ELI5: Why are CMYK conversions different using Pantones new Lab colors

[gig0]

Color management is my weakest link but Pantone's newly referenced Lab specs for spot colors has me confused. I understand Pantone needs to convey their colors on new media, e.g., tablets, digital billboards, etc., but from everything I've been taught, the CMYK gamut is a smaller color space than RGB and Lab. Why would Pantone change the CMYK recipe values in a smaller color space even though the objective to to emulate more accurate color on a wider gamut?

 

[Stephen Marsh]

CMYK is a device dependent space, the old “hard wired” CMYK values that Pantone used did not really match standard CMYK conditions (and the new Plus Bridge CMYK is the same, it does not match standard conditions). Basing the source values on Lab in theory results in more accurate conversions to the destination, as long as the destination profile reflects the final print conditions. Aside from theory, there will always be issues with both legacy work and with colour conversions where the source colour is “out of gamut” for the destination profile.


Stephen Marsh

 

[gig0]

Thanks Steven, but it's still out of my realm of understanding. As an Example, PMS 300 was broken down to (around) 100c 44m 0y, 0k. Now all of a sudden the magenta values jump up to 66%. I just don't understand how the outside Lab values would affect a stagnant smaller color space where everyone was happy. I simply don't get how they all of a sudden Pantone decided the CMYK recipe, which is a smaller gamut, just wasn't a good enough match. Just is a significant change for a smaller gamut that, in theory, should remain constant because it's inside that color space.

Perhaps my understanding is just over complicating this?

 

[gig0]

Thanks Steven, but it's still out of my realm of understanding. As an Example, PMS 300 was broken down to (around) 100c 44m 0y, 0k. Now all of a sudden the magenta values jump up to 66%. I just don't understand how the outside Lab values would affect a stagnant smaller color space where everyone was happy. I simply don't get how they all of a sudden Pantone decided the CMYK recipe, which is a smaller gamut, just wasn't a good enough match. Just is a significant change for a smaller gamut that, in theory, should remain constant because it's inside that color space.

Perhaps my understanding is just over complicating this?

 

[Stephen Marsh]

If you really wish to use hard wired CMYK values that may or may not match your actual final printing conditions, then use the Pantone Plus Colour Bridge values which are a CMYK source, not Lab.

The (rounded integer) Lab values of Pantone + 300 C =
L*: 36
a*: -10
b*: -62

Converting these Lab values to the Adobe Fogra 39 profile results in Lab values of:

L*: 38
a*: -9
b*: -48

Using the Colour Bridge hard wired CMYK values and assuming Fogra39 results in these Lab values:

L*: 41
a*: -12
b*: -50

Clear as mud? Do you really wish to know the dE values as to which is a closer match?

What works well for one colour may be totally off with another colour. There are no easy answers. This is part of our “value add” – managing colour and client expectations of colour.


Stephen Marsh

 

[gig0]

website is working wonky on my end so disregard the double post, fyi.

 

[gig0]

This is exactly where I don't understand. How has the outside Lab value suddenly change a "static" CMYK conversion value, which is a smaller color gamut? These were CYMK formulations thats been relied on upon years. How does one decide to change the recipe of color within a smaller color gamut? Thats my question.

 

[gig0]

Thank you Stephen but can you break it down into more simple terms because many of us don't even have the spectros you scientists have. Like I said, explain like I am five. Perhaps that will help others understand as well. I think it will convey things much more easily. Call me an idiot. Shut up Gordo.

 

[gordo]

Shut up Gordo.

Sadness :-(

 

[gig0]

Just Joking around with you, dude. Color just seems to be a very frustrating topic in this biz

 

[Stephen Marsh]

Thank you Stephen but can you break it down into more simple terms because many of us don't even have the spectros you scientists have. Like I said, explain like I am five. Perhaps that will help others understand as well. I think it will convey things much more easily. Call me an idiot. Shut up Gordo.

All of the previous info and Lab values were from Photoshop, no photons were harmed and I am definitely not a colour scientist :]

Explain in simple terms, hmmm…easier said than done! I would not have this conversation with a 5 year old. However, it is not really about age, it is more about one’s exposure to and understanding of the concepts of colour and colour management.

There is no one “single magic CMYK value", every CMYK device is different (device dependence).

There is one “single magic Lab value" (device independence).

The same CMYK values in SWOP create a different colour appearance than in say GRACoL or SNAP.

it is “simple” colour management to convert from a known Lab value to a known CMYK value.

With GCR, there are many combinations of CMYK values that can all create the same Lab value.

Of course, with legacy process simulations of spot colours, the goal is not really to match the original Pantone (which could be out of gamut anyway and not matchable), the goal is to match the previous CMYK print (which may be hard to do even if using the same CMYK values as the previous print).

When it comes to out of gamut colours such as Orange 021 being converted to CMYK, it comes down to artistic intent over science.


