- This topic has 31 replies, 4 voices, and was last updated 1 month ago by
.
-
Posts
-
The 79 patch does a lot more color. It is 2x as good to me. It test 100 level of primary and secondary color and 50 shades of all colors and secondarys It explains why one color was a 4.0x DE and now a 3 a 3 ish or less. sorry I got bad memory. It does lots of shades and all colors at 100. 100 is all we need to measure to set it.
I just had to share a Mac guide on a photograpy site. https://photographylife.com/how-to-calibrate-apple-mac-monitor
OK. I understand that I got good results by chance and not because I selected the right settings ?
So, i should rerun calibration with LCD PFS Phosphor WLED instead of White-LED, 6500 K white point, black level as measured, Gamma 2.2, set Single Curve+Matrix, as macOS doesn’t support LUT corrections and lower the calibration speed to low.
Is that correct ?
Then which testchart should I use in the profiling tab ? 34 patches ? 54 ? More ?
Think about is as a linear regresion trying to predict behavior with a few dots as input. It seems that measured blue in ti3 is off actual blue OR that ICC tried to fit the “linear regresion” (oversimplification, just to make explanation easier) to an overall lower error loosing accuracy on that outlier point (blue).
IDNK which one is happening, hence you can inspect ti3 or maybe create a single matrix profile with “more dots” o minimize outlier impact in final “simplified linear regresion” (ICC description of display as primaries + gamma).Redo as in your 1st post (recomnended settings for macOS: single curve + matrix, BPC + PFS phosphor mode) but choose as testchart “small” or “extended testchart for LUT” (115, 175)
Testchart is not profile type, it’s just the “dots” you are going to measure.
If error in blue goes up, try to choose the opposite direction “small testchart for matrix”. (34)
Anyway, try to find measured RGB 255 blue in both scenarios because IDNK which one of the two scenarios in my 1st paragraph is happening.I just had to share a Mac guide on a photograpy site. https://photographylife.com/how-to-calibrate-apple-mac-monitor
Mac use of i1Profiler or Dataclor or DIsplayCAL is the same as windows on old P3 macbooks/iMacs or 3rd party display without HW cal BUT due to macOS limitation of oversimplified porifiles (likely because they run color management on GPU shaders, just a guess) you should choose dafault seetings: display gamut described by primaries, gamma describd by idealized perfect grey (1 curve) and infinite contrast (BPC).
On a well behaved display you’ll use the same on a Windows machine to avoid 3rd party color management module to cause rounding errors due to 3 curves in TRC causing coloration on greys.I said “old P3 macs” becaue on new XDR displays you usually avoid all these process. DIsplay is likely to have exactly P3 gamut on a modified WLED PFS backlight, grey is highly expected to be perfectly neutral with sRGB-2.2 like gamma and the onñy issue may be matching white to wht yur measurement device sees as D65.
So for “XDR calibration” you take a current/native measurement of 255 whitepoint on CIE xy Y, then go to macOS settings, display, tune display and say to macOS “hey, my white point is not exactly D65 (or other value), it is <MEASURED CIE xyY)” and it will use GPU or RGB gain or mix of them to autocorrect whitepoint. And you leave as display ICC factory ICC.Since macbooks with XDR are relatively new, that is missing on the guide you linked.
For measuring actual whitepoint, better to use DIsplayCAL or even commandline ArgyllCMS (dispcal -R and the use with -X path to CCSS or -y NUMBER for UI list of modes). Do not use Calibrite Profiler as in some Calibrite ambassador Youtube channel suggest: that app lacks of colorimeter correction for WLED PFS (even if they appear on User interface, it’s a lie), and even if they hat the proper EDR for i1d3 colorimeter if you use suggested procedure on those videos it will apply no correction.Calibrite Display Pro HL on Amazon
Disclosure: As an Amazon Associate I earn from qualifying purchases.Well, I don’t see a huge difference. The one with the small matrix testchart (second one) is a LITTLE bit better than with the large LUT testchart, but both show pretty bad accuracy for this particular blue patch.
I tried to match the white point the best before calibration, but the increments are quite big and I couldn’t manage perfect balance between R,G and B (it’s still in the green zone <1.0 dE in DisplayCal though).
Do you have any idea how to improve ?
Try to keep one of the RGB gains at default. If color clips on test patterns I turn down contrast. I noticed if contrast goes from 50 to 49 I can raise green 3 gain and green do not clip and I keep the contrast. 900 contrast is almost the lowest contrast that might be good. Most monitors are 1000 + contrast. You wont have clipping on that monitor. No need to lower green especially. Lowering blue raises red a little. Lower red raises blue. Try adjusting 2 controls instead of 3.
That’s what I did. Green is still at 100 and I lowered the blue and red channels. Contrast is at default level (75/100) and brightness adjusted to match the 120 cd/m2 I chose for the environment I work in,
I’ll try to increase or decrease the contrast slightly to see if I can come closer to the perfect white point though. As you said, it could only need one click more or less to improve it.
It is good as it gets visually. Perhaps tomorrow your screen will rest and wakeup with .4 delta e difference of white 100. Mine almost does. It warms up though to .15 difference. I even suspiciously think the meter, software, and tv recalibrite. Not the Display Cal / argyii software.
I am not a expert though.
Even with a white point deltaE of 0.3 on the first screen (pre calibration, where you adjust the RGB sliders) the measurements are worse than my second try, even though the deltaE was at 0.8… I don’t know what to do now…
Should I use the profile made with the W-LED setting instead of the « right » ones made with Phosphor-W-LED, because all the values are better or is there something more I can do to do another calibration ?
Sorry , no answer. What it says is right does not make it right. Those reports look the same. HTML reports are prefered. You need a program like HCFR to do white balance manually in rgb controls. Even at default software color correction will fix it. Maybe default is best.
Yes, those are the same reports. I see now that I uploaded the report two times… my bad !
I’ll have a look at HCFR, thanks
Wincent was suggesting a larger patch set to help blue. somewhere from 0 to 255 the blue is getting to high with pfs white led. I looked and it the whole wavelength. The color match of screen to profile is wrong with PFS. If your monitor is right and default and is normaly good at default. Rtings could tell what the monitor is like normally.
I have to go off so cant post more tonight.
As said before, to find the source of that error you must check first if TI3 (post calibration measurement data) and predicted values from profile (XYZ, xyY “absolute values” on report) match.
I mean, is this result of some weird behavior after applying VCGT in post calibration measurement? (ICC profile creation) Does this happens when you load VCGT by macOS VCGT loader? (normal use)
Is this a result of ICC profile computation (“fixing points in a linear regresion”)? (ICC making issues)
TI3 are text files, row sample ID, col X or Y or Z.
Once you find the culprit you can try do do something about that (testing creating profile in another OS, sending TI3 and ICC to ArgyllCMS maillist.. etc)INDK what has ahppened, but TI3 file contains the data to shed light on this.
