red messing with my head

[stuck]

I have a photo that features a toddler wearing red leggings. Sorry, I can't post it here because although I had permission to take the picture, I don't have permission to make it public.

The SOOC JPEG is from Canon's 'standard' picture style so it's an sRGB image. When viewed with my monitor in sRGB mode, the red leggings don't look right, the red is dull. When I switch my monitor to aRGB mode the red looks 'better', more vibrant, and for what it's worth, nearer how I remember the colour.

When I view two TIFFs side by side, both processed from the RAW, in DxO PL5 (which uses a aRGB as it's 'native' working space') but one exported with an aRGB profile and the other with an sRGB profile then:

• with my monitor in aRGB mode there's not a huge difference between the reds in each image, in both images the red look about right.
• with my monitor in sRGB mode, again there's not a huge difference between the reds in each image but in both images the red is dull.

Please can someone try to explain, in words of one syllable why the red in both the aRGB and the sRGB TIFFs look more or less the same and OK in aRGB mode but although still the same, dull in sRGB mode.

Thanks

[DanK]

Check your settings. On all the Canon cameras I have had, the setting for color space is independent of the setting for picture style. So it's not necessarily correct that the color space is sRGB because you used the standard picture style.

So re the first question: I'm guessing that you took the JPEG in Adobe RGB, so it looks wrong in sRGB.

[Manfred M]

It could be a number of things; the most likely is the use of Perceptual rendering intent, which will cause changes in all the colours, depending on how much it has shifted the hues of the out of gamut colours. This colour space preserved gradations at the expense of colour accuracy. Using the Relative Colorimetric rendering intent could give you a "better" result. It could also be the device ICC profile and how the actual colours have been mapped when the profile was created

What could be happening is that the colours are being remapped and the sRGB colours are being shown as Adobe RGB values.


This is a complex issue and it is hard to figure out exactly what is happening without access to your system.

[stuck]

Check your settings. On all the Canon cameras I have had, the setting for color space is independent of the setting for picture style. So it's not necessarily correct that the color space is sRGB because you used the standard picture style.

So re the first question: I'm guessing that you took the JPEG in Adobe RGB, so it looks wrong in sRGB.

Sorry, my bad. You are correct, Canon Picture Styles do not dictate the colour space, which is a separate setting. However, the camera definitely is set to sRGB, i.e. the SOOC JPEG is not in aRGB.

...the use of Perceptual rendering intent...

How would I check that? As far as I can see there's nothing in DxO PL about rendering intent.

Thanks again

[Manfred M]

Another strike against PhotoLab; it seems that they only support Relative Colorimetric rendering intent. Add to that the working colour space being AdobeRGB and no possibility of working in L*a*b* colour space. I did not upgrade to PhotoLab 5 and it looks like my decision was the right one for me. The main reason I used it in the first place was the Prime noise reduction and there are now other alternatives out there.

[stuck]

I'm sorry again but I don't follow you. First you said:

...the most likely is the use of Perceptual rendering intent... ...Using the Relative Colorimetric rendering intent could give you a "better" result...

but now you've said:

...it seems that [DxO PL5] only support Relative Colorimetric rendering intent...

Which appears to contradict your first comment. Did you actually mean to say that is seems PL5 only does Perceptual?

[Manfred M]

No, I'm saying that from what I can tell PL5 only seems to support Relative Colorimetric. This suggests that the rendering intent choice is not what is impacting the result you are seeing. Other issues including the ICC profile used or mapping of colour values from one colour space to another could be the issue.

What you are seeing could be caused by a number of different factors; one of which has now been eliminated. If you can't change the rendering intent, that cannot be the issue.

[stuck]

No, I'm saying that from what I can tell PL5 only seems to support Relative Colorimetric. This suggests that the rendering intent choice is not what is impacting the result you are seeing. Other issues including the ICC profile used or mapping of colour values from one colour space to another could be the issue.

What you are seeing could be caused by a number of different factors; one of which has now been eliminated. If you can't change the rendering intent, that cannot be the issue.

OK, thanks, sorry for being slow.

The ICC profiles I used for the two TIFFs are the ones available in PL5, called 'AdobeRGB' and 'sRGB'. I would hope these profiles are 'correct'.

I'll thank you again and leave this topic there since, as you noted in your first reply:

This is a complex issue...

[stuck]

...I'll thank you again and leave this topic there...

OK, so I lied but by way of extra info I came across this statement from a member of DxO staff:

During image export, we use perceptual rendering intent and no black point compensation.

