Untitled Document

For several years I've been reading articles by Dan Margulis and many others about the dangers of sending tagged CMYK files to printers or service bureaus when there is no certainty that the recipient would not hose the files by converting them to a different profile. Every time I thought I understood the reasons, I was wrong.

In a last effort to get an answer on my own, I took a single RGB image and made four duplicates. Each was converted first to sRGB and then to one of four different CMYK profiles. Each was tagged with its own individual profile. All four looked identical. Then I converted all four to the same CMYK profile. All four, including their channels, were identical in appearance when I was finished. Frustrated, I converted each of the original four duplicates to one of four different RGB profiles and converted all four to the same CMYK profile. No visible difference. Finally, I converted all four CMYK images to a different CMYK profile. Again, there was no difference.

What, exactly, is the problem with a tagged CMYK image? This may have been explained many times in this Forum, but perhaps I was too dense to catch the meaning of the posts.

In simple language, what can happen to a mishandled CMYK image? What kind of mishandling are we talking about, and what is the result?

I will certainly appreciate getting an answer from someone who understands what is going on here. At this point I'll even be happy even to hear from someone who thinks they know the answer. While far from understanding everything about Photoshop, this is one of the few things in that still frustrates me.

Howard Smith

___________________________________________________________________________

Yesterday I received a message which began as follows:

**********************

Date: Wednesday, January 3, 2007 8:42:29 AM

From:

To: Dan Margulis

Dan,

I am sending this off line since I am not really interested in

whipping up the hard core calibrationists. I am interested in

knowledge, not reparte.

**************************

I have received in the last six weeks similar messages from two other list members, stating that they do not wish to discuss matters on this list because of the risk of provoking a flood of propaganda and abuse, and requesting a private response.

Similarly, after a particularly poisonous period this past fall, we lost a couple of the list's more knowledgeable members who left in disgust.

There is a clear consensus among the moderators that we will not permit further extended threads on the following topics:

1) Claims of general superiority for 16-bit correction, except when supported by specific color photographs on which it is claimed to be beneficial;

2) Discussions of ultra-wide gamut or ultra-low gamma RGBs, except when accompanied by reference to specific color photographs on which a use is claimed to be beneficial.

3) Claims of damage due to "quantization" "error" in conversions between any colorspaces, except when accompanied by specific color photographs on which the effect is visible.

The interminable discussions are available in the list archives, and there is no room for more unless someone has significantly new content to add.

Similarly, we have this morning seen a thread reopening the topic of whether profiles should be embedded in CMYK files before handing them out to printers. This subject has been beaten to death over the years and it is obvious from looking at the first response (which could have been cut and pasted verbatim from 1998) that those who would drag us back into the last century are incapable of learning from the experience of the last nine years.

Accordingly, with respect to the thread entitled "Tagged CMYK vs. Untagged CMYK":

1) One post per individual will be permitted, EXCEPT that I will reject any post feigning ignorance of the fact that errors have occurred in certain cases due to the inclusion of CMYK profiles that were used inappropriately by the printer or demanding evidence that such problems occur or links to the many posts on this list that document them.

2) Subsequent posts to this thread will be permitted at moderator discretion, and may be rejected if the moderator does not feel that they advance the ball.

3) I will probably post a brief response myself. This is not it.

Returning to the top of the message: the reason this list exists is so that those who are interested in my teachings can talk with other like-minded individuals and get assistance more quickly than I can offer it. In answering correspondence I have to give priority to clients, friends, this list, and former students, more or less in that order. Throughout most of 2005 and 2006, this meant that unsolicited technical questions sent to me privately usually took four to six weeks for a response. That's not fair, that's why this list exists, and we will attempt to make sure that nobody is intimidated from using it as it was intended.

Dan Margulis

___________________________________________________________________________

Hi Howard, comments in-line (I almost said ANSWERS in-line but that might be over-reaching a bit!)...

On Jan 4, 2007, at 1:09 AM, Howard Smith wrote:

Probably what you're missing is the fact that, as long as the CMYK profile you converted to is still assigned/tagged to the image, they will almost by definition look virtually identical. If you start soft-proofing these images using at least "Simulate Black Ink" checked, you'll start to see the differences.

