Paper color from ICC profile?

[hugowolf]

Is it possible to find/extract the RGB values, in a given color space, for a particular paper from the ICC profile for that paper in Photoshop?

Brian A

[Schewe]

Is it possible to find/extract the RGB values, in a given color space, for a particular paper from the ICC profile for that paper in Photoshop?

Readings in RGB from within Photoshop? Not that I know of…Lab values from within a profile reading software, yes. ColorThink and ColorSync utility should be able to give Lab values of the paper''s white point.

[digitaldog]

Is it possible to find/extract the RGB values, in a given color space, for a particular paper from the ICC profile for that paper in Photoshop?
Brian A

Take RGB data, convert using the profile based on the rendering intent you desire, ideally 1 pixel per color value. Then just read the values in Lab or otherwise using the color sampler set to 1x1.

[MarkH2]

Is it possible to find/extract the RGB values, in a given color space, for a particular paper from the ICC profile for that paper in Photoshop?

Brian A

Are you trying to see what the paper's white point is, for example, as contained in the ICC profile, using just Photoshop?  If so, convert to the paper ICC profile using relative colorimetric rendering, then convert back to a color space using absolute colorimetric rendering intent.  You can then see the paper's white point in Lab and RGB.

You can also extract the paper's ICC profile data using the free program "ICC Profile Inspector" from the ICC website.  Gets you  X, Y, and Z which you can convert to Lab (and then to RGB in space of your choice). 

[Ernst Dinkla]

Is that approach reliable given that many profiles are made with UV-cut spectrometers? For OBA papers it has to be seen whether that aspect is represented in the profile's paper white data and if present it can be either from a UV enabled spectrometer or, more common, from a software extrapolation of data beyond 420 NM.

Edit: tried the method "rel col > printer profile > abs col > color space" and compared with the SpectrumViz Lab numbers and the results are all over the place. With the ACE color engine in Photoshop there is some consistency but it indicates what I write above, many OBA papers fall back to b 0 where they should be negative. With the ICM color engine the results are way off to warm.


--
Met vriendelijke groet, Ernst

http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
July 2013, 500+ inkjet media white spectral plots.

[digitaldog]

Is that approach reliable given that many profiles are made with UV-cut spectrometers?

It is if you want to see the values the ICC profile believes to be correct for the paper and thus converts to.

[MarkH2]

Ernst,

There are two questions at play here that you address.  The first: can you extract the ICC Profile’s white point in Photoshop using my method of "rel col > printer profile > abs col > color space.”  And second: do ICC profile white points adequately represent paper white, especially for papers with Optical Brightening Agents?

My understanding of the definitions of relative colorimetric and absolute colorimetric conversion says my method of extraction should work.  I tried this a few years ago with a small sample and the results agreed closely with the white point contained in the ICC profiles.  I repeated this today with Epson R3000 profiles for Epson Velvet Fine Art Paper, Epson Premium Luster, Epson Exhibition Fiber Paper, and Hahnemuhle Fine Art Baryta; the Lab and RGB results for both ProPhoto RGB and sRGB agree closely with the profile values.  Using Adobe ACE in Photoshop.

What I was simply trying to see was if I could extract in Photoshop the profile white points.  My sample size is surely not sufficient to draw broad conclusions, but it worked in my cases.   My assumption is that Photoshop and Lightroom do something like the absolute colorimetric conversion using the profile’s white point when they show paper white during softproofing.

If I understand your method, you are comparing the ICC white point with your own spectral measurements of the paper (based on the extensive work you have done at SpectrumViz!)

When you say “ tried the method ‘rel col > printer profile > abs col > color space’ and compared with the SpectrumViz Lab numbers and the results are all over the place” I’m not sure if you’re saying the method does not work or that the white points in the profile are not correct.  Perhaps you’re saying both.

Ah, I see that Andrew Rodney  agrees my approach is reliable.  That settles it for me!

I think my method addresses what the op was asking.  You have given us more to think about: you may be able get the profile’s white point, but is it optimum, especially when OBAs are employed?

Mark

[Ernst Dinkla]

It is if you want to see the values the ICC profile believes to be correct for the paper and thus converts to.

Correct, my  thoughts too. Maybe Brian wants that, maybe not.

--
Met vriendelijke groet, Ernst

http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
July 2013, 500+ inkjet media white spectral plots.

[Ernst Dinkla]

What I was simply trying to see was if I could extract in Photoshop the profile white points.  My sample size is surely not sufficient to draw broad conclusions, but it worked in my cases.   My assumption is that Photoshop and Lightroom do something like the absolute colorimetric conversion using the profile’s white point when they show paper white during softproofing.

