Copy link to clipboard
Copied
This question/comment is regarding Mac OS colorsync mistmatch with PS/LR rendering.
Problem: Despite proper calibration and ICC profiles, rendering of shadows in quicklook/preview differs from PS/LR. My inclination is to trust PS/LR rendering, as the OS (I'm assuming Metal api) rendering of the same image (grey patch ramp from 1-40) crushes the blacks in the first row. Could have future implications for PS/LR if Metal rendering is further integrated.
Testing conditions: Mac OS 10.14.5/6
2018 Mac mini, 2015 Macbook Pro 13"
Latest versions of each app.
Monitor calibrated with i1Display Pro to sRGB target colorspace, gamma 2.2, 120 cd/m2, native contrast
Proper profile selected in system preferences.
Same image (located at www.lagom.nl) loaded in PS and Preview.
Issue can be replicated on both machines reliably.
Comments: This issue has been driving me insane, though the effect is obviously more subtle in real-world examples of images (photography). From what I can find online, it appears to be a longstanding issue at the OS level by Apple. For whatever reason, applying different gamma/blackpoint is suspected. Most visible in near black tones as demonstrated above. I've been in contact with NEC's director of software development for a number of problems relating to this.
Additionally, GPU rendering in LR is also subtly different than off. Not as severe as the above example, but noticeable when knowing what to look for.
It's likely an OS issue, which is very disappointing. I'm hoping Adobe can work with Apple to get to the root of this problem.
If you can replicate the issue on your own machines, please do chime in. It'll help develop a case.
Copy link to clipboard
Copied
Make sure to assign sRGB to the PNG in Photoshop. If it opens with Adobe RGB assigned, you get this black crush. Adobe RGB has regular gamma 2.2, while sRGB has an idiosyncratic and irregular tone response curve. This affects the conversion into your monitor profile.
When I right-clicked and saved the image from the lagom site, the saved PNG opens with whatever I have set as working RGB in Photoshop (I couldn't believe it so I tested). That's a baffling behavior that I've never seen before - if it doesn't have a profile it should open as "untagged", right?
So that's a mystery all by itself. Have they changed PNG handling of icc profiles yet again? God knows there are enough problems with PNG embedding already.
I should say that I'm on Windows, so I can't speak for OS issues. Just mentioning a possible explanation.
Copy link to clipboard
Copied
Thanks for your reply. Color settings can be found in the edit menu, where you can specify how it handles untagged photos. At least on the Mac, that particular image is saved with a correct sRGB tag. Any sRGB calibrations (that I've seen/used) are concerned, also specify 2.2 gamma. Color management is an animal at the best of times... especially when you're doing both web AND print workflows like I am.
We've (NEC team and I) have made some progress on the issue, and it's indeed MacOS specific. ICC profiles that use a parametric gamma curve instead of a point curve is interpolated incorrectly and results in either dark or light shadow values displayed. Apple have made a change to ColorSync in the lastest point release (10.14.6) that has finally broken it enough to show the issue clearly. Unfortunately, MacOS is in a sad state at the moment, which is why you have to be diligent with issues like these. You'd think a company that has a reputation for creative professionals would get something very basic (as in core) like this right every time. Perhaps those days are long gone.
Copy link to clipboard
Copied
Color settings can be found in the edit menu, where you can specify how it handles untagged photos.
Actually you can't. Whatever the color management policies, an untagged document should always open as "untagged". There is nothing to convert to or from. Of course, an untagged file will be displayed using the working space > monitor profile - but the notation should always be "untagged":
Any sRGB calibrations (that I've seen/used) are concerned, also specify 2.2 gamma.
Yes, for monitor calibration. I'm talking about the tone response curve in the sRGB IEC61966-2.1 profile specification, which is highly irregular, particularly in the shadow end. It is not a regular gamma 2.2 curve at all.
Anyway, this all appears to be a digression.
Copy link to clipboard
Copied
Hi Steve
what software do you use for calibration?
why do you target sRGB colour space in your display calibration?
[wide gamut screens are an issue in a web workflow if you must use non colour management savvy applications. They can ONLY provide correct appearance when running colour management savvy applications, but that’s not the issue here, just mentioning]
could it be you'd not have this problem if you calibrated the screen to "native" gamut, I suggest you try that.
to its native gamut
FYI: sRGB is actually not 2.2 gamma, it has its own strange curve
"Unlike most other RGB color spaces, the sRGB gamma cannot be expressed as a single numerical value. The overall gamma is approximately 2.2, consisting of a linear (gamma 1.0) section near black, and a non-linear section elsewhere involving a 2.4 exponent and a gamma (slope of log output versus log input) changing from 1.0 through about 2.3. The purpose of the linear section is so the curve does not have an infinite slope at zero, which could cause numerical problems."