If you want to see what's really going on, after you convert to the CMYK destination try assigning a common CMYK profile (US SWOP v2 would be a decent choice) to all the images. Now you'll start to see the different CMYK conversions rendered to a common/single printing condition. The differences between each should be more apparent.

What's generally alluded to is if your tagged CMYK image gets in the hands of some idiot that has his color management prefs set to AUTOMATICALLY convert the image with no questions asked. Even then, as long as this idiot's CMYK working space is appropriate for his specific printing conditions, there's potentially LESS damage going to be done with this conversion compared to leaving your image alone. The best, but not only, example I can think of is if you've separated for a total ink amount that's, say, >300% with perhaps light UCR/GCR but is going to be reproduced either on an uncoated sheet or perhaps newsprint. Your image if left untagged and therefore unconverted would look like hell if left alone. It would be totally appropriate in this case for the printer to convert/re-separate your image for their printing conditions.

There are a few SPECIFIC instances where a CMYK-to-CMYK conversion is undesirable but even these instances can be addressed by either using device link profiles in the workflow or converting using a destination profile with total ink/GCR amounts appropriate for the printing process and the image content.

I'll stick my neck out and say that as more and more printers become ICC and color management-aware that TAGGED CMYK images should be the default position and not the opposite. In the hands of a pro, tagged images, whether CMYK or RGB, put you in a better position than leaving this same professional with no information at all about how the image was prepared.

Regards,

Terry Wyse

_____________________________

WyseConsul

704.843.0858

http://www.wyseconsul.com

http://www.colormanagementgroup.com

___________________________________________________________________________

I would like to expand the query to the risks of tagged CMYK in PDF files:

Does Dan -- or do others -- consider the same risks for CMYK profiles being embedded in PDFs?

If so, what about the declaration of output intent (which can be embedded as an ICC profile) that is included in the PDF/X-1a spec? Is it a risk, too?

Whatever the consensus, it's sure a damn shame that a profile cannot safely be embedded in a delivered CMYK file for the purpose of any further editing or repurposing (say web to sheetfeed or newsprint).

Rick Gordon

___________________________________________________

RICK GORDON

EMERALD VALLEY GRAPHICS AND CONSULTING

___________________________________________________

WWW: http://www.shelterpub.com

___________________________________________________________________________

Howard,

I can't speak for Dan Margulis, but I can speak from the perspective of a printer. The problem with CMYK embedded profiles is not with how photoshop handles them, as your exercise demonstrated. The problem is how other software, namely rips, handles them. Currently it is common for proofing and film or C.T.P. output to be done with different Rips often made by different vendors, perhaps with different cores, ( Harlequin vs. Adobe). This raises the ugly possibility that tagged files may be handled differently by the two rips resulting in unexpected proof to press mismatches. In my own experience running CTP and Proofing Rips I have found what I consider to be unpredictable results from sending tagged CMYK files. I much prefer to send untagged files to the rip, that way I know that no additional 'black box' changes are happening to the color. One of the Adobe preference choices in In design says it for me, "Leave Color Unchanged." That is what I want to do in the CTP Rip. In the proofer rip I want to be able to change the global response of the rip to suit output conditions, a Gracol 7 target or SWOp, or a specific press, but I want this to affect the whole job and I don't want it confused with tagged profiles on images or files, possibly applying color management twice in some cases. Another problem of course is that profile to profile conversions can screw up the black generation, turning graphics and type that should be 100% black to something like 85c, 75m, 75y, 65k, etc.

So in general for a printer or service provider, and at this point in time, results are more predictable and controllable if untagged CMYK is sent to the rip. I do NOT object to clients sending us tagged CMYK files, but if they do we will do any necessary profile to profile conversions in Photoshop or InDesign prior to ripping, then send untagged CMYK to the rip. My preference for a client who is working in CMYK already, is for the client to convert those files to a workable profile, if necessary, before sending them to us untagged. That way we do not need to worry about what I call Double Color Managment when a file is converted to SWOP Coated v2 once, then converted to it again in a rip.

