Assigning an ICC Color Profile when Printing

[lacerto]

(...)

[R C-R]

5 hours ago, thomaso said:

@R C-R, you posted some days ago for the "Color Matching" menu: "I get a second window that offers a choice of every ColorSync profile installed on my Mac, like this: (screenhot)". Are you sure it's "any" profile? Your screenshot shows RBG only, too.

I think it is as @Lagarto mentioned: it depends on the printer ... somehow. Since the only color printer I have installed in System Preferences > Printers & Scanners is my old Canon MP620 (other than the fake one that does not actually work), there is not much testing I can do for this.

However, I did find at the bottom of the AP Color management help topic in the Installing ICC color profiles section instructions for assigning (associating?) an installed color profile with a printer. The steps are essentially the same as in the Change a device’s color profile using ColorSync Utility on Mac Apple support document.

Basically, on Macs this is done by selecting the device (the printer) in the utility, clicking on the "Current Profile" button & choosing "Other." From there you should be able to navigate to any folder that has icc or icm ColorSync profiles in it & assign that profile to the printer.

Since those profile files could be scattered all over your system, to make it easier to see them all what I suggest you may want to do is to type "icc" in the search box at the top of the window, wait for the "Kinds" submenu to pop up, select that, & then (important?) click on the plus button to add a search criteria set to System Files > are included:

On my Mac, this finds tons of profiles, including those installed in application support folders for apps like Raw Therapy & Graphics Converter, many of which will not be suitable for printed output, so it is up to you to choose wisely. Do not blame me if it all goes wrong -- consider this as an untested, try-at-your-own-risk, experimental suggestion, so take all necessary precautions if you try it.

That said, on a lark I assigned a very old profile I found this way, the EPSON Stylus COLOR 740 Standard one buried in the /Library/Printers/EPSON/InkjetPrinter2/ICCProfiles/SC740.profiles/Contents/Resources folder, to my fake printer. After restarting AP, it showed up as a Soft Proof Adjustment choice, as the default ColorSync profile in the Print dialog if I choose to print from the fake printer, & as a Document menu Assign or Convert choice -- IOW, everywhere in AP that users can make profile choices.

5 hours ago, thomaso said:

Also in Affinity the "Assign ICC Profile" appears strange: here, too, I get offered only profiles of the current document's color space, while with the "Convert" command I have free choice.

I have never been very clear about the difference between Convert & Assign in Affinity, but I think it is similar to the Assign Profile & Apply Profile choices in Apple's ColorSync Utility, as explained in the Modify image colors in ColorSync Utility on Mac support article:

Quote

• Assign Profile: Assigns the ColorSync profile for an image. ColorSync Utility does not modify the image saved in the file; it changes only the ColorSync profile for the image.
• Apply Profile: Modifies the pixels in an image to match the new color model and ColorSync profile, then assigns the image’s original ColorSync profile to it.

BTW, since all the profile setting stuff was done with the macOS ColorSync Utility app, it works in all my apps that use the standard macOS print dialog, including Preview, Safari, & so on.

[panelson3]

On 6/17/2020 at 2:22 PM, R C-R said:

I have never been very clear about the difference between Convert & Assign in Affinity, but I think it is similar to the Assign Profile & Apply Profile choices in Apple's ColorSync Utility, as explained in the Modify image colors in ColorSync Utility on Mac support article:

The difference between Assigning and Converting is confusing and, yet, pretty straight forward. Typically, you would Assign a profile when you open a file that does not have an embedded ICC profile, and you are unsure of what color space the image originated in. Assigning is like “trying on” a profile because it does not convert the color data in the file, but it will render to the screen the image in the Assigned color space. This gives you the opportunity to try different color profiles to see if you can either identify the unknown color of the image or find a color space that is pleasing. I usually find that the first profiles that I test are working spaces like sRGB or AdobeRGB since they are the most commonly used. If I can’t find what I like, I find a working space that looks the best, and then I Convert the image color to that color space.

