Color space questions

In PS CS, is there a way to *see* where a color locates in d

wrote:

In PS CS, is there a way to *see* where a color locates in different color spaces? If the rgb value of a color is known, e.g. 50-20-20, I would like to see where it physically locates in the working space AdobeRGB1998 and also when Soft Proofing to a printer media profile space. Colors like these are way off in Soft Proof and also in the prints. Because they do not show up as gamut warnings, I’m curious where they are in the color spaces.

LabMeter is a free download designed to do this:
http://www.curvemeister.com/tutorials/LabMeter/index.htm

The LabMeter image, together with Photoshop’s gamut warning will show you where the color is in the gamut.

To locate a specific color, set up the preview as per the instructions, set an eyedropper to display in RGB, and drag it around on the soft preview until it matches your numbers. You’ll also need to change with the adjustment layer to change the overall brightness to home in on the exact RGB value.
—

Mike Russell
www.curvemeister.com
www.geigy.2y.net

Bill Hilton wrote:

In PS CS, is there a way to *see* where a color locates in d­ifferent color spaces? If the rgb value of a color is known, e.g. 50-­20-20, I

would like to see where it physically locates in the working­ space AdobeRGB1998

Convert to profile and read it again?

OK, my question is not exactly very clear. When Soft Proofing in a printer media profile, colors like 50-20-20 look very different from the working space colors. Yet there is no Gamut Warning. I’m wondering if these colors are close to the edge of a gamut map. Hence the question.

also when Soft Proofing to a printer media ­profile
space.

In the Info palette click on the right-most eyedropper icon in the top row and change it to ‘proof color’ … if you have a soft proof active you can hold the eyedropper over a color and the first box shows the actual values, the 2nd one (the one you just changed) shows what the rgb values of the proof profile (ie, what the values will be when sent to the printer, for example).

This is very interesting indeed, but also brings up other questions on how to make use of the Proof Color values.

– One suggested workflow is to convert the file to the printer profile, send to the printer, and ask the printer driver not to color manage. In this case, how does the printer apply the Proof Colors? If the proof color value is x-y-z, will the printer print the color as x-y-z? If not, why would the printer print to a different value, and by what method?

– For the same profile, how should Proof Colors differ from Actual Colors if ONLY the rendering intent is changed from Perceptual to Relative Colorimetric for the colors that are within gamuts? And for the colors that are not?

I wish Adobe could provide more information on this.

Mike Russell wrote:

wrote:

In PS CS, is there a way to *see* where a color locates in different color spaces? If the rgb value of a color is known, e.g. 50-20-20, I would like to see where it physically locates in the working space AdobeRGB1998 and also when Soft Proofing to a printer media profile space. Colors like these are way off in Soft Proof and also in the prints. Because they do not show up as gamut warnings, I’m curious where they are in the color spaces.

LabMeter is a free download designed to do this:
http://www.curvemeister.com/tutorials/LabMeter/index.htm
The LabMeter image, together with Photoshop’s gamut warning will show you where the color is in the gamut.

To locate a specific color, set up the preview as per the instructions, set an eyedropper to display in RGB, and drag it around on the soft preview until it matches your numbers. You’ll also need to change with the adjustment layer to change the overall brightness to home in on the exact RGB value.

VERY interesting tool. I have two windows open – one for the working space and one for the profile. But 50-20-20 is at the same location on both maps. This does not tell me what values the profile is converting 50-20-20 to. Am I using this tool correctly?

wrote:

Bill Hilton wrote:

In PS CS, is there a way to *see* where a color locates in d–ifferent color spaces? If the rgb value of a color is known, e.g. 50-–20-20, I

would like to see where it physically locates in the working– space AdobeRGB1998

Convert to profile and read it again?

OK, my question is not exactly very clear. When Soft Proofing in a printer media profile, colors like 50-20-20 look very different from the working space colors. Yet there is no Gamut Warning. I’m wondering if these colors are close to the edge of a gamut map. Hence the question.

That depends on what rendering intent you use. If you use ‘Perceptual’, _all_ colors get shifted in order to preserve the overall look of the image, not only those that are (close to) out of gamut. Colors ‘on the edge’ get shifted more than colors somewhere in the middle, of course.

– One suggested workflow is to convert the file to the printer profile, send to the printer, and ask the printer driver not to color manage. In this case, how does the printer apply the Proof Colors? If the proof color value is x-y-z, will the printer print the color as x-y-z? If not, why would the printer print to a different value, and by what method?

