Working profile

Hi All,
I am in choosing my own workflow from raw decoding to print but after read here and there about color management I’m no able to understand which working profile to choose.
I have a monitor which is wide gamut ( I profiled it and it could be) therefore surely I’d choose a profile which is not smaller than it.
I saw there are a lot of profiles perceptual and some linear…
What are the things one have to evaluate??

Cheers
Gabriele

Of all the stuff I’ve sifted through about working profiles, @Elle Stone’s articles have been the most helpful. She has also produced a collection of working profiles (linear matrix) that I use for processing, particularly the Rec2020-elle-V4-g18.icc. Here’s a link to her article about them:

https://ninedegreesbelow.com/photography/lcms-make-icc-profiles.html

and from there you can find her very informative articles.

@dafrasaga, from what I read in your post, I think you have the right basic idea, select a working profile that chromatically (color-wise) encompasses your display and any output. For what that’s worth… :slight_smile:

Hi @ggbutcher - I’m confused about the phrase “linear matrix . . . Rec2020-elle-V4-g18.icc”, as “linear” refers to a linear gamma TRC, which would be my profiles that end in “-g10.icc”. Gamma=1.8 is a sort of compromise between linear gamma and more or less perceptually uniform TRCs, which include gamma=2.2, the sRGB TRC, and the LAB L TRC. “Matrix” refers to the type of ICC profile, meaning there’s a matrix for getting from RGB to XYZ and back, instead of a LUT. All my working profiles are matrix profiles, but only the ones with file names that end in “-g10.icc” are “linear (gamma) matrix” profiles.

But yes, the Rec.2020 primaries are an excellent choice for a large color space. When processing photographs, as long as one doesn’t add huge amounts of saturation, results should be fine. But it’s always a good idea to soft proof to your monitor profile from time to time, to make sure you aren’t making saturated colors you can’t see on your monitor, and likewise to your printer profile.

As an aside couple days ago I checked to see where top-of-the-line fine art printers+photographic papers are with respect to color gamuts - wow, those printable color gamuts keep getting larger, right along with the price tag for same.

@Elle, forgive the naive question: any reason to not use a very-wide gamut such as aces as working profile?

Yes. ACES is intended as a storage color space, not a working color space. It has a high percentage of imaginary colors, and it’s very easy to accidentally create imaginary colors (drive real colors into the realm of imaginary colors), just in the course of ordinary editing.

Even in a fully compliant ACES workflow, it’s not much used as the working RGB color space. Remember, add/subtract are color-space independent operations, at least when using unbounded editing. But results of multiply/divide/raise to a power, and also operations that involve directly using channel data, are very dependent on the color space primaries. “Different” results implies the possibility of “better/worse” and indeed for most editing tasks Rec.2020 and ACES-cg produces nicer results than ACES, according to people who test such things. My own feeble testing concurs.

1 Like

Ooff, I was thinking of the “linear” chromatic bounds, when you draw the xy triangle. That’s what I get for write-on-the-run… yes, the g18 profile is not linear, TRC-wise.

Anyone please, where do you put a working profile for RT to use? I can see the folders for input and output profiles, but can’t find the working ones (Ubuntu). If I search the whole of Home for “bruce” or “wide”, for example, none come up. Or maybe RT simply doesn’t support other working profiles? - I was going to try one of @Elle 's rec2020 out of curiosity (rather than the built-in rec2020).

nowhere. working profiles are built-in and not user-modifiable. you can only select from what is there

Hmm, that makes perfect sense description-wise, though not a standard application of the term “linear” wrt profiles. It also brings up the interesting question of what shape the “footprint” on the xy plane might be for something like a LUT printer profile. I don’t have a clue, but there’s not likely to be any straight lines for a LUT profile, except maybe for a synthetic LUT profile designed to mimic a matrix profile.

Yes, Old Thing here got a bit creative with the lexicon. Never a good formula for fomenting common understanding. I’ll stick to program, here on out… :smile:

I started to raise that question in my previous post, but didn’t have the time to form it well.

Engineering-wise, it’s pretty easy to specify a set of three primaries and a white point and call it a working profile. Doing that with LUT profiles would be a bit harder. Do you know of any LUT working profiles? Now that I’ve got (some sort of) religion regarding matrix profiles, I’d like to pick one apart, if they exist.