I also think it is also important that clients be able to make final tweaks to their files in the final working space that way they can be involved in the artistic decision that come from the fact of gamut compression, which rendering intents address only crudely. If have I images I am going to see printed by an offset press, I want to get them into their final color space, say CMYK Gracol 7, then fine tune the color, then tell the printer to leave the color unchanged. The best way to do this is to then send the printer untagged CMYK files.

Are other answers/workflows possible and workable, sure. Will this workflow change as technology improves, yes I think it will. Is proper in rip handling of tagged CMYK files possible now, it probably is and it may improve with Adobes new PDF Print Engine technology - we'll see, perhaps in rip color management will become as easy and standardized as In Rip Trapping has.

In summary, with so many variables in how different rips handle files tagged with profiles, the safest route is to send untagged CMYK. As a service provider our goal is to give the customer what he or she wants as quickly and efficiently as possible, so we don't like surprises in a production environment and we haven't got time for 'experiments' during a production shift.

See what others think, but that's my take on it.

John Gallagher

Prepress Manager, Val Print

___________________________________________________________________________

On 1/3/07 11:09 PM, "Howard Smith" wrote:

Only one as far as I©ˆm concerned and the ONLY reason I do strip CMYK profiles; size. Each takes up about 1.5mb. If I know for a fact that my output ready CMYK files will not be viewed or altered (and I don©ˆt want that anyway), the profile just takes up a lot of space. If you have a catalog of 100 images output ready and you know the numbers will go to the device Œas is©ˆ all you©ˆre doing is adding 150mb of unnecessary data.

Some would have you believe an embedded profile will cause Impotence and male balding but of course, this is the mindset of the uninformed and paranoid. I wonder how well they would run a press if their inks came in the same unlabeled containers. A profile is simply a label.

However, in the case where there©ˆs any question that someone needs to view the image or alter it, the profile is absolutely necessary and worthwhile at triple the file size.

Of course you could send 100 untagged CMYK images with a single profile and a note allowing the users on the other end to now know the meaning of the CMYK numbers. How hard is that?

Andrew Rodney

___________________________________________________________________________

On 1/3/07, Howard Smith wrote:

A mishandled CMYK image can get hosed by someone who doesn't know what they are doing or isn't paying attention.

In some shops where I've worked every workstation in the place had a different CMYK setting in use. Maybe one guy just did a Max black conversion, and another guy needed to convert to Newsprint, and the way they did that was by changing their color settings and going Image > Mode > CMYK. (Don't assume everybody uses Convert to Profile just because you do).

That doesn't really hurt anything when you are scanning untagged CMYK, as every color house used to do. It doesn't even hurt anything if you have your preferences set up correctly. It's a BIG PROBLEM if you don't.

Imagine that setup when you send files to a printer. You have tagged SWOPv2 in all your images just like you were told, but the guy who opens them has a different CMYK setup (newsprint) and has Photoshop set to Convert *automatically*. He got tired of seeing the profile mismatch dialog box and so he checked the box that said "stop telling me about whatever profile stuff you are doing."

In another scenario imagine a RIP that is set to automatically convert images that have a profile attached. The printer doesn't even know it's set that way because they don't attach CMYK profiles so it's never come up before.

Don't think those are realistic scenarios? Think again - I've seen both of them at placed where I have worked.

Sending tagged CMYK profiles shouldn't do any damage, but it can and does. Sending untagged CMYK prevents dummies from messing things up. Unfortunately it also makes it harder for experts to make intelligent conversions. What do you do?

What many have decided to do is to become the CMYK expert and send untagged CMYK files to the printer.

-----

J Walton

___________________________________________________________________________

Your basic Urban Myth created by Dan to sabotage ICC color management :)

Files in the hands of people that don©ˆt have a clue will get destroyed with or without a profile.

Profiles don©ˆt kill images People Kill images !

Its all in knowing when to convert, how too and when not too !

This argument has been kicked around HERE way too long......time to move on.

John R

___________________________________________________________________________

On 1/4/07 11:01 AM, "John Gallagher" wrote:

I may be over my post limit but hopefully asking for answers without comments will let this slip through. If not, please answer via email if possible.

Questions: Do you know how many of the more common RIPs will have such an issue? At least current rev©ˆs (not some 8 year old copy of the RIP)? Does having a Postscript compatible RIP make a difference?

