Dan Margulis Applied Color Theory

The Facts(?) About Color Management

Color management facts
Posted by: "Jeremy Schultz"
Fri Feb 20, 2009 1:32 pm (PST)

I attended an InDesign User Group meeting last night and the speaker was the director of color management for a local printing company. A lot of what he said flew in the face of what I have been taught by Dan and others about color management, so I wanted to put a list of some of his main points in front of this group and hear your reactions.

* You can set up a screen so a soft proof and a printed proof are identical.
* Designers should never convert images to CMYK‹printers should do it. * Printer?s profiles will give you more CMYK gamut.
* The GRACoL CMYK profile is best.
* Pantone colors are impossible to recreate with CMYK‹let printers do it.
* Let printers convert everything to CMYK because they know how to control GCR.
* On page layouts, leave spot colors as spot‹let printers convert to CMYK.
* A print job should look the same no matter where it?s printed.
* Pure CMY inks are possible, but they would be expensive to produce which is why the printer industry adds black for process color.
* Always give printers RGB files unless they absolutely demand CMYK.
* Use 16-bit images‹the speaker saw a totally underexposed 16-bit photo corrected into a perfect daytime shot.
* Printing on uncoated stock can look exactly like printing the same file on coated stock.
* All images need an embedded profile.

Jeremy Schultz
___________________________________________________________________________

Re: Color management facts
Posted by: "Peter Figen"
Fri Feb 20, 2009 3:52 pm (PST)

On Feb 20, 2009, at 12:26 PM, Jeremy Schultz wrote:

* You can set up a screen so a soft proof and a printed proof are
identical.

You can get them very close but never identical.

* Designers should never convert images to CMYK‹printers should do it.

In my experience the printer is usually the last person I would trust, but seeing as this is a printer making the recommendation...

* Printer's profiles will give you more CMYK gamut.

More gamut than what. The gamut of the press is determined by the inks and paper. A good profile will help you get the right color and density faster, but gamut?, not necessarily.

* The GRACoL CMYK profile is best.

There are more than one. If your press is running to the right densities, that may be true, but certainly not categorically.

* Pantone colors are impossible to recreate with CMYK‹let printers
do it.

Some are, but for the ones that fall within the capabilities of the press, you should be just as able as the printer. A good profile for that process will make the conversion numbers easier to ascertain.

* Let printers convert everything to CMYK because they know how to
control GCR.

Okay...if they insist. (just kidding)

* On page layouts, leave spot colors as spot‹let printers convert
to CMYK.
* A print job should look the same no matter where it?s printed.

Maybe it *should* but there's no way that it will.

* Pure CMY inks are possible, but they would be expensive to produce which
is why the printer industry adds black for process color.
* Always give printers RGB files unless they absolutely demand CMYK.

I would say ONLY give your printer RGB if you've had a specific discussion with their prepress department and they had a VERY good reason to have RGB. Too often, printers don't have the foggiest idea of how to best deal with RGB. This suggestion is asking for trouble. Again communication is all important.

* Use 16-bit images‹the speaker saw a totally underexposed 16-bit
photo corrected into a perfect daytime shot.
* Printing on uncoated stock can look exactly like printing the
same file on coated stock.

Did they really say this???

* All images need an embedded profile.

No comment needed.

Peter Figen
___________________________________________________________________________

Re: Color management facts
Posted by: Michael Jahn
Fri Feb 20, 2009 3:52 pm (PST)

On Fri, Feb 20, 2009 at 12:26 PM, Jeremy Schultz wrote:

* You can set up a screen so a soft proof and a printed proof are identical.

jahn comment - yes, this is exactly what they do at JCPenneys at the Plano Texas plant every day using EIZO monitors and the ORIS Soft Proof system - so, yes, it can be done and is being done, but it is not a simple or as inexpensive as most people think (or wish)

http://www.cgs.de/news/press/a081017_GoldenCylinder.php

* Designers should never convert images to CMYK‹printers should do it.

jahn comment - depends on the designers skill level. JCPenneys has people as skilled as the prepress people at the printers they use.