I am on latest Photoshop CC, Eizo Coloredge CG247 calibrated with a basICColor DISCUS instrument and basICColor display software (which accesses the Eizo hardware LUT). OSX 10.12.6
For NEC Spectraview screens - basICColor did provide their software rebadged as "Spectraview Profiler" for Europe but, I believe but not for the USA.
When I downloaded that png image you directed us to and opened it in Photoshop, sRGB was assigned [it does not open as untagged.
["North American General Purpose 2"was set is my default the moment, that means "missing profiles-ask when opening" is not checked [see below]. I can clearly distinguish patches 1-5 in the top row. I could see them via the web browser too.
Same appearance in Apple Preview.
When I reset Photoshop's colour settings to "Europe Prepress 2", "missing profiles-ask when opening" is checked
I see this warning:
IF I "leave as is" I get the same appearance as your screenshot. the first row is all invisible.
IF I assign Adobe RGB (as offered above) I get the same appearance as you, the first row is invisible.
so - if assigning a profile to that png file, then it MUST be sRGB
Plainly my Sierra install differs from your version of OSX, its older, but at least it proves the issue did not exist at the time when Sierra was released.
However there may be an issue with OSX 10.14.6 - see my next post
I suggest you join the: colorsync-users@lists.apple.com
then post under the title: "Possible ColorSync discontinuity bug introduced in OS 10.14.6 with display "r/g/bTRC" tags?"
suggestion:
What I write below is still worth a read I think - you might test that.
I've never seen this issue you describe on any mac screen calibrated with basICColor display software, why not download that SW and give it a shot, [free demo] - yes, it works with your i1 display pro
here basICColor display 5:
MAC: http://mylicense.biz/getProduct.asp?proId=180&downloadKey=w8hz-pe6q-2n2s
[full disclosure, yes I am a basICColor reseller]
I would test an LUT profile and the lower quality matrix profile too.
By the way:
The people replying to your enquiries here are volunteers, Adobe users, sometimes Adobe folk seem to get involved but its rare.
Don't presume that posting here is going to get Adobe and Apple sorting out this issue.
I hope this helps
if so, please do mark my reply as "helpful"
thanks
neil barstow, colourmanagement
Copy link to clipboard
Copied
Hi
research at colorsync users list indicates that you are not alone in seeing an issue with the macOS 10.14.6 update
"
Possible ColorSync discontinuity bug introduced in OS 10.14.6 with display "r/g/bTRC" tags?
| Tue, Aug 13, 4:21 AM (1 day ago) | |||
|
I hope this helps
thanks
neil barstow, colourmanagement
Copy link to clipboard
Copied
Yes, I'm one of those that contacted William. I trust the issue is in good hands now Thank you so much for your replies.
Also for i1Profiler, if you use the large patch set, the resultant ICC has the same issue. Medium or small patch options result in point response curves.
Copy link to clipboard
Copied
Steve,
just to be clear -
not sure what you mean by a "point curve"
that reads like a simple gamma curve with a single point on it?
generally display profiles are described as having a gamma curve ( a simple curve) or a look up table (LUT).
do you mean "encoded as an array of points"?
William wrote:
"
what I'm seeing is that when a display profile
has the r/g/bTRC "curv" tag encoded as a "simple gamma" value rather
than an array of points (typically 256), then there can be a big
luminance discontinuity between level RGB=0 and RGB=1"
I hope this helps
thanks
neil barstow, colourmanagement
Copy link to clipboard
Copied
This was asked in another forum, the answer is this: It's a bug in ColorSync, it doesn't affect Adobe products using ACE. Will at NEC is now aware of this as is Apple.
Copy link to clipboard
Copied
My mac just upgraded to Os 10.14.6 last week and My mac is unusable. My Eizo Flexscan S2411 is calibrated with a datacolor sypder 5 pro. The blacks have all blocked in and sever noise in the dark shadows. Is there any fix on this yet ?
Copy link to clipboard
Copied
Hi Paul, you may find this is an issue with the profile 'type' made by your Datacolor Spyder. I like basICColor display (and you can get a free 14 day demo to se if it fixes your problme.
I wrote the developers about the issue described above and got this reply:
"no professional profiling app would write a gamma (or other) equation into the vcg tag. Display 5 and 6 uses 256-point curves."
Why not test it?
basICColor display 5:
MAC: http://mylicense.biz/getProduct.asp?proId=180&downloadKey=w8hz-pe6q-2n2s
WIN: http://mylicense.biz/getProduct.asp?proId=181&downloadKey=uxr7-3m2p-bzx3
heads up - bias alert - we are basICColor dealers
I hope this helps
if so, please "like" my reply and if you're OK now, please mark it as "correct", so that others who have similar issues can see the solution
thanks
neil barstow, colourmanagement