MacOS struggles

Home Forums Help and Support MacOS struggles

This topic contains 52 replies, has 4 voices, and was last updated by  buckie (@buckie) 2 days, 23 hours ago.

Viewing 15 posts - 16 through 30 (of 53 total)
  • Author
    Posts
  • #18436

    buckie (@buckie)
    Participant
    • Offline

    Not trying to be difficult, just want to understand, how exactly my display stability issues which get into the profile make Photoshop and Preview display two different images of the same file at the same time on the same display? I fail to see how this is related because the ICC profile, however “untrue” it may be, is the same for both apps. Failing to see the logic here. Could you clarify?

    • This reply was modified 1 month, 3 weeks ago by  buckie.
    • This reply was modified 1 month, 3 weeks ago by  buckie.
    #18442

    Florian Höch (@fhoech)
    Administrator
    • Online

    Not trying to be difficult, just want to understand, how exactly my display stability issues which get into the profile make Photoshop and Preview display two different images of the same file at the same time on the same display?

    See the sudden slope at the top end in the TRC curves? That is very bad, because it is not your actual display response. You need to fix it, regardless of whether or not the resulting profile will work with Preview or not.

    #18476

    jlink (@jlink)
    Participant
    • Offline

    I’m very new to this whole topic but trying to get deeper in it.

    I have calibrated and profiled my MacBook Pro to Rec 709 BT 1886 2.4 with XYZ Lut + Matrix for working in Premiere Pro (using the new color management that’s implemented).

    First I ignored the message that told me to use single curve + matrix. Then I experienced these issues described in the stackexchange link. Preview behaving weired, when zoomed etc. I than tried to use single matrix and black point compensation ON but this did not change the behavior of Preview. Also OS and everything looked pretty similar to the XYZ Lut + Matrix Profile.

    Is that because I would have to use Standard with a Gamma of 2.2 to deal with that bug of OS X?

    I think in my case it would be better to stay in the video world, right? Even if that means that only Premiere and the other color managed CS products would show me the right colors?

    Thanks!

    Jonas

    #18477

    jlink (@jlink)
    Participant
    • Offline

    PS: I experience something strange, don’t know if it’s a bug in displayCAL: When I choose the XYZ Lut + Matrix Settings again profile type changes to single curve + matrix but the amount of patches stays at 1553 (with which I did the XYZ Lut Profile). So does the estimated time. It seems that only the display changes but in reality in stays on XYZ Lut…

    PPS: I’m on a late 2018 MacBook Pro, running Mojave 10.14.5 and using a i1 Display Pro.

    i1Display Pro on Amazon  
    Disclosure: As an Amazon Associate I earn from qualifying purchases.

    #18478

    Florian Höch (@fhoech)
    Administrator
    • Online

    I think in my case it would be better to stay in the video world, right? Even if that means that only Premiere and the other color managed CS products would show me the right colors?

    Yes.

    When I choose the XYZ Lut + Matrix Settings again profile type changes to single curve + matrix but the amount of patches stays at 1553 (with which I did the XYZ Lut Profile). So does the estimated time.

    That is intended – if you want to create another XYZ LUT profile from the same settings, you have to choose the profile type manually.

    #18480

    jlink (@jlink)
    Participant
    • Offline

    Thanks, Florian, for the fast reply!

    And vice versa it means XYZ Lut ist the best/most accurate type for me, i.e. I would not benefit at all (expect concerning time) from choosing single curve + matrix?

    #18482

    Florian Höch (@fhoech)
    Administrator
    • Online

    And vice versa it means XYZ Lut ist the best/most accurate type for me, i.e. I would not benefit at all (expect concerning time) from choosing single curve + matrix?

    Yes.

    #19131

    Wire (@wire)
    Participant
    • Offline

    I’m usung Mac Mojave with DisplayCal recommended settings on conventionak RGB displays and not seeing the diff you describe between Preview and Adobe CMS.

    My take per Fkorian’s response is given MacOS is a blackbox with you note has documented problems pertaining to custom profiles, the instability of your display magnifies the chance of problems.

    You say your display is plasma…, is it attached via HDMI, and if so is it running in YCbCr mode (eg YUV 4:2:x) like a TV or in RGB mode like DVI graphics?  I would expect YCbCr  on Mac to be more prone to rendering glitches than RGB, but this is purely a hunch

    #19183

    buckie (@buckie)
    Participant
    • Offline

    No, instabilities don’t “magnify” that particular problem. You can download the profile I created, it’s posted earlier in this thread, apply it and see if Preview misbehaves on your machine in any way with that profile.

    The panel is forced into full RGB mode (0-255) so it is treated like a regular display by the OS.