* Printer?s profiles will give you more CMYK gamut.

jahn comment - perhaps - and perhaps this is a good or not so good thing. if this is the only printer you use, and he has one single press condition, i am certain one can find another printer who prints with more gamut than this one printer - ask anyone printing Mattels boxes running Barbie Pink instead of Magenta)

* The GRACoL CMYK profile is best.

jahn comment - this would assume that this press condition is actually able to print to the GRACoL specifcation, and since there are many (and since they do not seem to mention which one) who knows what they actually meant to say for you to use...for example, there are three profiles here;

http://www.gracol.com/resources/iccaccept.asp

- did they mean GRACoL 2007 (GRACoL 7) - and did they mention if they were using the 'practice' many refer to as G7 to print ?

http://www.gracol.org/specification/

* Pantone colors are impossible to recreate with CMYK‹let printers do it.

jahn comment - if this were the case, how do they print the print the pantone book, using magic fairy dust ? (okay, so we will agree that "48 percent of the Pantone colors cannot be spectrally simulated using a standard CMYK inks, whatever THAT means (call your local ink supplier and ask for "SWOP Inks" or "GRACoL Inks" and they will start asking questions about your press and what paper you are using...

* Let printers convert everything to CMYK because they know how to control
GCR.

jahn comment - I am glad they said this. If they do the conversions and you are unhappy with their proofs, you can ask them to try again. And then, when the press sheets do not match the proofs, you can ask them to try that agin to.

* On page layouts, leave spot colors as spot‹let printers convert to CMYK.

jahn comment - same answer as above.

* A print job should look the same no matter where it?s printed.

jahn comment - unless it was printed at Anderson Litho, where they will make it look much better (that is their sales pitch, I am just parroting)

* Pure CMY inks are possible, but they would be expensive to produce which
is why the printer industry adds black for process color.

jahn comment - kinda sorta true in most heatset offset printing, but not so true for other printing processes.

* Always give printers RGB files unless they absolutely demand CMYK.
* Use 16-bit images‹the speaker saw a totally underexposed 16-bit photo
corrected into a perfect daytime shot.

jahn comment - that would assume that the image had image data in that underexposed area. This is a very old trick. Anyone can do the same with a 16 bit overexposed image. This has little to do with the bit depth.

* Printing on uncoated stock can look exactly like printing the same file on
coated stock.

jahn comment - sure, if you print the coated stock with less ink and with some flattening curves, you can even make coated stock look exactly like newspaper, right down to printing grey over that nice white paper !

* All images need an embedded profile.

jahn comment - except for 1 bit TIFF files.

--
Michael Jahn
___________________________________________________________________________

Re: Color management facts
Posted by: "Todd Shirley"
Fri Feb 20, 2009 3:52 pm (PST)

On Feb 20, 2009, at 3:26 PM, Jeremy Schultz wrote:

* You can set up a screen so a soft proof and a printed proof are
identical.

True, but only under extremely controlled conditions. You have to tweak the white point of the monitor and dim the booth so they match perfectly. JUST normalicht has an automatic system for doing this where the booth is connected via USB, otherwise it can take a while. The JUST system works well and I can say that I've actually seen a hard proof exactly match a screen using their booth and software, though I've never set set up such a system myself.

* Designers should never convert images to CMYK—printers should do it.

In an ideal color managed world sure. In reality virtually all printers want CMYK, which essentially breaks the color management chain right there. We supply RGB files to one organization who has
invested a lot in soft and hard proofing color management and they re-separate for a variety of uses and it works great. But they are a rarity. I wish it worked this way for more printers.

* Printer’s profiles will give you more CMYK gamut.
* The GRACoL CMYK profile is best.

These statements seem contradictory. I guess a printer COULD make a custom profile of a press that provided more gamut, but if GRACoL is best, why would they do it? Of course "best" is pretty vague, as obviously its not going to be best for all possible printing conditions. It's certainly a reasonable visual target for sheetfed offset, but there are others (ECI Coated V2 comes to mine) that are just as valid.

* Pantone colors are impossible to recreate with CMYK—let printers
do it.