Stephen Marsh

 

[gig0]

Sorry. I'm outta time here. The lady is getting pissed and I'm on a 4 day weekend and I'm still talking about work.

Thanks, Stephen for the breakdown but I still can't seem to understand why a smaller color gamut (CMYK) would change value when the outside gamut (Lab) was introduced to represent a spot color. Lab encompasses CMYK, right?

All things being equal with the same output intent, the values should be the same? Or No?


Here are two files, one generated from IDCS3 and one from IDCS2014. Both using 021 as the spot color and both files exported using the same intent, however, when rasterized in Photoshop (CS3) via CMYK conversion, there are two different recipes. Color management was set to CMS OFF in both versions of InDesign.

Why would the CMYK recipe change? This is what I don't get.

 

[fletcher0854]

...it comes down to artistic intent over science.


Stephen Marsh

Hi gig0,

I think this line from Stephen might be where you are getting caught up on. My explanation to a 5 year old; Because they want to. I understand you are looking for more of a scientific explanation than that, but I'm not sure there is one. Think of it this way:

CMYK: Process
Lab: Measurement
Pantone: Company

While CMYK and Lab values ultimately won't change, Pantone can say their colors are whatever they want.

I hope that helps, or is remotely an answer you are looking for.

 

[gordo]

First off, let's get the terms and concepts right...

CMYK, just like RGB, does not describe color.

CMYK is a composite of 4 greyscale (achromatic) images. Each greyscale image can represent 100 tones/levels of grey. The color that you see represented by a composite CMYK image is dependent on the hues used to print each component greyscale image. Change the hue of each component and the final color changes. The range of colors that can be represented using a CMYK model is dependent on the hues of the CMYK components, the observer, the substrate, and the illumination.

RGB is a composite of 3 greyscale (achromatic) images. Each greyscale image can represent 256 tones/levels of grey. The color that you see represented by a composite RGB image is dependent on the hues used to print each component greyscale image to a screen display. Change the hue of each component and the final color changes. The range of colors that can be represented using an RGB model is dependent on the hues of the RGB components, the observer, the display, and the ambient illumination.

CIE L*a*b*, on the other hand, is a 3 coordinate locational system for defining a color. If two measured samples have the same coordinate values then they are (theoretically) identical. The color identified using a CIE L*a*b* model is dependent on the instrument used to determine the coordinate values. I.e. Two different instruments can give two different coordinate values for the same sample.

Pantone is a proprietary custom color library. Each color in the library is defined by a recipe of specified ink/pigment components. I.e. to get a specific hue, you mix X parts of one ink with parts of other inks. AFAIK, Pantone includes CIE L*a*b* coordinates only as a convenience for color systems that use ICC profiles or LUTs to convert Pantone's library into various color models (RGB, CMYK, etc).

Below is part of the Pantone library that is licensed to RIP vendors:

In this image 021 Orange is highlighted and you can see the CIE L*a*b* values that Pantone has provided to a RIP vendor.
In the box is the same 021 Orange as represented in PhotoShop's Pantone color library that Pantone has provided to Adobe.
Notice that the CIE L*a*b* values are different.
Since the CIE L*a*b* values are different, the resulting RGB/CMYK values will also be different, and hence the resulting colors simulating the target 021 Orange will be different.
Pantone does alter the CIE L*a*b* values it licenses to third parties as time goes by. The third parties may or may not be up to date with implementing the different versions of Pantone's libraries, nor do they necessarily ensure that the libraries used in individual vendor's applications are the same.
Which may be the problem in the OP's case.

 

[Stephen Marsh]

Using Lab as a spot colour alternate colour build is about accurately describing the source colour. Even using Lab is limited in it’s ability to describe spot colour behaviour (tints, overprints and substrate), which is why spectral reflectance value colour management is used by some proprietary systems and is the basis of an upcoming open ISO standard to communicate spot colours.

Using CMYK as a description to build a spot colour in many cases will not create an accurate depiction of the spot colour, but it will make a CMYK simulation more convenient for some.

As to your uploaded files, one file uses Lab and the other file CMYK as the alternate colour build to describe the same spot (so they are potentially different colours based on their colour builds, even though they are the same named colour). The reason that the conversion to CMYK or RGB changes is that the source for both files differs, even if the destination does not. This is why most workflows and proofing RIPs offer named colour lookup tables and the ability to override or use the files source values to describe a named spot colour. Photoshop does not offer such a control when rasterising a PDF file.

As Orange 021 is “out of gamut” for common press CMYK, artistic intent enters the picture, as using a Lab source will likely result in an unacceptable conversion that has too much magenta and is basically a red. The CMYK based alternate colour values would likely represent a better CMYK orange, as the colour values have been built with common sense, which current colour conversions using simple colorimetric intents lack. Perhaps when colour appearance model based conversions are commonplace, such conversions may possibly result in better results. The flip-side is of course that for “in-gamut” colours, using a Lab based source usually creates a better conversion (if the destination profile is correct).