Converting, unlike Assigning, actually changes the data in the file, so this can be a bit dangerous if you aren’t careful. In my photography workflow, I put all of the color data that comes off of my camera (in a raw file) into the largest color working space I can, ProPhotoRGB. This way I am retaining as much of the color that the camera captured as I can. I edit the photo in ProPhotoRGB. When I am done editing and ready for output, I’ll convert the data depending on the output destination. If I am going to deliver the file to a printer service, I will convert down to the color space that they require. It is most often AdobeRGB. If it’s a book, I will convert to the CMYK color space that they require, like a generic US Web Coated (SWOP) or something like that. Often, CMYK printers will have a specific custom space that they want you to convert to, like blurb’s custom color space that is some kind of derivative of US Sheetfed. If I am sending my image to the web, I always convert to sRGB.

In my case, the color workflow is always one of managing the color conversion from capture device, my camera which captures a lot of color, down to a working space so that I can work on it, to an output space. The output gamut is almost always smaller than my working space. So the color workflow is really a process of controlling how much color you are throwing away as you move through it. Once you have thrown the color away (and save the file) you can’t get it back. If you take a file that is in a large color gamut like ProPhotoRGB and convert it down to sRGB which is a much smaller color gamut, everything that was outside sRGB is converted down to sRGB. If you then convert back to ProPhoto, you will not get the colors back that were originally in ProPhoto. You will simply have the sRGB color inside the ProPhotoRGB color gamut.

[thomaso]

13 minutes ago, panelson3 said:

Assigning is like “trying on” a profile because it does not convert the color data in the file, but it will render to the screen the image in the Assigned color space. This gives you the opportunity to try different color profiles to see if you can either identify the unknown color of the image or find a color space that is pleasing.

I would expect that I am allowed to "try on" the same amount of profiles as I am allowed to convert to. But "Assign" offers less than "Convert": it only offers profiles in the same space, so in RGB only RGB profiles. Whereas I get full choice when converting. Why?

So, "Assign" as kind of "try on" forces me to convert from RGB to CMYK to be able to "try on" various CMYK profiles. That's kind of destructive "try on", isn't it?

[panelson3]

1 hour ago, thomaso said:

I would expect that I am allowed to "try on" the same amount of profiles as I am allowed to convert to. But "Assign" offers less than "Convert": it only offers profiles in the same space, so in RGB only RGB profiles. Whereas I get full choice when converting. Why?

Hello @thomaso, Great question!

If you are opening an untagged RGB file, you don’t know what RGB color space it’s in because no RGB profile was embedded with it, but you know that it’s colors are being represented by the RGB color model. One of the points of Assigning is to try to determine what color gamut your RGB (or CMYK, if the file is in the CMYK color mode) file is in or which color gamut is going to give you the most pleasing results. That’s why you only see RGB profiles for files using the RGB color model, CMYK profiles for files using the CMYK color model, etc. I hope that makes sense.

1 hour ago, thomaso said:

So, "Assign" as kind of "try on" forces me to convert from RGB to CMYK to be able to "try on" various CMYK profiles. That's kind of destructive "try on", isn't it?

So, again, think of Assigning as a way to determine the color space of an untagged file that you have just opened. Once you have figured that out or found a profile that you prefer, you can then Convert to preferred color space. Let’s say it’s an RGB file, but no RGB color profile was embedded with the file. You open it and have no idea what RGB color space your file is in. So you Assign sRGB. The color shifts to something you think is accurate, ie: you like what you see, and assume that the file probably originated as an sRGB, but the sRGB profile was never embedded with it. You are happy with the color, so you now Convert, shifting the color data in the file to sRGB. The RGB file is now in the sRGB color space. Now that you have done this conversion, you wonder what the file would look like if you were going to send it to a print service that only takes CMYK. This is when you soft proof. You set up a soft proof adjustment layer with the printer’s preferred CMYK color profile, and now you can get an idea of what your sRGB color will look like when the printer outputs it on his CMYK press.

I hope this answers the question.

[lacerto]

(...)

[lacerto]

(...)