I think it something less than 40% that can be reasonably simulated by CMYK. I agree that a printer could probably do a better job of it, but then you are left not knowing how a particular spot simulation will come out till the printer gives you a proof. If you do it yourself, and you are able to make a GRACoL (or whatever printing condition) proof, you will be able to control exactly in what way it doesn't match. Of course if it is an ACTUAL spot ink, then of course the printer has to do it.

* Let printers convert everything to CMYK because they know how to
control GCR.

Same as your 2nd point. Sure, great, except most printers don't want RGB. Also, many have pretty good software to RE-separate CMYK images to get the desired GCR, so it doesn't necessarily matter.

* On page layouts, leave spot colors as spot—let printers convert to
CMYK.

Same as your other spot color point. If you don't have a good way to proof your CMYK simulation, and you really just want it to match the spot ink as close as possible and don't care that it may or may not look like your screen, then have the printer do it.

* A print job should look the same no matter where it’s printed.

Ha ha. In an ideal world. Now that more people in the US are making certified proofs and more printers are printing to visual specs like GRACoL, this is a real possibility, but far from a common thing. Lots of printers don't have the process control to even match themselves from run to run.

* Pure CMY inks are possible, but they would be expensive to produce
which is why the printer industry adds black for process color.

I've heard this too but don't know much about it. If they are possible, someone must have done it, but I've never heard of it actually happening.

* Always give printers RGB files unless they absolutely demand CMYK.

Most of them absolutely demand CMYK.

* Use 16-bit images—the speaker saw a totally underexposed 16-bit
photo corrected into a perfect daytime shot.

No comment.

* Printing on uncoated stock can look exactly like printing the same
file on coated stock.

I've never seen it happen. Exactly is a pretty strong word. I think part of the point of using different stocks is that they print a little different.

* All images need an embedded profile.

This I totally agree with. You are implicitly using a profile all the time in photoshop. You should just attach it. Why throw it out? If you really think that the printer is going to mess it up by converting incorrectly, then you are dealing with a pretty low-end printer (there are plenty) and they will probably mess it up anyway. Any printer of reasonable quality would want the profile embedded.

My 2¢

-Todd Shirley
___________________________________________________________________________

Re: Color management facts
Posted by: "Alex Kent"
Fri Feb 20, 2009 3:52 pm (PST)

i will only address two of these:

- a transmitted light image / screen cannot ever look the identical a reflected light image / print.

- afaik there isn't a single camera or digital back available which actually really truly captures 16bit images. also, if you're getting as far as sending files to a printer and your images are still 'totally underexposed' then you have bigger problems in your workflow than 8bit vs. 16bit.

alex kent.
___________________________________________________________________________

Re: Color management facts
Posted by: Terence Wyse
Fri Feb 20, 2009 8:48 pm (PST)

Excuse me while I chase a few worms that just escaped from the can I just opened up.

:-)

Comments inserted...

On Feb 20, 2009, at 3:26 PM, Jeremy Schultz wrote:

* You can set up a screen so a soft proof and a printed proof are
identical.

It can be VERY VERY close but it usually takes more than just an off-the-shelf solution if you want to some sort of "certified" soft-proofing solution. Think Remote Director and a number of other solutions.

* Designers should never convert images to CMYK—printers should do it.

I guess it depends. A color-management savvy printer should know more about their pressroom than the designer so on the surface I would have to agree with this. But what I would PREFER happen is that a "standard" profile (GRACoL, SWOP, ISO) could be used by the designer and that the printer has calibrated their pressroom to a point that a standard profile should work well. I don't like "custom" profiled presses because it says to me "non-standard" or out-of-spec.

* Printer’s profiles will give you more CMYK gamut.

If it's a good custom press profile, maybe. But it will only give you more CMYK gamut if the press has more CMYK gamut (duh!) than what you assumed to begin with.

* The GRACoL CMYK profile is best.

It's an awesome profile, it really is....for commercial sheetfed offset printing. If you're printing web offset publication work however, you should be looking at using one of the "G7" SWOP profiles instead. But, in theory, if you've separated using the GRACoL2006_Coated1 profile, it should print well with other "G7" printing conditions just with perhaps an overall different color gamut. You could do way worse than using the GRACoL profile for separations.