If you convert the file to the printer profile, your original value is already changed to what it should be for printing. Consequently, the ‘proof colors’ are already applied. So yes, the printer will print the x-y-z value of the proof.

—
Johan W. Elzenga johan<<at>>johanfoto.nl Editor / Photographer http://www.johanfoto.nl/

When Soft Proofin

for example a pale yellow (in gamut) o

wrote:

[re use of LabMeter]

http://www.curvemeister.com/tutorials/LabMeter/index.htm

VERY interesting tool. I have two windows open – one for the working space and one for the profile. But 50-20-20 is at the same location on both maps. This does not tell me what values the profile is converting 50-20-20 to. Am I using this tool correctly?

Hi Jean,

The location and Lab values will be the same, but you can set the eyedropper to show the color values for your RGB and CMYK working spaces.

Note that the eyedropper values are for your working spaces, and do not reflect the Custom settings of your soft proof.

Thanks for a good question – I’ll add it to the web page. Keep em coming 🙂
—

Mike Russell
www.curvemeister.com
www.geigy.2y.net

Bill Hilton wrote:

When Soft Proofin­g in a
printer media profile, colors like 50-20-20 look very differ­ent from

the

working space colors. Yet there is no Gamut Warning. I’m won­dering if

these colors are close to the edge of a gamut map. Hence the­ question

You can use a tool like Color Think by Chromix to plot the gamut of your image and then plot how it maps to a printer profile. This should answer this type of question.

One suggested workflow is to convert the file to the print­er profile,

send to the printer, and ask the printer driver not to color­ manage.

In

this case, how does the printer apply the Proof Colors?

You don’t want the printer to apply the shift in this case since it was already done once when you converted.

Understood and agreed.

For the same profile, how should Proof Colors differ from ­Actual Colors if ONLY the rendering intent is changed from Perceptu­al to Relative Colorimetric for the colors that are within gamuts?

Open up a test pattern and watch the numbers change as you change rendering intents … for example a pale yellow (in gamut) on a pattern I have is 255/148/176 and proofed to the Epson 4000 Enhanced Matte profile it becomes 255/244/200 with perceptual rendering, 254/248/198 with rel col (not too different, in-gamut colors look similar in both intents).

And for the colors that are not (in gamut)?

Saturated yellow on the same pattern is 255/242/46, mapping to 255/241/122 (perceptual) or 254/252/90 relcol … note that some profiles don’t have separate values for perceptual and rel col, the spec only requires the profile to have one or the other, not both.

By definition, colors proofed with perceptual intent or colors that are oog would cause the values to shift. But why would colors that are in gamut shift when proofed with rel col intent?

View | Gamut Warning is also confusing. I’ll start another thread about that.

I wish Adobe could provide more information on this.

"Real World Color Management" by Fraser, Murphy, Bunting is good. Actually the Adobe Help files are good for soft proofing and rendering intents too.

Thanks.

Mike Russell wrote:

wrote:

[re use of LabMeter]

http://www.curvemeister.com/tutorials/LabMeter/index.htm

VERY interesting tool. I have two windows open – one for the working space and one for the profile. But 50-20-20 is at the same location on both maps. This does not tell me what values the profile is converting 50-20-20 to. Am I using this tool correctly?

Hi Jean,

The location and Lab values will be the same, but you can set the eyedropper to show the color values for your RGB and CMYK working spaces.
Note that the eyedropper values are for your working spaces, and do not reflect the Custom settings of your soft proof.

Bill’s post on using the Info Palette Proof Color is the solution to comparing real and proof colors.

Thanks for a good question – I’ll add it to the web page. Keep em coming 🙂

You are welcome, and thanks for the tool.

By definition, colors proofed with perceptual intent or colors that

are

oog would cause the values to shift. But why would colors that are in gamut shift when proofed with rel col intent?

Hi Jean, I’m not an expert on this but I’ve studied it quite a bit … I can think of at least two reasons for a shift even with in-gamut colors. One is caused by ‘black point compensation’, which (to quote from "Real World Color Mangement") "ensures that black in the source is always mapped to black in the destination, so that the entire dynamic range of the input is mapped to the entire dynamic range of the output".