Hmm, as a matter of fact, there are such things. Here’s one that’s been around forever (not a free license, so I can’t directly load a copy):

http://www.photogamut.org/E_ICC_profile.html

And the ICC (color.org) has several LUT versions of standard RGB working profiles, for use in a print-oriented workflow. Every time I go to look, they’ve changed the list, but here’s the link to their profiles page (again, these don’t have free licenses, and in fact you have to accept “terms and conditions” before you can download the profiles):

As of today, there are two sRGB profiles on the above page that are LUT profiles, that can be downloaded here:

I’d be very interested in what you ascertain about the sRGB “appearance” and “preference” profiles.

For anyone tempted to use these profiles, please note that they are not matrix profiles, so don’t support floating point conversions. They do support perceptual intent, for anyone who wants to experiment with compressing an image with a color gamut that exceeds sRGB (matrix) color gamut, down to the sRGB color gamut.

The color.org profiles don’t have free licenses, so read the fine print before modifying or distributing any profiles from color.org.

LUT profiles require a source color gamut, on which the intent tables are based. I don’t know what the source color gamut is for the photogamut profile. For the color.org profiles, it’s the “PRMG” (Perceptual Reference Medium gamut), which is a standardized color gamut:

See this page on how the ICC recommends to use the LUT sRGB profiles:

See this page for a discussion of limitations and benefits of using the PRMG (introduced in the V4 specs), vs using a source color gamut tailored to one’s own specific conversion and printing needs, as is the case in a V2 workflow:

http://argyllcms.com/doc/iccgamutmapping.html

The ICC provides some code and utilities that are useful for examining profiles:

http://www.color.org/profileview2.xalter

The ICC Profile Inspector (link is on the page above) doesn’t provide source code and only runs on Windows, but it does run under Wine, though I think it might be necessary to supply a dll that Wine doesn’t, or didn’t, supply - I haven’t installed or used Wine in quite a few years now so I’m going from memory. But if you have Wine installed, the Profile Inspector is very useful. I wish Linux had such a program.

Hmm, there’s also this, that seems to be new (edit, not so new on closer look, the repository was established three years ago):

and seems to include the functionality of the older sampleicc and iccXML. Guess it’s time for me to clone and build some software . . .

I did that yesterday. If you compile per their instructions, it puts the executables and libraries in directories in ~/.local/

If you keep them there, you’ll have to point your PATH and LD_LIBRARY_PATH environment variables to ~/.local/bin and ~/.local/lib, respectively.

I only got as far as dumping a few of the profiles to XML. I’m going to have to study up on LUTs and their encoding to get where I can make sense of it, says bear-of-little-brain…

Awesome! Usually I put the executables and libraries for such programs in /usr/local rather than in my user prefix, just to avoid pointing stuff to various places (being lazy about such things, and having a short attention span that precludes remembering details about what is located where . . . )

Then, just change the cmake command line to:

cmake -DCMAKE_INSTALL_PREFIX=/usr/local ../Build/Cmake

I also have a short attention span, and very well understand the need for… Squirrel!!!

For viewing icc profile I use iccexamin…is it as Profile Inspector??
I return to working profile… Ok it has to be not too larger than your display profile because you could be colors which you do not view…

Linear or perceptual ?? what is one against the other??

For matrix profiles, exiftool seems to provide all the data that the Inspector provides.

1 Like

The ICC profile inspector has a nice User Interface, that allows to see the Channel curves even for LUT profiles, and also allows to make selected changes to the profile.

Yes, exiftool indeed has the same information and also allows to make changes if one knows exactly what the required text wouldbe. But exiftool isn’t user-friendly for viewing/modifying ICC profiles. For example, one could extract the curve:

A To B0 : (Binary data 328632 bytes, use -b option to extract)

and then find a plotter to view it, but such isn’t easy or convenient.

icc_examin is a really cool utility, with nice views on the profile… I don’t have it installed at the moment (last I checked there were some compatibility issues on Gentoo). I don’t remember whether icc_examin allows to view individual channels and/or to modify selected data, but it’s a super-awesome viewer.

People have been editing images in color spaces that vastly exceed their monitor’s color gamut for many years now. The trick is to keep in mind that “what you see” isn’t always “what you get” and also to spend some time becoming familiar with your monitor’s actual color gamut. People who make prints often want to keep colors they can’t see, as long as the colors are printable. So again, the trick is to softproof to the printer profile, checking out of gamut areas, and make enough actual prints that you have an idea of what those colors will look like.

The other alternative of course is to pick an editing space that’s close to the size of your monitor’s color space. But soft proofing and familiarity with the limits of various color space gamuts will still be required for converting to sRGB for the web and for making prints.