* Pantone colors are impossible to recreate with CMYK—let printers
do it.

Assuming they've profiled their pressroom, they potentially could have a better CMYK recipe for Pantone process simulation than the default values. But I don't know of many printers that will take the time to do that for a customer unless the customer explicitly asks for that kind of service and is willing to pay for it.

* Let printers convert everything to CMYK because they know how to
control GCR.

Well, they should know what sort of GCR or separation parameters work best in their pressroom...but how many take the time to determine that is probably a crap shoot.

* On page layouts, leave spot colors as spot—let printers convert to
CMYK.

No opinion.

* A print job should look the same no matter where it’s printed.

Huh? So a sheetfed press, web offset press and flexo press all print the same? Sure, color management (profile or device link conversions) will attempt to maintain a similar appearance across multiple printing conditions but there's no way a job printed on a web press with #5 stock will look the same as a sheetfed press on a #1 stock. Color gamut is worlds apart.

* Pure CMY inks are possible, but they would be expensive to produce
which is why the printer industry adds black for process color.

That's color theory 101 and is mostly correct. If we had perfect CMY inks, we'd have better inherent gray balance (C=M=Y). Most of the reason for the gray balance values we assume today (50c40m40y for example) is because of impure CMY colorants but we use black ink mainly because CMY alone cannot produce an acceptable density range. But even with perfect CMY inks, we'd probably still want to add black ink to extend the dynamic range a bit.

* Always give printers RGB files unless they absolutely demand CMYK.

As long as the you've embedded the RGB profile and they honor it, sure, why not. But it's probably about a 50/50 chance that that will happen. Talk to them first to get a read of their CM IQ level.

* Use 16-bit images—the speaker saw a totally underexposed 16-bit
photo corrected into a perfect daytime shot.

16 bit can do no harm and storage is cheap so, yea, why not keep things in 16 bit?

* Printing on uncoated stock can look exactly like printing the same
file on coated stock.

Horse hockey. The dynamic range of uncoated stock is no where near a coated stock. That aside, if they're simply saying the results can look very similar except for pure saturated colors and maximum blacks then, yes, it's possible.

* All images need an embedded profile.

No question, embedding profiles do less harm in the hands of a *knowledgeable* and color-management savvy prepress person than not having any profile at all.

Regards,
Terry Wyse
___________________________________________________________________________

Re: Color management facts
Posted by: Dan Margulis
Sat Feb 21, 2009 6:39 am (PST)

Jeremy writes,

I attended an InDesign User Group meeting last night and the speaker was the
director of color management for a local printing company. A lot of what he
said flew in the face of what I have been taught by Dan and others about
color management, so I wanted to put a list of some of his main points in
front of this group and hear your reactions.

An InDesign user group and this list are two different animals. It is conceivable if not probable that the user group has few people in it with retouching experience. In that case, the printer's representative may actually know more than they do. For ourselves, however, it's yet another of the constant reminders that we get that the typical printer knows almost nothing about prepress and their advice on the subject is generally to be ignored.

Trusting printers to correctly convert files or to know what profiles should be used is an exercise in masochism. This particular guy should clearly not be allowed to make RGB>CMYK conversions--and he's the company's guru! He's not the guy who will actually be doing the work, who presumably knows even less. So my advice must remain, unless you have verified that this is one of the few printers who has some clue about what's going on, make your best estimate of his printing conditions, separate accordingly, don't embed a profile, and do attach an instruction saying that if the file is reseparated without your consent you will not pay for the work.

Meanwhile, in the "I told you so" department, an astonishing color management mea culpa has just been issued by the Seybold Report, blaming itself for being taken in by color management claims in the 1990s and conceding that little progress has been made. In their lead editiorial, they resurrected a piece of the 2001 article that I co-authored, which quoted several industry leaders at length, including the Seybold Report itself, making predictions about the demise of CMYK and how printing would become a commodity--predictions that had been proven incorrect by 2001 and look ludicrous today.