How does the RIP Œknow©ˆ what the initial CMYK numbers represent? How do you tell it ©¯these CMYK numbers are in this color space?©˜ I have to assume that a RIP that would chock on an ICC profile still needs to make some assumption about the source color space of a file to convert to another color space. How do you communicate this?

If you set the Rip to leave the numbers untouched, doesn©ˆt that present a lot of ugly output? Obviously getting an untagged CMYK file that©ˆs fit for output will fly but what happens in the other cases?

Andrew Rodney

___________________________________________________________________________

On Jan 4, 2007, at 1:02 PM, J Walton wrote:

If his printing conditions are truly newsprint, this would likely still produce a better result than sending SWOP directly to his press printing on newsprint.

Then, again, they didn't take the time to get to know their RIPing system better. It's their responsibility to know how the system works or get trained on it so they do. I spent several years installing and training on RAMpage RIPs and it was my responsibility to make sure they were aware of these sorts of potential issues.

Don't hire dummies and put them in charge of pre-flighting and handling customer supplied images.

Regards,

Terry Wyse

_____________________________

WyseConsul

704.843.0858

http://www.wyseconsul.com

http://www.colormanagementgroup.com

___________________________________________________________________________

Den 4. jan. 2007 kl. 17.52 skrev John Romano:

Agreed. In addition: The best rule to follow in my opinion is to always talk to the printer. Ask them what they would like. Do they want CMYK-files at all? If so, which profile are you supposed to use? Do they want the profile embedded or not? It all depends on what the printer replies to these questions. If they don't know, it's a clear signal that you should chose another printer.

It is surprising that so few of us who actually communicate all the time are able to ask such simple questions.

Per Arne

___________________________________________________________________________

Hi John,

In terms of proofing RIPs, tagged files should generally not be a problem. Every proofing RIP that I've come in contact with (ColorBurst, GMG, Oris Color Tuner, EFI Colorproof XF, etc.) either ignores ICC profiles outright (GMG MX4 device link workflow for example) or has an option to honor or ignore embedded ICC profiles. In a proofing RIP workflow it's generally desirable to ignore embedded profiles as the assumption is that you want a proof that represents your specific print conditions. In some scenarios it's actually desirable to create two queues/hot folders; one that ignore profiles and one that honors profiles. This way you see a HARD proof of what you're going to get on press vs. what the customer expected to get and then note if there's a potential issue here.

I would suggest if you're running a proofing RIP system that doesn't give you the option to honor or ignore embedded ICC profiles, it's time to investigate getting a different proofing system. :-)

In terms of full "workflow" RIPs such as Harlequin, RAMpage, Prinergy, et al, they generally either don't honor profiles at all or the default setting is to ignore. With most you must make a special effort to enable in-RIP color management. If there's any color management at all going on here, it's usually in the form of device link profiles that don't so much "convert" the data from one CMYK space to another but instead are simply "normalizing" files for a specific black generation (more GCR typically) and total ink amount. I believe in the future, as "standards-based" printing takes more of a foothold (SWOP and now GRACoL, ISO 12647-2, Fogra39 and more), we'll see more printers relying less on simple 2D plate curve transforms and more on device link profiles as a way to normalize jobs going to their pressroom. Supplying these printers tagged CMYK gives them an even better chance of meeting your expectation.

Regards,

Terry Wyse

_____________________________

WyseConsul

704.843.0858

http://www.wyseconsul.com

http://www.colormanagementgroup.com

___________________________________________________________________________

On 1/4/07, Terry Wyse wrote:

See above. The SWOPv2 profile he was told to use would indicate sheetfed conditions.

You're right - they didn't do their job correctly in this scenario. Does that change the fact that YOUR print job looks bad? It's kind of like crossing the street - you have the walk sign, you have the right-of-way - you can still be DEAD right.

Again, see above. The question is: "What do YOU do?" *not* "What should THEY do?" They shouldn't hire dummies, but in many cases they have. Now what do you do?

Who knows if this post makes it through. Maybe Dan's right about this subject.

