File talk:Stuttgart - West - Johanneskirche im zugefrorenen Feuersee zur blauen Stunde (AdobeRGB).jpg

Test

Content bellow copy-pasted from this discussion -- Basile Morin (talk) 23:30, 24 April 2023 (UTC)

Here is a quick test everybody who is interested can try: (A) AdobeRGB file with correct profile, (B) sRGB file with correct profile, (C) AdobeRGB file with sRGB profile (intentional error). Please open (A), (B) and (C), each in full size, in three tabs or windows. Then compare the colours. If (A) looks like (B), your machine cannot display the extra colours of the AdobeRGB colour space, but correctly matches them to sRGB – fine, colour management is working. If (A) looks like (B), but with a deeper blue in the sky, then congrats, your machine can even display some of the extra colours in AdobeRGB. If (A) looks like (C), your machine does not understand colour profiles at all, but handles all RGB files as sRGB. This is what the old warning meant when speaking about “subdued” or faded colours. And if (A) looks totally different from (B) and (C), the colour management is working, but buggy. --Aristeas (talk) 19:00, 24 April 2023 (UTC)

• 
  (A) AdobeRGB file with correct profile
• 
  (B) sRGB file with correct profile
• 
  (C) AdobeRGB file with sRGB profile (intentional error)

 Question Aristeas, what do you mean by "working, but buggy"?

 Comment Interesting but this test says nothing about the intensity of the observed differences. I tried it on two different devices, and notice contrasting results.

How did you build the 2 derivatives sRGB files? The weird thing is that Photoshop does not recognize sRGB in (B) when clicking on the arrow at the bottom left of the document > Document Profile. It just says "RGB space without description". Idem (C). Online metadata viewers say "uncalibrated" for (A), nothing wrong for (B), and "sRGB" for (C). It seems to me that we can obtain a fair sRGB result from (A) by converting the color space on Photoshop and save it as a new file. Then the colors and saturation are comparable to (A), and the format standard -- Basile Morin (talk) 05:47, 25 April 2023 (UTC)

This was just a phrase for “something is going wrong due to severe misconfiguration or bugs in the software”. --Aristeas (talk) 13:40, 25 April 2023 (UTC)

The two files (A) and (B) were created directly from the raw image file with Photoshop some years ago; (C) was created just by opening (A) with Photoshop, assigning the (wrong!) sRGB profile with “Edit > Assign profile …” and saving it. To avoid any confusion due to different Photoshop versions, I have now re-created all three files directly from the raw image file with the current Photoshop version (2023, 24.4.1). (A) is the result of developing the raw image in ACR (Adobe Camera Raw) using 16 bit AdobeRGB as working colour space, the JPEG file is just the result of reducing it to 8 bit. (B) is the result of converting the 16 bit AdobeRGB data to sRGB using Photoshops “Edit > Convert to profile …” (the correct way to convert photos between colour spaces), reducing it to 8 bit and saving. (C) is the result of assigning the (wrong!) sRGB profile with “Edit > Assign profile …”, reducing it to 8 bit and saving. Now all three files should be as uniform as possible, also regarding their metadata. The small info gadget at the bottom left of each window now says “Adobe RGB (1998) (8bpc)” for (A) and “sRGB IEC61966-2.1 (8bpc)” for (B) and (C).

This was just intended as a quick test with a photo for which I had already uploaded two different versions. Even this quick test shows, on all machines and screens I have tried, cleary the strong loss in contrast and visible colour depth which occurs when software misinterprets an AdobeRGB photo as sRGB, as it is done in (C). On my own computer I see clear differences between all three files – (A) and (B) looking almost identical regarding overall contrast and color balance, but with a purer, more brilliant blue in the sky. On other machines I have tested – two Linux desktop computers with office screens, two smartphones, one running Android and one running /e/ – that difference is more or less invisible, their screen does not support that blue, but (A) and (B) look almost identical. This is how it should be when the software recognizes that (A) contains a colour profile and must be displayed with colour conversion. And on all devices I see that (C) shows a clear loss in and saturation, the blacks not being really black, and a flat, slightly purplish sky. This is a big difference. If somebody sees (A) on their screen more or less like (C), then we know that the device they use ignores colour space information completely and treats a file with AdobeRGB (or any other RGB) colour profile just like a sRGB file.

Certainly this test says nothing about the intensity of differences etc. It was just a quick test. We should make a better one which also shows more differences in the orange and turquise tones (two areas where sRGB and AdobeRGB differ even in the midtones), with a better photo from a better photographer – feel free to suggest one. Some of your wonderful sunrise/sunsets from Laos could be excellent candidates, you just would have to develop the raw image file in a wider colour space (e.g. ProPhoto or WideGamut or, at least, AdobeRGB) and then create derivativ files with different colour spaces settings. That’s it for now, I got other things to do and certainly the same is true for you. Best, --Aristeas (talk) 13:40, 25 April 2023 (UTC)

Publications

This file has been used in: Feuerseeplatz (in en). CitySeeker.com. Archived from the original on 2023-04-26. Retrieved on 2023-04-26.

See also File talk:Stuttgart - West - Johanneskirche im zugefrorenen Feuersee zur blauen Stunde (sRGB).jpg#Publications -- Basile Morin (talk) 08:47, 26 April 2023 (UTC)