. For whatever reason, I’m getting better results,though!
- This reply was modified 6 years, 10 months ago by charles d.
Attachments:
You must be logged in to view attached files.2017-06-21 at 14:52 #7593With these two back to back profiles, the white point was adjusted and checked to read 0dE with the interactive display adjustment window before each profile run.
That won’t help against display white drift though. White drift was nearing 5 dE during the course of the profiling measurements.
Like you mentioned, I can use relative colorimetric rendering or just adjust the white balance subsequently.
Adjusting the display after the fact is not a good idea as it invalidates the profiling.
2017-06-21 at 19:19 #7594So, I guess the best thing in this case is to do a full grayscale calibration with the tv’s controls first then use relative colorametric rendering intent for the profile?
I was hoping since verification errors and color checker measurements were still good/consistent after adjusting the grayscale that it was workable.
- This reply was modified 6 years, 10 months ago by charles d.
2017-06-21 at 19:41 #7596So, I guess the best thing in this case is to do a full grayscale calibration with the tv’s controls first
Not necessarily. My impression from other user reports is that the LG OLEDs perform best if not adjusted too much.
use relative colorametric rendering intent for the profile?
For the 3D LUT, yes.
I was hoping since verification errors and color checker measurements were still good/consistent after adjusting the grayscale that it was workable.
After you adjust it, it’ll drift again anyway, so you’re not gaining much.
2017-06-22 at 17:56 #7605After you adjust it, it’ll drift again anyway, so you’re not gaining much
Here’s the adjusted grayscale of the last I posted from the 1500 patch “minimize display response delay” and a second measurement I just did, so about 24 hours between the two. It doesn’t look like there’s much drift with normal content. The drift seems to be related to running extended profile patch sequences. Maybe its just a matter of experimenting with patch sequence and number as well as size, APL etc.
Thanks for all your feedback!
Attachments:
You must be logged in to view attached files.2017-06-23 at 15:55 #7609The drift seems to be related to running extended profile patch sequences.
That seems plausible. In that case, running fewer patches might help.
-
AuthorPosts