A bigger cause is that most ink sets aren’t perfectly gray-balanced, meaning if you have equal RGB values you don’t necessarily get a neutral tone, so there’s a shift to compensate for this. Here’s a plot I made of three ink/paper combos, check the last two to see what I mean …. the graph has black on the left, white on the right and the gray diagonal line is the corresponding RGB gray-scale values … the red, green and blue lines show the shift needed (according to the ICC profile) to actually get gray at any given point. Most of this is in-gamut so even here you see a fairly big shift (the top plot is from a $130,000 Lightjet laser printer profiled by one of the sharpest minds in the digital game and it’s much better gray-balanced). http://members.aol.com/bhilton665/grayscale.jpg
Bill

Bill Hilton wrote:

By definition, colors proofed with perceptual intent or colors that

are

oog would cause the values to shift. But why would colors that are in gamut shift when proofed with rel col intent?

Hi Jean, I’m not an expert on this but I’ve studied it quite a bit … I can think of at least two reasons for a shift even with in-gamut colors. One is caused by ‘black point compensation’, which (to quote from "Real World Color Mangement") "ensures that black in the source is always mapped to black in the destination, so that the entire dynamic range of the input is mapped to the entire dynamic range of the output".

A bigger cause is that most ink sets aren’t perfectly gray-balanced, meaning if you have equal RGB values you don’t necessarily get a neutral tone, so there’s a shift to compensate for this. Here’s a plot I made of three ink/paper combos, check the last two to see what I mean … the graph has black on the left, white on the right and the gray diagonal line is the corresponding RGB gray-scale values … the red, green and blue lines show the shift needed (according to the ICC profile) to actually get gray at any given point. Most of this is in-gamut so even here you see a fairly big shift (the top plot is from a $130,000 Lightjet laser printer profiled by one of the sharpest minds in the digital game and it’s much better gray-balanced). http://members.aol.com/bhilton665/grayscale.jpg
Bill

Thanks for the great explanation. Looking at your Profile Inspector plots brings up more questions. The plots can tell you how much shifting a profile is causing, but how can you tell whether it is good or bad? Ultimately, shouldn’t the goodness of a printer profile be judged by looking at the final print? And how can the plots lead you there?

Norman Koren evaluates profiles by using a test chart.
http://www.normankoren.com/color_management_4.html

Here, you judge a profile by looking at the converted chart or its print. Although the evaluation is subjective, I find it more understandable and hits closer to home.

One profile vendor uses Koren’s method to demonstrate how he creates profiles.
http://www.colourprofiles.com/sitemap.htm
Click on the "Visual demonstration…" link on this page.

There are many profile vendors out there. I wonder why the "good" ones don’t do something similar to promote and distinguish their products.

Mike,

IMO the gamut warning by PhS7 is wrong for profiles
which were made by software which uses internally
the Logo Steinfurt color system.
The profiles themselves are OK.

An example is here (updated doc) :
http://www.fho-emden.de/~hoffmann/hungams17042004.pdf

Would you please check this issue by your tools ?

Best regards –Gernot Hoffmann

The plots can tell you how much shifting
a profile is causing, but how can you tell whether it is good or bad?

Best way to evaluate an output (usually printer) profile is to print a pattern with known values and read them with a spectrophotometer and compare what you measured with what was in the file. Most top-end profiler packages provide the software to do this … read ch 9 in "Real World Color Management" for details.

Ultimately, shouldn’t the goodness of a printer profile be judged by looking at the final print?

A couple of problems with relying solely on prints … 1) typically you are comparing the print to what you see on the monitor so there are two profiles involved. If there are problems it can be hard to tell which profile is off. 2) typically the human eye adjusts for color shifts and white balance differences, making it difficult to accurately judge. Using a measuring tool like the spectrophotometer gives you a more objective evaluation down to the number of delta-E’s your profile is off for any given color patch, though of course at the end of the day what really matters is how well the viewer likes the prints.

Bill

wrote:

Mike,

IMO the gamut warning by PhS7 is wrong for profiles
which were made by software which uses internally
the Logo Steinfurt color system.
The profiles themselves are OK.

An example is here (updated doc) :
http://www.fho-emden.de/~hoffmann/hungams17042004.pdf

Would you please check this issue by your tools ?

Best regards –Gernot Hoffmann

Gernot,

Very interesting. Have you tried switching engines to see if the problem is with the engine? If you can point to a copy of the lexmark profile I’ll be happy to take a look.

