Colour Space - SRGB or Adobe RGB?

[MDM]

1 hour ago, M.Chapman said:

Another oddity I really don't understand.....

 

If I follow my own instructions...

 

Another option is to create you own comparison images using a RAW image file that you think is demanding and should benefit from the increased gamut of AdobeRGB. Open the RAW file in LR and export an AdobeRGB and sRGB version. Then open both in PS (do not convert the profile) and toggle between them. On my monitor (which is not wide gamut) the difference between sRGB and AdobeRGB is quite subtle, to my eyes anyway. 

 

Although the difference in the images is very subtle, the histograms are quite different.....

 

The sRGB image (bottom) has the full luminance range. But the AdobeRGB version doesn't. It's one of the reasons I ended up standardising on sRGB for my workflow, so that the end result was predictable and should match what my Alamy and my other customers receive. Maybe I should have turned on soft proofing and keep swapping to AdobeRGB or sRGB before adjusting the histogram and exporting? All too complicated for me. So I stick with sRGB, simples...

 

Mark

 

 

As I said I am not going to get into a detailed argument about any of this. The points I am making are general and not specific in relation to Alamy.

 

However, I would suggest you get those two Color Checker Passport images you uploaded, open them in Photoshop without converting the profiles and then use the color sampler tool on various patches. You may not be able to see the differences although I definitely can on my wide gamut monitor but there are very significant differences between the colour patches. That in itself does not argue in favour of AdobeRGB - it simply demonstrates that there are differences. However, I will trust the colour scientists who have the equipment to accurately measure the differences in the various colour spaces and believe that AdobeRGB does offer a much wider colour gamut than sRGB. 

 

It is easy to overthink colour management. I don't need to do an exam in the theory which is complex. All I need to know is how to use it in my workflow and that is what I do and have been doing for almost 20 years now.  I started using colour management to try to get consistent inkjet prints and I am still doing that, nowadays to a much higher quality and consistency than in those days but the principle remains the same. I don't need a degree in optics to use a camera, I don't need a degree in colour science to use colour management and I don't need to be qualified as a printer to get really good prints. I just follow the recipes I have developed and adapted over the years. Of course I want a certain amount of understanding in order to control what I do  but I don't want to get bogged down in the theory as the practice is simple. 

 

I have recently started trying to figure out colour management in relation to video and that is a whole new story but I intend to use the same approach. 

Edited May 21, 2020 by MDM

[M.Chapman]

13 hours ago, MDM said:

However, I would suggest you get those two Color Checker Passport images you uploaded, open them in Photoshop without converting the profiles and then use the color sampler tool on various patches. You may not be able to see the differences although I definitely can on my wide gamut monitor but there are very significant differences between the colour patches.

Michael,

 

Thanks, you've led me further into understanding what's going on with my histogram and those colour sampler numbers. I find that histogram and RGB colour sample values are based on the RGB values in the image data before the colour profile is applied and so can't be used to measure the differences  we may (or may not) see on screen after the profile has been applied. To see this in action try the following. Create a blank image in PS with R=64, G=128, B=192 in AbobeRGB colour space. Add 1% gaussian noise (Filter>Noise>Add Noise) so you can see the RGB lines on the histogram. You'll see they are nicely spaced at 64, 128, 192, and the colour sampling tool shows values that are R=64, G=128, B=192 with a bit of noise. So at this stage, everything is as expected. This image was created in AdobeRGB colour space and is being displayed using an AdobeRGB profile.

 

Now assign (don't convert) to an sRGB profile instead. (Edit>Assign Profile..). Watch what happens to the colour shown on screen, it changes significantly - but the histogram and RGB colour sample values do not change as they are based on the underlying data before the profile is applied. ie. the histogram and colour samples don't indicate what's on screen. They are based on the RGB values before any profile is applied

 

Undo the last step (to go back to an AdobeRGB profile), and now convert the image to sRGB (Edit>Convert Profile...). Watch again what happens to the colour shown on screen (it barely changes this time) - but the histogram and colour sample values change significantly (especially the red). Conversion to sRGB is a more "legitimate" step as that's what happens when saving an sRGB image whilst working in AdobeRGB colour space.