    #19185

    Wire (@wire)
    Participant
    • Offline

    I appreciate the attention to language…

    What I was trying to convey is that given that we know that MacOS CMM doesn’t handle certain types of profiles well, then for those types we might expect very erratic behavior based on parameters of such profiles which are borderline—IOW there’s no reason to expect a graceful degradation; an unstable display might produce a profile that for reasons of deeper incompatibility causes wildly erratic results.

    But this point is academic…

    I installed your PIONEER profile as my display profile. On my Dell IPS the calibration produces a slight uniform overall green color cast which I should think is perfectly appropriate to alignment of your display.

    I then looked the lagom.nl test images for BW-gradient, white, and black patches. I see perfect rendering in both Photoshop and Preview: The gradient is smooth, and top and bottom-end patches are right where they should be.

    I also tried assigning your profile to same images in PS and didn’t see anything the least bit odd.

    As to your situation with JPEG artifacts, I can’t reproduce that experiment precisely. I tried saving the gradient from PS as a very low quality JPEG (Qual 2) and see appropriate artifacts that don’t do anything funny with tonality beyond expected banding. Based on what I see I can’t imagine how I would get your results, unless there was something wonky with the specific image you refer to.

    I am running MacOS Mojave 10.14.6 build 18G84 on a Mac Pro 3,1 (unsupported by Apple).

    HTH

    #19186

    Wire (@wire)
    Participant
    • Offline

    Also just to be sure about your config, you can check you mac
    (Apple Menu) About This Mac > System Report > Graphics/Displays
    to see if the listed parameters for the display include “Television” which would indicate YCbCr mode.

    Here’s a couple threads on the topic of MacOS  insisting on YCbCr for certain displays.

    https://apple.stackexchange.com/questions/259751/what-is-television-in-system-information-report-for-an-external-monitor-that

    Fixing the External Monitor Color Problem with My 2018 MacBook Pro

    #19187

    Wire (@wire)
    Participant
    • Offline

    Related question:

    I just used DisplayCal 3.8.1 to make an sRGB XYZ LUT display profile for a 2012 Macbook Pro running Mojave 10.14.6 18G87 (latest release) and the display looks great plus Preview seems to be handling images with various profiles properly.

    What’s the right way to check for the known problem with LUT display profiles and MacOS CMM ?

    #19196

    buckie (@buckie)
    Participant
    • Offline

    It could also be due to me testing all of this out on a hackintoshed Mojave running on AMD Ryzen. Could partially explain why I never had issues in El Capitan which ran on a 2008 Mac Pro. I would have expected a crash, but it may very well be due to CPU instruction incompatibilities that trigger some kind of fallback behavior which in turn produce weird results.

    I think the test is pretty straightforward: if the profile I linked doesn’t cause any obvious visual discrepancies on your end, then it’s pretty much the only remaining guess.

    Speaking of TV vs display, I already did an override according to an article linked inside your second link because it’s the first thing Mojave greets you with after installingx. It absolutely runs in 0-255 mode because the panel itself is set strictly to that mode. Any attempt to feed ycbcr to it in that mode results in incorrect colors everywhere regardless of profiles.

    It’s pretty hard to photograph the display but you can approximately see what’s going on: the gradient is completely smooth in PS, the same image saved in PNG and opened in Preview exhibits complete black crush at around 20% in the gradient and banding is also very visible. Both images being previewed at 100% 1:1 scale.

    Attachments:
    You must be logged in to view attached files.
    #19198

    Wire (@wire)
    Participant
    • Offline

    One more thought:

    Does the MacOS System Report describe the display as “Television” ?

    YCbCr vs RGB mode may be separable from 16–235 vs 0-255 (“limited / full”) range HDMI ?

    Your unit being a TV and the odd response at bottom / top suggest to look a bit more at this aspect of config.

    Also how to know when to choose 16–235 vs 0-255 in  DisplayCal settings?

    #19199

    buckie (@buckie)
    Participant
    • Offline

    Enable “Show Advanced Options” and you will see three options for RGB ranges.

    System report does indeed describe the display as Television, but I don’t believe that affects the RGB now that the display profile had been patched. Detecting it as Television enables overscan feature in displays config panel which may be useful.

    It’s not the limited range on the panel that’s at issue here. Enabling 16-255 mode on the panel simply cuts out the blacks (similar to how Preview misbehaves but it looks sane and smooth) and that is readily visible in Photoshop. Enabling 0-255 mode back immediately brings out the data in the shadows that was invisible in 16-255 mode.

Viewing 15 posts - 16 through 30 (of 53 total)

You must be logged in to reply to this topic.

Log in or Register

Display Calibration and Characterization powered by ArgyllCMS