[INSERTED BY MODERATOR]

Yes, it makes it through, but no more, please. Terry, J, please take it offline. As you can see, it's just going to revert to the same old thing we've been hearing for more than a decade: it is OK to have a job ruined provided it's somebody else's fault; it's all a matter of educating the user; we've finally turned the corner on this, adoption will be real rapid now, etc., etc. --DM

{/moderator insert]

-----

J Walton

___________________________________________________________________________

I've got one chance to say that having read all of the arguments, I'm more convinced than ever that the only safe thing to do is to provide tagged RGB files and let the people who are possession of the actual printing conditions do the conversion. If they're monkeys, then it's their fault if the job is hosed, but for me to assume what paper and press the job will be printed on (or even the country it's going to for printing) is an accident waiting to happen. It's not always possible to do it this way, but I think it's safer if I can't find out how it will be printed.

john castronovo

___________________________________________________________________________

Dear Dan,

I know that you are fed up with this subject but ...

Last week our entire magazine staff was invited by our printer for the inauguration of their new ManRoland Rotoman press (impressive!) and afterward we had a meeting with some of the printer's key staff members.

My question to prepress: "what will happen if I send you untagged CMYK files that I have separated with a profile that you may not expect such as U.S. Web Coated (SWOP) v2?"

His answer: "We have agreed that you are sending us untagged files and that you have separated with our profile if you wish to change that then let me know the details when you are sending me the documents. It's not a problem we can deal with whatever you decide to do."

I can pick the phone up and be talking to Bob within 30 seconds (or so) to let him know about the files I'm sending in. NO problem!

André Dumas

___________________________________________________________________________

I work close enough with designers, clients and printers to deliver CMYK files for the work I provide. At this time, I have CMYK profiles from the better printers in the area, which are used 99% of the time by my clients. Every profile is matched to a specific "house" paper stock, and the paper grades range from a #4 sheet to a premium #1 sheet. Because of this relationship with these printers, I deliver tagged CMYK files. It has made my workflow easier, my results better and my clients happier.

I never, under any circumstances, deliver RGB files. As a photographer, I'm the first one in front of the firing squad when an image is reproduced poorly, so I've learned to become outrageously stubborn when someone requests --or demands-- RGB files. I'd rather not shoot the job than shoot it, only to have it reproduced poorly and then get blamed for the problem (yes, this has happened and it's not worth it).

I will say that in the pre-press departments at these printers (one of which is R.R. Donnelley), most of the guys I deal with are not sold on profiled printers. They feel there's too much press variation within a single print run and are hesitant to provide an ICC profile that's based on a press condition plucked from a micro-moment in time.

I know the regional sales reps for several of these printers and I *always* ask them how they felt a particular job went and how I can improve my work to make their life better. I get so much mileage out of this approach, that when a proof comes out of their pre-press shop that looks odd or problematic the sales reps do not hesitate to call and ask me to take a look to see if something should be done. Again, everyone benefits. The printer feels better about the color they're printing, the client is happy everyone's hustling to get the job right, and my work looks better. This is a benefit tagged CMYK files can't touch.

~ Chris Brown

___________________________________________________________________________

Dan, I've read all your books from Makeready to Professional Photoshop 5th ed., rereading one of them until it literally fell apart in my hands. In addition I've read hundreds of posts about sending tagged vs. untagged CMYK files to printers and service bureaus. Please correct me if this is my error, but not once do I recall having seen specific reasons for not sending tagged files. Instead I kept running into quasi explanations like "A tagged file can be ruined if some ignoramus mishandles it." or "A tagged file may be hosed." , or "It's dangerous to send tagged CMYK files to strangers." etc. That doesn't tell me a thing except that tagged CMYK files are frequently subject to mishandling. When I sought more technical, informative answers, quite often I was told that this had already been covered and it wasn't a subject of any further interest to anyone on the Forum. Covered where? Recently, in one of your magazine articles, you spoke of a file, and perhaps even a reputation, being ruined because somebody somewhere along the line failed to honor an Adobe RGB tag and instead converted the file to sRGB before separating it for printing. You presumed that everyone who read your article would have had enough practical experience to understand what had happened. Not me. It took awhile for me to figure it out, but I presume that the culprit used Assign Profile to achieve this. That, at least, could be reasoned out. But until now I could not find anything that came right out and said that some RIP's were double managing color and thereby ruining perfectly good files. How can someone who wouldn't know a RIP if he saw it could possibly figure that one out? Even now all I know about RIP's is that they are some form of expensive software used to separate files for printing. Granted that the business about RIP's is not the only answer because most human beings are quite capable of doing considerable damage either through their ignorance or by trying to correct what they perceive as someone else's error. The argument could be made that I would know all about RIP's if I was working with files being sent out for offset printing. Yet, from what I've seen in this forum, many a hapless person is given the task of preparing and forwarding files when they have no or very limited experience with any aspect of commercial printing.