Mike

Mike,

the two profiles for the Lexmark 4079plus are here:

http://www.fho-emden.de/~hoffmann/PS4079.ICM
http://www.fho-emden.de/~hoffmann/Lex29122004.icc

The same bug (wrong gamut warning) happened for
high end inkjet profiles by PosterShop Pro/Profiler,
which uses internally the Logo Steinfurt system as well.

Color engine was always ACE.

Best regards –Gernot Hoffmann

wrote:

Mike,

the two profiles for the Lexmark 4079plus are here:

http://www.fho-emden.de/~hoffmann/PS4079.ICM
http://www.fho-emden.de/~hoffmann/Lex29122004.icc

The same bug (wrong gamut warning) happened for
high end inkjet profiles by PosterShop Pro/Profiler,
which uses internally the Logo Steinfurt system as well.

I don’t think Lex29122004 is a valid profile, or at least it does not function well on my system.

I’m getting large areas of black when I select that profile in Photoshop’s Assign Profile command. That’s with the Adobe engine. when I switch to Microsoft’s CMM I get pure white for all rendering intents.

The first profile, PS4079, works as expected, and the gamut looks reasonable in LabMeter.
—
Mike Russell
www.curvemeister.com
www.geigy.2y.net

Mike,

thanks for the investigation.

The Lex.. profile was used for printing by PhS
last December.

An actual test shows:
PhS’s softproof interprets the profile reasonably .

GMB ProfileEditor/GamutView shows the Lab volume
alike the manufacturer’s profile PS4079.

Of course you are right – something is wrong.
But, as already explained, similar problems with
at least one other Logo Steinfurt profile
(Lex… by GMB ProfileMaker, other inkjet profile
by PosterShop/Profiler).

Best regards –Gernot Hoffmann
(Sorry for posting in wrong order – Google-Beta
doesn´t work in the moment)

Gernot Hoffmann wrote:

Mike,

thanks for the investigation.

The Lex.. profile was used for printing by PhS
last December.

An actual test shows:
PhS’s softproof interprets the profile reasonably .

GMB ProfileEditor/GamutView shows the Lab volume
alike the manufacturer’s profile PS4079.

Of course you are right – something is wrong.
But, as already explained, similar problems with
at least one other Logo Steinfurt profile
(Lex… by GMB ProfileMaker, other inkjet profile
by PosterShop/Profiler).

I have to think it’s the profile.

On my system the soft proofs were not even close to reality, and the results are dependent on the engine. I got a pure white soft proof using the Microsoft CMM, and large areas of black covering most of the image with the Adobe engine.
—
Mike Russell
www.curvemeister.com
www.geigy.2y.net

Mike,

that´s strange. My tests for Lex29122004.icc:

1. Print by PhS 7
2. Soft proof by PhS 7, ACE and MCE, four rendering intents, with and without BPC.
Everything as expected (of course a small gamut, but nothing garbled)

3. Gamut Warning by PhS 7
Very wrong

Another profile with the same behaviour (3MB):
http://www.fho-emden.de/~hoffmann/mutoh950S.icc

Made by PosterShop/Profiler (Logo Steinfurt)
Very good print results on Mutoh 6100

Best regards –Gernot Hoffmann

wrote:

Mike,

that

Mike,

thanks for the tests.
Again: the odd gamut warnings appear for all rendering intents. But for the moment we can leave it at that.

My opinion about RelCol and AbsCol:

The viewing light for print products is D50. Therefore the reference white for the PCS CIELab is D50 as well.

If the RGB space is defined for D50, like ProPhoto, then AbsCol and Relcol is the same.

If the RGB space is defined for D65, like sRGB, then
AbsCol and RelCol is different. It’s just the coordinate transform of the D65 gray axis to the D50 gray axis.

Essentially, AbsCol isn’t intended as a printing space.
For me it´s the rendering intent for CMYK–CMYK conversions for proof printing.
That means: the final gamut RelCol is valid. Source CMYK is converted to Lab and Lab is converted to destination CMYK. The output profile (inkjet) was made for RelCol. The conver- sions are done without white point shift (AbsCol)

This is so far valid without taking paper white into consi- deration.
RelCol can apply corrections for paper white, but this is IMO (and in my practical experience) rather doubtful.
Much more important than paper white is IMO the black point compensation, which is often applied in Perceptual, but in a somewhat unspecified manner. RelCol doesn’t have a BPC
by ICC standards.
PhS RelCol with BPC is often similar to Perceptual.

