1

I'm trying to understand color management, and have stumbled across an issue:

I've seen many people, and some tools (like the OSX system screen shot utility) that will embed the icc profile of the monitor inside the picture file. My understanding is that the embedded profile should be the icc profile of an absolute colorspace such as sRGB/adobeRGB etc, not that of the particular monitor it was displayed on.

What am I missing here? Help me find the fallacy in my reasoning:

When you design an image on your computer with a profiled display you get to see the colors in your monitor's native colorspace, but the RGB color values are stored in some other (reference) colorspace. When you save the image, you save along the information of what this reference colorspace is, so when the next person opens it, his computer will transform this reference colorspace into his own display's native colorspace and reproduce the actual color.

Why would anybody embed the color profile of his display? How would this help? Even if one did, that would require to define it against a reference colorspace; then why do it in the first place?

  • Its perfectly valid to embed the color space of your device. – joojaa Feb 5 '17 at 14:21
  • thank you @joojaa, I understand that it may be the case, but I was interested the the cause of my misunderstanding. I'll edit my question adding some more details on my reasoning so that I can help people help me debug my thought process – FotisK Feb 5 '17 at 14:34
2

All color spaces are alike, theres is nothing special about the standard RGB color spaces as opposed to the device colorspaces. They are exactly the same, they both describe conversion form that space into an absolute color space (Neither sRGB or Adobe RGB are absolute spaces). You can then convert back form that absolute color space to any RGB space. Normal caveats apply. You would do this for a scanner, but you could do the same for a monitor.

Why standard spaces? Theoretically it gives you a target to calibrate against. So that you can make 2 monitors display the same even without color transformations (this is how web is supposed to work). It would also be possible to factory calibrate things. In practice this does not work, except perhaps for sensors, monitors not so much.

On the other hand if you are going to color transform you might as well utilize the entire size of your monitors space. Then you would just rely on profiling your device and convert to that space. Its a perfectly valid strategy, theres no need to calibrate as long as you have a profile. Un-profiled stuff is now wrong but theres no way to know what they were supposed to be anyway. And whaen you monitor travels in its color reproduction capability due to age and use you get different results but yes still valid (same would apply to standard spaces unless your monitor is significantly larger in size and can crop off some parts).

  • thank you @joojaa - I'm not sure I understood your points on the 2nd and 3rd paragraphs though (I'm not a native speaker). Let me summarize what you say and see if I got it right: On the 2nd paragraph you state that embedding/referencing a standard colorspace (such as sRGB/adobeRGB, right?) in reality is a moot point since almost nobody has devices calibrated into that space out of the box. On the 3rd paragraph you make an argument in favor of embedding your display's color icc profile into the image, since this means fewer conversions (preserving the original gamut/gradation) – FotisK Feb 5 '17 at 21:18
  • 1
    @FotisK It is a tradeof, neither is better, they both have pros and cons. So yes there are benefits to both approaches that the other does not have. For example it should be obvious that the benefit of using the entire available colorspace of your monotor may not be such a great benefit if your monitor is very special and generates lots of out of gamut situations and or if the monitor is out of gamut most of the time. But – joojaa Feb 5 '17 at 21:29
  • I think I'm starting to understand! But what would be the problem when you embed the profile of your (let's say) unnaturally wide gamut display? The receiver of the file would have to transform this to his Display's native colorspace (with gamut/gradation loss), but what other options did you have? If you embedded a standard colorspace such as sRGB, wouldn't you be baking-in the transformation and limiting the gamut/graduation yourself? What would the benefit be in this case? – FotisK Feb 5 '17 at 22:13
  • 1
    @FotisK depends on what you do, your software has working space for this reason. If the image is intended for web use it should be sRGB. If not then you can revel in the knowledge that 99,999% of people will not see the colors as you intended, they have no means to do so. The benefit for you is that you can fit most cmyk and rgb spaces in your view. – joojaa Feb 6 '17 at 5:13
  • yes, that makes sense! – FotisK Feb 6 '17 at 8:07

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.