If I understand your method, you are comparing the ICC white point with your own spectral measurements of the paper (based on the extensive work you have done at SpectrumViz!)

When you say “ tried the method ‘rel col > printer profile > abs col > color space’ and compared with the SpectrumViz Lab numbers and the results are all over the place” I’m not sure if you’re saying the method does not work or that the white points in the profile are not correct.  Perhaps you’re saying both.

Mark

Mark,

Your method is sound enough for me for the conditions Adobe ACE + Photoshop. Actually a fast way to see what spectrometers + profile creators deliver in profile white point that I can compare to my SpectrumViz measurements.

Huge deviations I see in the use of Microsoft ICM + Photoshop with your method. Wonder what LCMS does in Qmage, have to find a correct route for that > tried with 2x Print to File and delivers the same result as ACE in Photoshop (edit).

Measured with your method + ACE I see huge deviations in profile white points of  b 0 for OBA papers with a b -7 to -10 in my measurements. It says something about the odd way profiles are made.  I have some difficulty with the fact that UV-cut spectrometers are considered to give better profiles in practice with whatever paper. And I know from my own experience that it is like that for most cases. It is like building a stone house on wooden pillars and it holds. Most of the time.


--
Met vriendelijke groet, Ernst

http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
July 2013, 500+ inkjet media white spectral plots.

[MarkH2]

…Huge deviations I see in the use of Microsoft ICM + Photoshop with your method…
--
Met vriendelijke groet, Ernst

I tried a few with Microsoft ICM and see what you're talking about, especially when converting back to sRGB.  Earlier this year Jeff did some digging into Windows color management and its (inscrutable) intricacies (http://www.luminous-landscape.com/forum/index.php?topic=76312.0 ).  I’ll just stick with Adobe ACE for now.

Mark

[TylerB]

I think it can be pretty easy.. if you just want Photoshop's LAB whole number rounding.. make a new white file any size, assign your output profile, then look at LAB in your info palette. Set Color Settings Conversion option to Absolute, you don't even have to save the setting, just look at  your info with the dialog box open.
Tyler

[MarkH2]

I think it can be pretty easy.. if you just want Photoshop's LAB whole number rounding.. make a new white file any size, assign your output profile, then look at LAB in your info palette. Set Color Settings Conversion option to Absolute, you don't even have to save the setting, just look at  your info with the dialog box open.
Tyler

Tyler,

It works to see the white point if you assign the profile, then convert to a color space using absolute rendering.

Note that this will only work for the white point.  If you want to see how a photo looks when printing (as in softproofing) then the realignment of the neutral axis during perceptual or relative rendering will affect other colors as well.  Just assigning a profile (which preserves the rgb values) then converting back with absolute rendering (preserving lab values) will give a different result.  The imperfections of a-to-b and b-to-a profile conversions affect all this as well.

Mark

[TylerB]

Mark, if you just LAB want the numbers, you don't have to convert, but you do need your intent set to abcol in your color settings, at least temporarily. You can simply open the settings dialog change to abcol, put your curser in that white and look at the info LAB numbers, no need to even save the color setting.. just back on out. I thought WP numbers were all the thread was about, no? What they are in RGB depends on what RGB space the OP is concerned with which hasn't been specified but gets more to the usefulness of the exercise anyway, then yes, conversion will be necessary to see RGB numbers.

[MarkH2]

Tyler,
Yes, I see what you mean.  Not the behavior I would expect.  Assign a profile to a file.  Then afterward change the color settings, which I would expect to apply only to future operations, but instead the assignment previously made changes.  Useful in this case, but not what I would have predicted.  When I initially tried your suggestion it did not work - must have been pilot error on my part.
Mark

[hugowolf]

Sorry, I actually forgot I had started this thread:

What I want to do is replicate the color of one paper, on another brighter while paper

Jeff: PC so no ColorSync utility. Would love ColorThink, but can’t really justify the cost; although I have heard many say it has proved more useful than they though it would.

Andrew: I didn’t understand

Mark: I had thought of this, but when I tried it, the values seemed too neutral, the paper didn’t seem that gray. In 8-bit I got Lab(98,0,0), in 16 bit Lab(98,0,1), in 32 bit Covert to Profile… is grayed out. (Interestingly if I change the mode to 32 bit from either 8- or 16-bit, the image turns very, very pink.) I will try printing it on whiter paper and see what happens.

Ernst: No OBAs in the paper, but interesting points none-the-less.

Tyler: I get the same results as doing this a la Mark, and interestingly the same pink when doing it with a 32 bit file.

Brian A