"While progress has been made by organizations such as the International Color Consortium (ICC), Fogra and IDEAlliance, color management is still something of a mystery to many designers, publishers, brand owners and advertisers. Quark and Adobe have implemented profile-based color management in their design applications, and proofing vendors (hard copy and virtual) have taken heroic measures to incorporate sophisticated color models into increasingly more intuitive packages. However, "push-button color" - if it really exists - is still a long way off."

After repeating the 1990s quotes with their lame predictions, at
http: //www.seyboldreport.com/color-management-predictions-revisited ,
the editors concluded:

"Besides teaching us some humility, these pronouncements shed some light on the issue itself. Prepress color has, arguably, come closer to a manageable level - particularly in high-end prepress and premedia shops. However, the design world is still struggling to cope with predictable color over multiple media. Brand-related color consistency has a track record of sorts in the print advertising world, but online and mobile color control is in its infancy. Because color is not a "done deal" - even in the world of print - we will continue to follow developments in color workflow, standards and control. There is much to be learned, especially on the non-print side. We will also take a lesson from our past, and refrain from making color predictions not warranted by real-world conditions."

Dan Margulis
___________________________________________________________________________ .

Re: Color management facts
Posted by: "john castronovo"
Sat Feb 21, 2009 6:39 am (PST)

I'll address just one of these since I agree with what's already been said. The photographic color process uses no black. All of the color and dynamic range in a transparency or photographic print comes from just CMY dye colorants and it's typically far more gamut than can be printed on a four color ink press, so it's obviously possible to print full color without black, but just not practical for today's presses.
___________________________________________________________________________

Re: Color management facts
Posted by: "Jeremy Schultz"
Sat Feb 21, 2009 10:17 am (PST)

Responding to Dan's comment about an IDUG being far different than this list, I totally agree‹the attendees were almost all graphic designers specializing in layouts and probably not much color correction, if they do any at all.

The main points I wondered about in that list was (1) the existence of pure CMY inks, which I thought did not exist, and (2) the superiority of an RGB-only workflow. We?re taught that CMYK is a valuable colorspace for some corrections, and yet the speaker advocated never going into CMYK.

Jeremy Schultz
___________________________________________________________________________

Re: Color management facts
Posted by: "Jeremy Schultz"
Sat Feb 21, 2009 10:17 am (PST)

On 2/20/09 9:45 PM, "Terence Wyse" wrote:

That's color theory 101 and is mostly correct. If we had perfect CMY
inks, we'd have better inherent gray balance (C=M=Y). Most of the
reason for the gray balance values we assume today (50c40m40y for
example) is because of impure CMY colorants but we use black ink
mainly because CMY alone cannot produce an acceptable density range.
But even with perfect CMY inks, we'd probably still want to add black
ink to extend the dynamic range a bit.

This is one that I wondered about. I always thought that CMY inks were simply not able to be so pure as to recreate CMYK without black. This statement suggested that these inks were possible but just too cost-prohibitive.

Jeremy Schultz
___________________________________________________________________________

Re: Color management facts
Posted by: Dan Margulis
Sun Feb 22, 2009 12:14 pm (PST)

Jeremy writes,

The main points I wondered about in that list was (1) the existence
of pure CMY inks, which I thought did not exist,

The theoretical arguments about why we need black in color reproduction are an attempt to find a logical explanation for something that may not seem obvious to the image-centric. Photographs aren't the only thing on the printed page. There is usually text as well, and the text must be printed in black ink. So, the reason we use black ink in color reproduction has nothing to do with theory and everything to do with the fact that black ink is already guaranteed to be present for text purposes and the printer will charge a great deal more if we demand a fifth ink.

and (2) the superiority of an
RGB-only workflow. We’re taught that CMYK is a valuable colorspace
for some corrections, and yet the speaker advocated never going into CMYK.

Stressing that we are speaking about an RGB-*only* workflow--no CMYK, no LAB, no way, no how--what the speaker was advocating is essentially the same as suggesting that we should all go out and campaign for Bob Dole, as otherwise Clinton may be re-elected. There was near-unanimity among industry prognosticators in the early 1990s that the RGB-only workflow was inevitable. That view, as the Seybold Report commentary indicated, has long since hit the ashcan of history.

Dan Margulis