The responses to my post do not completely answer my question, yet it is extremely unlikely that I, personally, will ever have a need to prepare a file for offset printing. Nevertheless, while I had enough scientific curiosity in the beginning not to be satisfied with earlier, vague answers about a subject that is deemed by so many to be of such vital importance, I have had enough information provided this time to satisfy my curiosity about this subject.

Not meaning to beat another subject to death, either, but the endless discussions and heated arguments about ProPhoto vs. more conservative color spaces has been of invaluable assistance to me in my own work. The fiery rhetoric that permeated many of the posts was of no interest to me and certainly never offended or intimidated me. The good part was that even the most disagreeable posts contained bits and pieces of information that could be put together in a most productive way. For all of the useful information that I was able to obtain, I would have been more than happy to wade through far more brutal rhetoric than I've seen so far. And I might add-even it some of it had been aimed directly at me.

Perhaps those who are offended, those who are intimidated, and those who have their feelings hurt by the occasional vitriolic outbursts should seek a less inflammatory forum in which to participate. They won't learn much, but they will feel good. Forgive me for being so blunt about all this, but to my notion there's entirely too much of a tendency for participants to get their feelings hurt or their egos bruised, especially when no real malice is intended. I still remember being horrified by the many attacks mounted against someone who playfully suggested that it would be fun to see your head "explode". I don't recall the circumstances behind the joking comment, but it never occurred to me that the author bore any ill feelings toward you or toward anyone else. Why so many forum participants felt it necessary to rally to your defense, I did not understand and still don't understand. Of all the people I have encountered in my professional life, you are at the top of the list of those who are well able to defend themselves.

Dan, I genuinely admire you. You have no idea how much you've taught me, though you expressed your doubts at one point that I was ready for your advanced class without having gone through your introductory course. You underestimate your teaching ability, but that's your honest opinion, and I respect it. There is no way for you to know just how much I have learned, primarily from you, but also to a significant extent from those other recognized experts who have been so generous with their time and their knowledge. This is not to imply that my knowledge approaches yours. There is no substitute for the years of experience you have in working in the color correction field. If I was much younger, I would be competing with you. At my age, however, I much prefer to spend my time teaching my grandchildren some of the valuable things I've learned over the years in a variety of fields.

And for what may be a final word, though hopefully I won't be thrown off the forum after this outburst, I would urge all participants to leave their personal feelings in a dark closet when they come to the Forum. The only reason for participating in the first place is to advance our knowledge. Those who live for the sake of arguing need to get married. If they are married, for the life of me I don't see how they can possibly have enough energy left to fight here.

For all of the time and effort they expended trying to help me, I want to offer my heartfelt thanks to Andrew Rodney, Terry Wyse, J. Walton, Rick Gordon, and especially to John Romano who offered valuable insights from the viewpoint of a printer with experience on the frontlines. If I missed someone, it was not intentional. Every reply to one of my posts is appreciated, whether the reply is helpful or critical. All I want from this forum is more knowledge, in whatever form it may take.

Howard Smith

___________________________________________________________________________

on 1/5/07 2:02 AM, Howard Smith wrote:

I, for one, am glad you asked.

No, you did not get your question answered exactly in the way you posed it. I can't add anything better than what's been written. However I like when we re-visit a topics. New info as well as current (and maybe revised) thinking on the topic shows up.

The archives are a great resource but you're not alone in wishing that our more "famous" topics had an end-of-thread summation entry.

As the person who created that mess I'm surprised at how quickly many wanted to continue their defense of Dan even after I offered a public apology.

I hope others besides Dan have noticed the loss of many writers in the last 8 months. We could easily begin the new year by inviting them back.