[R C-R]

3 hours ago, Lagarto said:

So IMO the correct procedure would be to convert colors to sRGB (if that is what the service provider expects and which does not narrow down the target color space) and just see that the colors look right, rather than fiddling with off-gamut colors manually using soft proof adjustments (especially in Affinity apps where this feature does not work appropriately).

Can you explain more about how in your opinion soft proof adjustments in AD do not work appropriately? It seems at least more or less consistent with what I have read from various sources that discuss how to use soft proofing, for example here or here.

From what I can tell, the biggest difference is that in AP soft proof adjustments are implemented as an ordinary adjustment layer (I suppose so that as in the AP help topic it could be used on selected layers for artistic effects?) while in Photoshop it is implemented as a view mode.

[panelson3]

6 hours ago, Lagarto said:

No, they use printer color profiles when outputting to the printer, but assume typically the input to be in sRGB (which it is for the majority of customers using these kinds of services meant for general public) and probably strip off any embedded profiles (at least if there is any kind of automated delivery). So they offer their proofing profiles so that the customers can realistically see (in apps like Photoshop and Lightroom) what can be expected when they get their photos printed. So IMO the correct procedure would be to convert colors to sRGB (if that is what the service provider expects and which does not narrow down the target color space) and just see that the colors look right, rather than fiddling with off-gamut colors manually using soft proof adjustments (especially in Affinity apps where this feature does not work appropriately). If colors are not mapped as wished by the conversion, then the image can be manually adjusted within the expected input color space e.g. to bring out specific tones and color areas using manual fine tuning, but just trying to do manual color management and do something that is much better done by a profile conversion is not sensible.

This is the way that I understand it as well.

[lacerto]

(...)

[panelson3]

@Lagarto, do you know what color conversion engine (CMM) the Affinity apps use to convert color? Do you think that they rely on the color engine provided by the operating system?

Also, in the soft proof adjustment layer, why is Absolute Colorimetric the default? I’m assuming this is so you can see the paper white, but that doesn’t help give an accurate color rendering if your output is going to be converted using Perceptual or Relative Colorimetric.

[R C-R]

1 hour ago, Lagarto said:

As different rendering intents do not cause different color areas to be flagged as out-of-gamut, I think the feature is broken at the moment, or only partially implemented.

Why wouldn't different rendering intents cause different color areas to be flagged as out-of-gamut in a soft proof? That seems consistent with what I have read, for example in the Cambridge in Colour reference I just cited. Isn't that what is being illustrated in the diagram on that page comparing how Relative Colorimetric crushes all the out-of-gamut colors to the extremes while Perceptual does not?

 

1 hour ago, panelson3 said:

... do you know what color conversion engine (CMM) the Affinity apps use to convert color? Do you think that they rely on the color engine provided by the operating system?

From Cambridge in Colour's Color Space Conversion tutorial there is this:

Quote

Even though perceptual rendering compresses the entire gamut, note how it remaps the central tones more precisely than those at the edges of the gamut. The exact conversion depends on what CMM is used for the conversion; Adobe ACE, Microsoft ICM and Apple ColorSynch are some of the most common.

As I understand it, on Macs AP uses ColorSync as the CMM (Color Management Module), as do other Mac apps that do not provide their own built-in CMM (like Adobe ACE). I suppose on Windows the equivalent would be the Microsoft ICM CMM.

 

1 hour ago, panelson3 said:

Also, in the soft proof adjustment layer, why is Absolute Colorimetric the default?

Just a guess, but maybe it is because, also from the Cambridge in Colour tutorial I just cited, there is this:

Quote

Relative colorimetric skews the colors within gamut so that the white point of one space aligns with that of the other, while absolute colorimetric preserves colors exactly (without regard to changing white point).

Of course, if you want to use some other rendering intent in AP you could create a new preset that uses that intent with the proof profile of your choice, with or without the black point and/or gamut check options enabled.

As mentioned in one of the AP help topics, you could even stack several different soft proof adjustment layers in one document & turn them on one at a time to see how they differ.