 

But what this indicates is, if the required image format is sRGB (Alamy always end up there), but editing is being performed in AdobeRGB, the histograms and RGB values shown in PS can't be relied upon. In fact, careful adjustments to the histogram to ensure a tonal range from 0-255 are a waste of time in AdobeRGB as they will be messed up by the conversion to sRGB.

 

It's simple to test this.  Take an image with some nice rich colours. Working in AdobeRGB in PS adjust your histogram to give close to 0-255. Save as sRGB jpg ready for Alamy. Now open the jpg file in PS and look at the histogram. It may be badly clipped. Here's an example.

 

Working in AdobeRGB - nice histogram.

 

But after conversion to sRGB - significant clipping of the red channel

 

 

OK. So surely there's a simple solution? Turn on Soft Proofing for sRGB before making adjustments to the histogram? But no, that doesn't work either. The histogram doesn't change when soft proofing is turned on. Mmmmm.....

 

OK how about the sRGB images exported directly from LR. Same problem, clipped reds. But I find soft Proofing works in LR. If sRGB is set as the target profile, the histogram changes and it's clear that clipping will occur so corrective action can be taken. I wonder why soft proofing in PS doesn't work the same way??

 

IMO the bottom line is; If using PS and wanting to submit optimised images to Alamy, then adjustments have to be made whilst working in sRGB mode and sRGB images have to be submitted directly to Alamy.... If working in LR then be sure to turn soft proofing on and set sRGB as the target profile to check everything's OK before exporting as an sRGB jpg.

 

Submitting AdobeRGB images to Alamy doesn't solve the problem. The clipping will still occur when they convert the image to sRGB (I've checked this happens).

 

I've no idea if these comments/problems with the histogram apply to other image editors.

 

Mark

Edited May 22, 2020 by M.Chapman
Updated to include comments about soft proofing in LR

[Jansos]

23 hours ago, Niels Quist said:

https://discussion.alamy.com/topic/11753-colour-space/page/2/

Link worked - thanks a lot!

[Jansos]

23 hours ago, Harry Harrison said:

Would that be for Alamy or just generally? Mark Chapman did extensive tests that proved, and eventually elicited a response from Alamy confirming it, that all downloads are in the sRGB colour space, but have no colour space embedded. Anything uploaded in a different colour space is converted to sRGB. We don't know whether that conversion takes place on upload or download. 

Well at the moment I'm only uploading to Alamy but also interested from a more general quality perspective! Thanks 🙂

[Jansos]

22 hours ago, M.Chapman said:

 

For images submitted to Alamy it doesn't matter because they convert everything to sRGB and then (contrary to best practice) strip the profile. I work in sRGB because it provides me with a consistent viewing and editing experience and simplified my workflow, and the libraries I submit to take sRGB. I do however keep my RAW files so I can always go back and recover any colour information the conversion to sRGB may have lost. I don't do much printing.

 

You may find this interesting https://hubpages.com/art/SRGB-AdobeRGB-and-ProPhotoRGB-colour-spaces, there are links to some sRGB, AdobeRGB and ProPhotoRGB test images towards the end.

 

The biggest differences (on screen) IMO occur when images encoded with one profile (e.g. sRGB) are rendered using the wrong profile (e.g. AdobeRGB). This can make viewers believe the differences between sRGB and AdobeRGB are more significant than they are. 

 

Mark

Mark, Thanks for the comprehensive reply - much appreciated and to everyone else who has chipped in. You are (collectively) a very generous bunch of peeps! Thanks for your insight, advice and professionalism. I'm now off to try and get my rusty cogs processing some of the info. Any advice or which gear ratio to use? 😉

[spacecadet]

15 hours ago, M.Chapman said:

 

I just updated the links on the Granger to 16 bit TIFF versions. Did you download them? If I view in my browser, the Granger charts look pretty much identical too, but I think Google drive is preparing an sRGB preview. The true difference should be visible if you download the TIFFs and then open. 

