- This topic has 10 replies, 4 voices, and was last updated 6 years, 5 months ago by
.
-
Posts
-
I have been having issues with using DisplayCal 3.3.5.0 with Sierra 10.12.6 and an LG 24MP68 monitor, calibrated with a ColorMunki Design. (I have also tried to calibrate with a i1Display – with the same issue, and I have also swapped out Nvidia 980ti <-> AMD 7970 gpu with same outcome- so it doesn’t look to be a specific GPU issue)
The generated profiles have a strange pink tint in the grey tones. I have tried with using Low & High HDMI level, and also setting DisplayCal to advanced mode and setting Fullrange manually- and TV range manually.
Everything I try simply makes the pink more noticeable.
Settings for calibration are simply 2.2 gamma- 120 brightness- white point as measured.
Currently the only profile that looks ok on this display is the default ‘LG Calibrated’ profile.
I haven’t come across this issue before.
Let me know what I can supply to help trouble shoot this issue.
Possibly it could be corrupt plist files somewhere… but I don’t quite know which preference’s are used for colour profiles in MacOS.
Thanks in advance.
UPDATE: I just noticed after trying the same hardware with an Arch Linux install that DisplayCAL running in MacOS Sierra does not change the monitors profile as it enters into calibration mode.
In Arch Linux (and I am guessing correct functionality) once the calibration has started the currently loaded display profile (monitor colour) is swapped out for something much different.
For example if I have a bad profile loaded it simply continues to take measurements with that loaded.
I am assuming that DisplayCAL has similar behaviour regardless of OS.
I am guessing that I could just use the profile generated in Arch to load into Sierra…
The generated profiles have a strange pink tint in the grey tones.
What calibration whitepoint target are you using?
DisplayCAL running in MacOS Sierra does not change the monitors profile as it enters into calibration mode.
The only thing that happens during calibration is that any existing calibration in the graphics card gamma tables (videoLUT) is reset to linear. Profile assignments are not touched.
Possibly it could be corrupt plist files somewhere… but I don’t quite know which preference’s are used for colour profiles in MacOS.
Color profiles are not related to plist files.
What calibration whitepoint target are you using?
both 6500 and native
The only thing that happens during calibration is that any existing calibration in the graphics card gamma tables (videoLUT) is reset to linear. Profile assignments are not touched.
In MacOS I can’t see that happen- but it may not be the issue. I have since taken the Arch Linux icc profiles and placed them in MacOS- yet the same Pink issue exists.
Color profiles are not related to plist files.
ok, I’m just starting to think this is some sort of bug somewhere… as this issue seems to only happen in Finder (its side bar).
The colour change is a hard value, not in the display- as I used pixie (macOS colour picker) to see values of pixel colours. I can clearly see the value is pink in pixie. Wondering if Finder is not correctly colour managed- or if a quantisation is happening in MacOS interface elements?
Just a quick update: I was able to load the attached icc profile on another computer, and noticed the same issue… so I am thinking that I will increase the time of the profiling in DisplayCAL. Possibly this monitor has a bit of a spike at some point in the spectrum, so hopefully a longer and higher quality scan could fix this?
Would I be correct in this summation?
I’m having the exact same issue with DisplayCAL 3.3.5, High Sierra 10.13.1 an Asus MG279Q and a ColorMunki Display.
Every light gray is pink, it’s very noticeable in the finder side bar, the system preferences displays window and the save dialogs.
The weird thing is, if I drag the window that is displaying the pinkish gray, to another display and then bring it back, the gray pink is gone until the window is opened again.
I’m calibrating with LCD generic without correction, speed set to High, color temperature 6500k, 120 cd/m2 and gamma 2.2.
The profile generated by x-rite’s software doesn’t have the same issue.
Calibrite Display SL on Amazon
Disclosure: As an Amazon Associate I earn from qualifying purchases.Hi Pedro,
There are two kind of profiles in Xrite software, one of the differeneces between them is in 1xTRC (grey neutrality is taken as “perfect”) vs 3xTRC (one for each channel).
I think that “lower end” software like Color Munki Display software bundled with your colorimeter only supports the less complex, less accurate type of profile: 1xTRC (+ matrix). And you said that “Pink issue” is not happening with that software.That point to one possible source of your problems: please open DisplayCAL and inspect (“i” button) DisplayCAL’s profile. Select that profile as default profile for screen before starting DisplayCAL. On DisplayCAL’s profile info window there is a combo box. Select tone response curve (TRC) and post here the graph. If channels dift away then any color managed program will try to “compensate” it and if done with low bitdepth calculations this results in this kind of green/magenta cast on greys.
That is one of the reasons why “desktop color management” is a huge mistake done by Apple. Apps should be explicitly color management aware (by themselves or relying on OS CMM api) not “all windows on UI” by default… but thats another story.If this is happening to you, I mean: “channels drifting form each other on DisplayCAL’s TRC graph” then re-run a DisplayCAL calibration&profilling with 1xTRC profile (single curve+matrix). As long as there is no GPU banding artifacts (intel iGPU, nvidia) on calibration, you should not see that issue.
Another alternative is to keep the more accurate 3xTRC( + matrix/XYZLUT) and do not use OSX.Thanks for the explanation Vincent! I will give this a go soon and report back.
As MacOS is a bit challenged in its use of full-colour management throughout UI, maybe there could be a warning in DisplayCAL?
“If problems are noticed in colour, please use 1xTRC profile“
(just an idea)