[R C-R]

FWIW, I stumbled on https://helpx.adobe.com/photoshop/kb/adobe-color-management-module.html, providing separate downloads of the Adobe CMM component of the Adobe Engine for Mac & Windows systems. According to one of the FAQ's on the Read Me page:

Quote

The Adobe CMM takes the ACE color conversion engine and packages it into an OS compliant CMM that can be used by applications that support external CMMs using the Windows ICM2/WCS and Mac OS® X ColorSync APIs.

I have no idea if this can be used with either the Mac or Windows version of AP.

Edit: since it is from 2007, I doubt that it can be used with any app running a modern version of either OS. I mentioned it only because the info in the Read Me file might in some way be relevant to this discussion.

Edited June 19, 2020 by R C-R

[lacerto]

(...)

[panelson3]

41 minutes ago, Lagarto said:

a) Here is an actual conversion done from sRGB color profile to Canon Pro 9000 Mark II Seriers Glossy (üsing Adobe ACE; using Microsoft ICM produces more or less identical conversion):

Adobe ACE, and Microsoft’s ICM and Apple’s ColorSync, and any other CMM created for ICC Color Managementn is going to have to adhere to the ICC’s specifications for performing a color conversion. It would stand to reason that the results you see are very similar. That being said, if you dug deep enough you might be able to find some differences, but they may be negligible.

45 minutes ago, Lagarto said:

b) Here are the out of gamut warnings (shown in gray) between these two profiles if using Absolute Colorimetric and Relative Colorimetric rendeing intents (no difference in color gamut warning, and offering "Black point compensation" as an option in Absolute Colorimetric is also wrong, but it does not matter as it does not have any effect anyway)

As you know, the color gamut warning is going to indicate the out of gamut colors when comparing the source color space with the destination color space (sRGB to Canon). Because the rendering intent is a method for converting out of gamut color to in gamut color, no matter which rendering intent you choose, you will not see a change in the gamut warning. The gamut warning is simply showing you the colors that are outside the destination color space.

There are plenty of sources online that provide definitions of rendering intents. I borrowed this from John Paul Caponigro...

“Perceptual
A perceptual rendering intent preserves the overall color appearance by changing all colors in the source space to fit the destination space. The perceptual rendering intent is favored for images that contain many out-of-gamut colors.“

It compresses the colors into the destination color space to best maintain tonal and color relationships.

“Relative Colorimetric
A relative calorimetric rendering intent maps the white of the source space to the white of the output. It reproduces in-gamut colors exactly and clips out-of-gamut colors to the nearest reproducible color. The relative calorimetric rendering intent is a good choice for images where more of the colors are in-gamut than out-of-gamut.

Absolute Colorimetric
An absolute colorimetric rendering intent differs from relative colorimetric because it doesn’t map the source white to the destination white. It reproduces hues absolutely. If the source is a clean white reproduced on yellow paper the result will be a yellow white. If the source is a cool white reproduced on a warmer paper, cyan ink will be used to simulate the cool white of the source. Th absolute colorimetric rendering intent is intended for cross-rendering simulations of output condition with another.”

This is best used for proofing.

“Saturation
A Saturation rendering intent converts saturated colors in the source space to saturated colors in the destination space. It favors reproducing vibrant colors and will do so at the expense of reproducing hue or luminosity accurately. The saturation rendering intent is useful for reproducing graphics with high color impact.

Here’s the bottom line. To make the highest quality prints possible, choose either relative colorimetric or perceptual. As results vary from image to image, softproof an image to choose which rendering intent is best for it, before you print it.”

[lacerto]

(...)

[R C-R]

9 hours ago, Lagarto said:

I do not understand why you do not run these kinds of tests yourself.

Because as I mentioned earlier, I only have one color printer to test with. I also do not have Photoshop to compare the results with, & because I use a Mac, only ColorSync profiles to experiment with.

One thing I noticed from my experiments with Apple's ColorSync Utility described above is that all my printer-specific ColorSync profiles (including the one I tricked into appearing for the EPSON Stylus Color 740 I no longer have) include a "gamt" tag, described in the ColorSync Utility as an "8-bit, PCS to gamut check table."