Stephen Marsh

Sorry. I'm outta time here. The lady is getting pissed and I'm on a 4 day weekend and I'm still talking about work.

Thanks, Stephen for the breakdown but I still can't seem to understand why a smaller color gamut (CMYK) would change value when the outside gamut (Lab) was introduced to represent a spot color. Lab encompasses CMYK, right?

All things being equal with the same output intent, the values should be the same? Or No?


Here are two files, one generated from IDCS3 and one from IDCS2014. Both using 021 as the spot color and both files exported using the same intent, however, when rasterized in Photoshop (CS3) via CMYK conversion, there are two different recipes. Color management was set to CMS OFF in both versions of InDesign.

Why would the CMYK recipe change? This is what I don't get.

 

[gig0]

Many thanks, fellas, for helping me out in these difficult times! *tips hat* I think my problem was an old way of thinking and the frustration of not having a good enough understanding of the impact and the solutions. Stephen pretty much nailed it right here for me:

"Of course, with legacy process simulations of spot colours, the goal is not really to match the original Pantone (which could be out of gamut anyway and not matchable), the goal is to match the previous CMYK print (which may be hard to do even if using the same CMYK values as the previous print)."

I had the attitude of, 'oh god, this is going to be a nightmare for legacy process work where branding is vital'. Our biggest client, uses PMS 300 in all their work and logos and I have no control as far as where these files originated from. We worked with them to supply the process work using specific CMYK values but some files still come using the spot color. Obviously I don't have time to call all their outsourced design studios around the world and tell them to stop using the Pantone+ library for process work but now understand how much more important color managed workflows have become, hence an earlier post of mine in regards to referencing a legacy Pantone library in our system for color conversions. This makes more sense but I have to wonder, how long can one rely on legacy color conversions in the world of printing?

 

[gig0]

Pantone does alter the CIE L*a*b* values it licenses to third parties as time goes by.

Gordo, just a follow up question; I've heard this from you guys many times before, but why would Pantone change the values over time? From what I understand, CIE Lab basically covers the entire color spectrum the human eye can see and is the most accurate perceptual representation of color. If theres a reference table for color, why would those Lab values need to change?

 

[gordo]

Gordo, just a follow up question; I've heard this from you guys many times before, but why would Pantone change the values over time? From what I understand, CIE Lab basically covers the entire color spectrum the human eye can see and is the most accurate perceptual representation of color. If theres a reference table for color, why would those Lab values need to change?

This is speculation....
Pantone first developed their custom spot color system when very few, if any, printers had spectrophotometers. Heck, many didn't even have a densitometer. If a client wanted a custom spot color they would bring a sample to the printer and the printer would match the sample the same way an artist would mix their paints.

In that environment Pantone created a color specifying system based on color samples (the swatchbook) and the ink mixing recipes needed for the printshop to replicate the spot color (using Pantone's inks). Designers used the Pantone swatchbooks for spot colors and CMYK color Atlases for screen tint builds.

It is the same basic method/system that's used by house paint vendors.

It's only recently that the industry has become more sophisticated in color (i.e. CIE Lab measurement) and Pantone has been trying to evolve while maintaining its core system methodology. Changing the Pantone system to one based on Lab values would run counter to Pantone's core system of recipes and would likely contribute to even more confusion when it comes to setting and meeting print buyer color expectations for spot/custom colors.

A CIE Lab color specification can be just as inaccurate a specification of color as a CMYK specification because it is instrument-based and instruments vary in what they report - even the same instrument from the same manufacturer may report different values for a specific measured target (The Print Guide: Top reasons why color instruments don't agree ). Also, if a Lab value is given then, because nothing is perfect, a tolerance needs to be given. How far away from PMS 123 can you be and the hue is still considered PMS 123? Is Pantone the one responsible for defining that tolerance? With a Lab value as the target, how would a printer compensate for color deviation from the target due to substrate, lighting, context, etc.?

When you say that "CIE Lab basically covers the entire color spectrum the human eye can see and is the most accurate perceptual representation of color" I don't agree. CIE Lab is a 3D coordinate mathematical model of human vision. It often does align with what we see - but not always, since our wet-ware varies between individuals and why so-called optical illusions like simultaneous contrast and Mach Bands are so fascinating.

 

[gig0]

and thus explains your signature

 

[Correct Color]

Just to add a little bit to some good info...

Gordo, just a follow up question; I've heard this from you guys many times before, but why would Pantone change the values over time?

And this isn't speculation...

The honest truth is that they haven't. Lots of people say they have, and lots more believe that they have, but the fact is that the Lab values that correspond to each original Pantone color have only changed once.

And the only reason they changed is that Pantone changed the paper on which they print the books, so as to be "greener." That paper change resulted in slightly different Lab values for many of the ink formulas when printed on the new media vs. the old -- and, as Gordo has pointed out -- the ink formulas are the true actual Pantone colors and have never changed since their inception.

That is the only reason for any differing Lab values.


Mike Adams
Correct Color