2 up vertical duplicate image colour shift (bug?)

[Gibson Editions]

I am trying to resolve an issue. I am seeing a color/ luminance shift when viewing two identical images created with image>duplicate in two up vertical view. (see screenshot attached)

Is this a known issue?

PS V 24.1.0

OSX 12.6.2

M1 Max processor (macbook pro 16")

Eizo CG279X monitor

Custom color profile 90cd,0.4cd, 6800k, 2.2 (v4 profile)

 

[D Fosse]

When I overlay these two and flip between them, I am absolutely unable to discern any difference whatsoever. Difference blend mode is pitch black until finally with a maximum curve you get a very faint uniform noise, probably because the screenshot isn't at 100%. Using an Eizo CG2730.

Generally, v4 is slightly risky, but LUT-based profiles can be problematic even in Photoshop. Both are defaults in ColorNavigator. Matrix profiles (called  gamma (eotf) value in CN) are much safer and much more reliable, and I would recommend you change this in Preferences.

[Gibson Editions]

Many Thanks i will try this. However surely, LUT based profiles should be theoretically more accurate than Matrix based ? 

see attached a 100% view. To my eye the left hand side is reading cooler and darker/ slightly more contrast than the right. I am talking very subtly. Probably not viewable in difference unless applified 

[Gibson Editions]

So ive just had a colleague using the same test image duplicated and displayed 2 up vertical. this time on a different machine (Mac pro 6.1) with a different Eizo CG277 and same calibration target and they are seeing the same shifts that i am on screen. 

"left hand side is reading cooler and darker/ slightly more contrast than the right." 

Any ideas Adobe? 

[Michael J. Hoffman]

When you see the image on the left as cooler, can you swap the position of the two images? Does the anomaly stay in the same are of the screen? Sounds to me more like a screen/environment issue, as the images you are posting here are completely identical.

[Gibson Editions]

Thanks Michael, I was thinking this too. However, My colleague has performed the same test on a different machine, in a completely environment (different building). Which makes me believe this is software bug, The cooler representation is always on the left. I can even see the difference using a blank white canvas 

[Gibson Editions]

I've just performed the following test. I made a blank document White background, and duplicated it then arranged them 2 up vertical. 

I then measured the two sides with i1pro2 and see the results. This is for sure a bug

[TheDigitalDog]

I've just performed the following test. I made a blank document White background, and duplicated it then arranged them 2 up vertical. 

I then measured the two sides with i1pro2 and see the results. This is for sure a bug

By @Gibson Editions

You're measuring what is emitting from the display. That doesn't point the finger at a Photoshop bug. 

Give us Lab values; then we can calculate a more useful metric; the deltaE of the two areas. 

Minus 'noise' in measurement devices (measure the same spot 5 times in a row; depending on the device, don't expect identical Lab values). 

[D Fosse]

There are only two possibilities here:

• you have a defective Eizo. Not something you come across often, but there's a first for everything.
• an optical illusion, possibly something in the room lighting

Your two screenshot examples are identical, end of. And that should be simple enough for yourself to test, not something we need to do.

[D Fosse]

If so, it doesn't come through to your screenshots. I can't think of any more ways to show that fact than I've already done. How do you account for the "zebra" example if there really was a difference?

Switch the order left-right and see if you see the same thing. I'm pretty sure this is an optical illusion.

[TheDigitalDog]

So ive just had a colleague using the same test image duplicated and displayed 2 up vertical. this time on a different machine (Mac pro 6.1) with a different Eizo CG277 and same calibration target and they are seeing the same shifts that i am on screen. 

"left hand side is reading cooler and darker/ slightly more contrast than the right." 

Any ideas Adobe? 

By @Gibson Editions

So the RGB numbers are the same, right? So it's your display or a display issue. 

Check out: 

http://digitaldog.net/files/20Testing%20your%20display.pdf

You could try the Apple Digital Color Meter (which is often misunderstood as measuring something in the image; no). 

The Apple Color Picker magnifying glass reads the data converted to the current display profile i.e., from the frame buffer. Which is what you should be looking at. 

The purity over your display could be really poor! As such, you see differences in different areas with the same RGB values. 

Make a document in Photoshop that's Lstar 50 only, and fill the screen; does it look gray evenly over the entire display? 

And there's nothing Adobe can do about this; their product is simply using a large portion of your display path; good bad or ugly. 

[Gibson Editions]

Many thanks for the article, it passes those tests. 

The values of the original images are indeed the identical.

Apple color picker wise on a L 50 is reading consistent values + or - 1 across the screen. It looks pretty consistently gray across the board its, potentially slight hue shift left to right, but its very difficult to discern. 

I understand that if its my screen there is nothing that adobe can do to remedy it, and i would accept it is simply my screen, if i hadnt been able to replicate it on a different setup. 

see a very rough phone shot that illustrates what i can see on the screen with the same image side by side, as you can see one is very clearly warm vs one is cool, its not a gradual shift across the panel. If the images are not side by side and a switch between them in quick sucession there is no visual difference...

I would accept it is simply my screen, if i hadnt been able to replicate the phenomenon on other machines, and confirmed it exists visually with with a colleague. 

[TheDigitalDog]

You pass the tests so to speak.

The readings from Digital Color Meter state it's all fine. 

A few here have said the two look the same (backing up the 'measurements'). 

Not much more we can do here. 

Next test (maybe):

https://www.xritephoto.com/ph_toolframe.aspx?action=coloriq&sms_ss=blogger&at_xt=4d3437bbad19d542%2C...

There is zero evidence and the opposite this has anything to do with Photoshop. 

[D Fosse]

There is no difference. Here's an overlay with zebra-striped mask:

[Gibson Editions]

Many thanks for checking, In theory they are identical

[TheDigitalDog]

V4 profiles bring nothing to the party! I know of no software creating them that supports the PRMG. Just select V2, your life will be easier 😋

[Gibson Editions]

and what about LUT vs Gamma EOTF ?

[TheDigitalDog]

and what about LUT vs Gamma EOTF ?

By @Gibson Editions

YMMY, depending on software (making and using) LUTs. 

Mine are in the panel; better.

[Gibson Editions]

Apologies, what is YMMY? 

[TheDigitalDog]

Your Mileage May Vary. IOW (In Other Words), it depends. 

[Gibson Editions]

Basically any image i take and duplicate and view side by side shows a difference. Can anyone else repicate this on their setup. 

Currently i have been able to replicate the issue on a M1 mac with CG279X and  Intel Mac pro 6.1 with CG277 both running PS24.1

[TheDigitalDog]

Basically any image i take and duplicate and view side by side shows a difference. Can anyone else repicate this on their setup. 

Currently i have been able to replicate the issue on a M1 mac with CG279X and  Intel Mac pro 6.1 with CG277 both running PS24.1

By @Gibson Editions

I cannot. I'm using an NEC SpectraView PA271Q. M1 Mac, 24.1 version of Photoshop, the latest version of macOS. 

FWIW, I'm using a much, much better test image to compare:

http://www.digitaldog.net/files/Gamut_Test_File_Flat.tif

[Gibson Editions]

Many Thanks for having a look, I really appreciate it. I will use your test image for comparision in future. 

[D Fosse]

I still think you should check your Eizo for panel unevenness. It shouldn't happen, but it seems likely here.

And no, I don't see anything like that here on my CG2730.

[Gibson Editions]

Many thanks for checking. I will do and report back