Here is my attempt how to decipher the ICC specs:
http://www.fho-emden.de/~hoffmann/cmsicc08102003.pdf

Comments and corrections are welcome, maybe by private e-mail.

Best regards –Gernot Hoffmann

Bill Hilton wrote:

The plots can tell you how much shifting
a profile is causing, but how can you tell whether it is good or bad?

Best way to evaluate an output (usually printer) profile is to print a pattern with known values and read them with a spectrophotometer and compare what you measured with what was in the file. Most top-end profiler packages provide the software to do this … read ch 9 in "Real World Color Management" for details.

Ultimately, shouldn’t the goodness of a printer profile be judged by looking at the final print?

A couple of problems with relying solely on prints … 1) typically you are comparing the print to what you see on the monitor so there are two profiles involved. If there are problems it can be hard to tell which profile is off. 2) typically the human eye adjusts for color shifts and white balance differences, making it difficult to accurately judge. Using a measuring tool like the spectrophotometer gives you a more objective evaluation down to the number of delta-E’s your profile is off for any given color patch, though of course at the end of the day what really matters is how well the viewer likes the prints.

Agreed with your comments, as usual. My statement that "the goodness of a printer profile [should] be judged by looking at the final print" is from the perspective of hobbyists with reasonably well profiled monitors, and not from the professionals with spectrophotometers. Those professionals probably don’t come here for advice.

Getting back to the real question: what can one learn from Profile Inspector plots?

Mike,

the profile Lex29122004.icc was checked again:

PhS7: Print OK (using ACE)
PageMaker: Print OK (using Kodak Color Engine)
GMB ProfileEditor/GamutView: OK
IccInspect (little.cms): OK
X-Rite Colorshop: OK

No idea why this profile should be wrong.

This profile and the large format inkjet profile mutoh950S.icc are both OK.
But both show wrong gamut warnings in PhS 7.0.
Both use Logo Steinfurt software (embedded in GMB or Onyx programs).

Best regards –Gernot Hoffmann

wrote:

Mike,

the profile Lex29122004.icc was checked again:

PhS7: Print OK (using ACE)
PageMaker: Print OK (using Kodak Color Engine)
GMB ProfileEditor/GamutView: OK
IccInspect (little.cms): OK
X-Rite Colorshop: OK

No idea why this profile should be wrong.

This profile and the large format inkjet profile mutoh950S.icc are both OK.
But both show wrong gamut warnings in PhS 7.0.
Both use Logo Steinfurt software (embedded in GMB or Onyx programs).

Best regards –Gernot Hoffmann

Gernot ,

In PS 6’s soft proof, I got large areas of black with Lex29122004.icc, using both the MS and Adobe CMM engines. I’ll set aside some time later today to try it in PS CS on a different system.

BTW – I’ve found your online articles very valuable – thanks for providing them. I’m ramping up on ICC internals, and find that the actual ICC spec reminds me of the Byzantine DICOM specification. Your articles are serving as a very valuable Rosetta stone.
—
Mike Russell
www.curvemeister.com
www.geigy.2y.net

Mike,

thanks for your friendly comment.
At present my work is probably more helpful for explaining Byzantine architecture than for deciphering the ICC specs:
I’m printing 42 posters for an exhibition about the church/mosque/ museum Hagia Sophia in Istanbul (research by one of my brothers) which will be shown there in April and in Berlin in May.

Curvemeister is interesting. I see, the histogram is embedded in the Curve diagram. Shouldn´t it be drawn in horizontal direction ? The original histogram could be drawn by vertical bars, but each modi- fication means new values – horizontal bars then.

I’m wondering why Photoshop doesn´t show the histogram in Curves.

Best regards –Gernot Hoffmann

wrote:

Mike,

thanks for your friendly comment.
At present my work is probably more helpful for explaining Byzantine architecture than for deciphering the ICC specs:
I’m printing 42 posters for an exhibition about the church/mosque/ museum Hagia Sophia in Istanbul (research by one of my brothers) which will be shown there in April and in Berlin in May.

What a wonderful subject for a photographic series. This is of the greatest buildings in the world – no pillars in that great space- even without taking into account it’s great age, and the fact that there are earthquakes in the area. I hope your brothers images are on line as I would enjoy seeing them.

Curvemeister is interesting. I see, the histogram is embedded in the Curve diagram. Shouldn