 

Mark

Now identical.

[M.Chapman]

1 hour ago, spacecadet said:

Now identical.

 

They shouldn't look identical. If you're viewing in PS make sure you have the following colour management settings.

 

>Edit>Color settings....

>Color management policies...

      Set RGB to Preserve Embedded Profiles.

      Profile mismatches - Ask when opening

 

Reopen the TIFF files and then, if asked what to do about the profile mismatch to your working space - select "Use embedded profile"

 

You should see a noticeable differences in the distribution of the colours, and there should be some intense colours in the AdobeRGB version that appear slightly more muted in the sRGB version. The degree of extra vividness you see in the AdobeRGB version will give some insight into how much extra (beyond sRGB) your monitor can display. It's not a perfect test, but gives an insight. If the same test is done with ordinary photos the difference will be less noticeable, because most photos aren't full of colours that are at the extremes (unlike the Granger chart which does). 

 

Mark

 

Edited May 22, 2020 by M.Chapman

[spacecadet]

1 hour ago, M.Chapman said:

 

They shouldn't look identical. If you're viewing in PS make sure you have the following colour management settings.

 

>Edit>Color settings....

>Color management policies...

      Set RGB to Preserve Embedded Profiles.

      Profile mismatches - Ask when opening

 

Reopen the TIFF files and then, if asked what to do about the profile mismatch to your working space - select "Use embedded profile"

 

You should see a noticeable differences in the distribution of the colours, and there should be some intense colours in the AdobeRGB version that appear slightly more muted in the sRGB version. The degree of extra vividness you see in the AdobeRGB version will give some insight into how much extra (beyond sRGB) your monitor can display. It's not a perfect test, but gives an insight. If the same test is done with ordinary photos the difference will be less noticeable, because most photos aren't full of colours that are at the extremes (unlike the Granger chart which does). 

 

Mark

 

I was viewing in FF. I won't be doing that because I hardly use PS and I don't have any concerns about my colour profile. I thought the idea of looking was to see what the differences might be on various displays.

[M.Chapman]

28 minutes ago, spacecadet said:

I was viewing in FF. I won't be doing that because I hardly use PS and I don't have any concerns about my colour profile. I thought the idea of looking was to see what the differences might be on various displays.

 

Yes it was. But unfortunately when Google drive provides the low res previews it converts both images to sRGB before you view them in your colour managed browser. Which is why I mentioned the images need to be downloaded to make a true comparison. I apologise for the confusion. It's my fault, I should have found an image hosting site that provides "un-adulterated" image previews to make it easier to make a comparison without needing to download. I'll see if I can find a suitable hosting site. If I do I'll update the links. Maybe Dropbox works?

 

Mark

Edited May 22, 2020 by M.Chapman

[spacecadet]

47 minutes ago, M.Chapman said:

 

Yes it was. But unfortunately when Google drive provides the low res previews it converts both images to sRGB before you view them in your colour managed browser. Which is why I mentioned the images need to be downloaded to make a true comparison. I apologise for the confusion. It's my fault, I should have found an image hosting site that provides "un-adulterated" image previews to make it easier to make a comparison without needing to download. I'll see if I can find a suitable hosting site. If I do I'll update the links. Maybe Dropbox works?

 

Mark

I get the slight differences again when I do that, viewing in Picasa viewer as it happens. As you say I'm not likely to notice those differences day-to-day. I'll carry on exporting in sRGB.

Although I did macrophotograph some cherry tomatoes the other day (yes, things are that bad) and they're quite intense red-yellow-orange, so maybe I'll do a test.

[spacecadet]

I've tried it.. No difference to my eye.