As a side note, in the Print dialog, under Windows, you can choose the color intent.

here: https://feedback.dxo.com/t/convertin...profile/7798/3

[AaronRoberts]

My best guess here is that your monitor is what’s making the difference. A few basic but important things:

a) sRGB is a smaller color space than Adobe RGB
b) Pure red (for instance) as represented by RGB values 0,255,0 is brighter and more saturated in Adobe RGB than it is in sRGB
c) Assigning a color space is different from Converting a color space. When you “assign” a color space, you accept the numerical RGB values and show the color that corresponds to it. When you “convert” a color space, you accept the color itself and re-map the numerical RGB values to new ones that correspond to the original color.

When you display the sRGB photo on your monitor in “sRGB” mode, the look you get is most likely accurate, in that it represents what’s actually in the file. When you display the same sRGB photo on your monitor in “aRGB” mode and it appears more vibrant, this is tantamount to saying that your monitor is “assigning” Adobe RGB and not “converting.”

When you export from PhotoLab, either in sRGB or in Adobe RGB, PhotoLab is clearly “converting” to the chosen color space. This means that the exported photos are meant to look as close to the same as possible, irrespective of the color space you chose. It’s your monitor setting that is making them look either dull or vibrant.

Clearly, you prefer a more vibrant rendition of the photo. This is something you should accomplish in your editing software before you export. When exporting, the decision as to which color space to use is really dependent on who will eventually view the image, and whether they are likely to have software and a monitor that is capable of (i) taking a color space into account, and (ii) displaying Adobe RGB properly. Because this is usually not the case, most folks usually recommend exporting to sRGB even when you edit in a much larger color space.

[Manfred M]

In my trawling the web, I found this article that covers this topic quite well, but is a bit heavy on the tech side:

https://color-image.com/2012/06/comp...d-srgb-colors/

I had been thinking along the chromaticity coordinate side and based on the article, this is clearly not the right approach with the reasons outlined here and I suspect this is the root cause of what you have been seeing.

[stuck]

...
c) Assigning a color space is different from Converting a color space. When you “assign” a color space, you accept the numerical RGB values and show the color that corresponds to it. When you “convert” a color space, you accept the color itself and re-map the numerical RGB values to new ones that correspond to the original color...

Thank you, this is helpful to me.

this article that covers this topic quite well, but is a bit heavy on the tech side

Thank you too. The details of the matrix maths are beyond me but I think I follow the general thrust of the article, it too is helpful to me.

[dje]

When I view two TIFFs side by side, both processed from the RAW, in DxO PL5 (which uses a aRGB as it's 'native' working space') but one exported with an aRGB profile and the other with an sRGB profile then:

• with my monitor in aRGB mode there's not a huge difference between the reds in each image, in both images the red look about right.
• with my monitor in sRGB mode, again there's not a huge difference between the reds in each image but in both images the red is dull.

It sounds to me like you need to change your display profile over to sRGB when you change the color mode of the display to sRGB. I'd say it's set to the native monitor color space as a default and this is probably close to Adobe RGB. This would explain why both images look right when you are in Adobe RGB mode. The display profile is used by the Color Management System to translate between the color space of the image and that of the display. To change the profile, open the Windows Control Panel and select Color management.

Dave

[BobGilbody]

I found this series of posts as useful as having quantum physics explained to me by way of the manner in which goldfish swim anti-clockwise in a glass bowl.
Yesterday I realised I know nothing about the different kinds of photo files. I doubt I ever will. Things advance at a rate that means by the time I get to where I wanted to be everyone else will be somewhere else.
I am not saying the discussion is in any way redundant, rather I'm saying that it is not a fatal flaw to know nothing about certain things. What does annoy me is that Canon can make amazing cameras yet fail to provide a manual that allows a senile amateur like me to know what is going on when I take a photo. The R7 does not need use the battery it comes with. It works fine with the same batteries I used in my 80D and do use in my 5Ds. The R7 can use Canon speedlite 600EX-RT and Canon speedlite 580EXII. By using these speedlites it will affect the waterproofing but then if it is raining I don't go out picture taking. Topaz Denoise AI does not handle Raw files from an R7 but will do soon ??? Just pay us $70 for the latest upgrade. Then again Topaz Sharpen AI does, maybe, perhaps, I no longer profess to know.

[stuck]

...rather I'm saying that it is not a fatal flaw to know nothing about certain things...

I agree and this thread I started is an example. I've known about different colour spaces and profiles for a long time but until now I've stuck with the simple approach of always working in sRGB. Initially, even though I had an AdobeRGB compliant monitor, because I knew it was not fatal to do that and there were other more basic things to learn about. Next I was forced to work in sRGB because I had to switch to a different monitor that did not cover the AdobeRGB gamut. Now though, I'm back on a monitor that can handle AdobeRGB and while I'm not in the same league as people here like Manfred and Dan, I do like to print up to A3 so it seems to me it is worth my effort to get my brain around these sort of things so that I can better see on screen what will come out of my printer.

I've a long way to go but I am learning.