I have no idea if or to what extent this has any effect on out-of-gamut warnings in the Mac version of AP. Ditto for the various 'intent' tags & the "media white-point tristimulus" tag -- the various ColorSync profiles vary greatly in the number, type, & (byte?) size of tags they include. I am not even 100% sure that the Mac version of AP uses ColorSync profiles alone for color management; or if or to what extent 8 bit vs. 16 bit images or how far out-of-gamut some colors might affect any of this.

Because there is so much guesswork involved, particularly about the possible differences in the Mac & Windows versions, & because my understanding of such things is limited at best, as I mentioned earlier I am just adding things to the topic I hope are somehow relevant & might give someone else a clue about how it all works. 

[thomaso]

Sorry I can't support with tests, too. I still struggle with understanding unambiguously a correct setup, e.g. in my idea (not: experience) assign a profile and soft proof would cause the same but with different UI only. Also I still need to find out why a print profile doesn't replace the need for soft proof + image data manipulation simply by converting according to its optimum and the print device + paper properties. And, sorry again, I still did not take my time to read the related answers of you 3 guys in all their details, attracted/distracted by various articles here and there. ... But this one might interest Lagarto, too:

10 hours ago, Lagarto said:

Affinity Photo based out-of-gamut view would certainly be a poor starting point

It might comfort you that also PS and LR apparently show out-of-gamut overlay with different results, as shown in 2015 by Andrew Rodney, with his final recommendation not to trust out-of-gamut overlay in both apps as soft proof information but to prefer your eyes instead:

 

[thomaso]

39 minutes ago, BofG said:

Now say one prominent colour is out of the print gamut, and just makes the design look off. You can create adjustment layers (...)

I would expect that I don't need to do color manipulation myself because the output profile (f. print or screen) is quite aware of its boarders (limits, 'virtual fence') and therefore would by itself and automatically, during conversion, move the out-of-gamut colors to a position inside their gamut. Solely instructed by my render intent selection.

Like it does work for document color space conversion from RGB to CMYK, where the profile causes out-of-gamut colors being moved inside its limits of CMYK. There I also do not manually manipulate too vivid greens or pinks with an adjustment layer or sponge brush to make them fit to the printing inks cyan and magenta.

[R C-R]

FWIW, here are a few things I have gleaned from the info included in this discussion & from other web sources (particularly the tutorials on the Cambridge In Colour web site) that seem relevant here:

1. Soft proofs are useful for approximating what output to a different device (particularly to a printer) will look like.
2. Rarely if ever is this 100% accurate.
3. Soft proofs do not perform any color correction.
4. For soft proofs to be of any use at all, the chosen profile must closely match the characteristics of the output device. For printers, this must include the characteristics of the ink, paper, etc.
5. OOG overlays are useful only as very crude indicators of what might be out of gamut, quite possibly not even for that, are probably implemented differently in different apps, & in any case are of little to no use in indicating how far out of gamut a color might be.
6. The profile & rendering intent together determine what the soft proof will look like. This is what is intended to be used as a guide to decide if or to what extent manual adjustments are necessary to get the appropriate or desired output.
7. There is no "one size fits all" answer for this. Some images may need little or no manual adjustment. Others may respond well to certain adjustments, but which (if any) work well will be strongly dependent on image content & the desired effect, which may be to achieve some artistic or stylistic effect rather than a technically 'correct' one.

The short version is (or so I think) using & interpreting soft proofs is an acquired skill, one that I am not even remotely close to mastering (& probably never will). 😢

[R C-R]

1 minute ago, BofG said:

All soft proofing does is give you that view on the output at design time, what you do with that view is up to you.

I think that sums it up quite well.

[panelson3]

5 hours ago, R C-R said:

I think that sums it up quite well.

Absolutely agree.

[panelson3]

For those of you who are printing with the Affinity products, are you happy with the results you are getting?

[panelson3]

I forgot to ask this with my last post. During your time on this forum, have you ever seen anyone from Serif/Affinity respond to a question or comment on a post?

[R C-R]

1 hour ago, panelson3 said:

I forgot to ask this with my last post. During your time on this forum, have you ever seen anyone from Serif/Affinity respond to a question or comment on a post?

Often.