I regret you felt your next position was to qualify, clarify and sum up what to me was a reasonable question.

Lee Clawson

2/\V/\7 Studio

___________________________________________________________________________

I am now back from holidays, so this is a quick reply to numerous threads:

* On CMYK PDF files:

I would supply a colour normalised PDF that is untagged. The syncing of colours would either be done manually in Photoshop and Illustrator, where the final CMYK values were fixed in place - or at PDF creation where all tagged and assumed CMYK is converted to a single final destination space (I prefer to do this manually). If one has to use a destination profile in the PDF, even as a descriptor of the document output intent (and not at the object level) then it is very interesting to compare a proper ICC profile vs. the legacy Custom CMYK (a simple CMYK and K only gradient and step wedge helps to illustrate the point if the image does not contain dark near neutral areas of higher K values). If the numbers are all that matters, then the profile should not enter the story - but if the profile is being used to control the output in some way then the description of 100%K being at say 0L or 91L (LAB numbers) can make a big difference.

* For Howard on the dangers of CMYK profiles in some cases (the benefits have been expanded upon at length in the past too) -

See my first point on the introduction of the "Safe CMYK Workflow" colour management policy by Adobe in Illus/InDesign CS2, where CMYK profiles are ignored and the CMYK numbers are simply output (in earlier versions they thought conversions were a good idea but eventually understood that end users need to maintain critical colour values.").

Also:

http: //tech.groups.yahoo.com/group/colortheory/message/15879 (Dec 3, 2006)

I'll post a quote from the full reply below:

===

" * I usually create a pre-separated CMYK PDF and then recombine the seps and run a rough proof before sending off the file for film output at the local service provider. In this case there is only time to send a *composite* PDF (mistake #2). If you prefer not to trust that the next person will make the seps as expected, do it yourself - which is my standard workflow when I have the time.

- But in "theory" there should be no problem with sending a composite CMYK PDF to a service provider to make a set of negs and an inkjet press simulation 'proof' of the data on the negs. The CMYK numbers go in and the same CMYK numbers come out, don't they?

* I receive a call from the SP that due to an ICC profile being present in my composite PDF (mistake #3) that their RIP had automatically reseparated the input CMYK values based on the ICC profile and had created new seps. Thus all the black only elements such as hairline rules and text was now in four colours and 90%K rather than 100K only and the pastel blue background had turned mauve with higher magenta values. The film was output. I was asked to resupply the PDF with the profile stripped and that we would be charged for new film a proof. If they remove the ICC profile, strange things may happen to the PDF such as text failing to output or other output glitches.

Simply having no ICC profile in the composite PDF stops their RIP from automatically reseparating the job.

* Usually I do not tag output ready files, but obviously not this time. They are useful in-house, but less so when sending a final job out that will never need any future reseparation.

We have had this similar story repeated many times in the past by many different people. Yes, we know that service providers can be less than helpful and appear to sabotage jobs and if only they would educate themselves and use the right settings then there would be no problem. The problem is not the ICC profile but the implemented CM workflow, software and people using it. If one does encounter these problems, simply vote with your wallet and shop elsewhere etc.

Meanwhile Dan, myself and other practical minded people have no problems accepting that theory does not always work as expected when applied in the real world and that it is best to avoid problems in the first place.

(Note: the RIP was a Black Magic and I am not sure if this was only for proofing or for film and proofing but both sets of data were incorrectly mistreated by the global document level profile in the PDF file, I think the film setter would have been a Scitex Dolev, but I am not sure).

Agreed, but the world is not ideal. It pays to observe and think rather than just parrot conventional wisdom that may have little application to your particular environment.

This is the message that Dan tries to impart to his students. It starts with simple curves and image evaluation by the numbers and spreads into many other areas of our professions.

Rich has recently been asking Dan for specific recommendations on various workflow aspects (that for the most part are a side issue to the main focus of Dan's CMYK output experience/commentary). For me, that is not what Dan has ever been about. The reason for his popularity as an author is that he teaches his readers to *think for themselves*, rather than just presenting a cook book or canned reply. Dan can do that too, but I believe that the greatest contribution that Dan makes as an author and educator is encapsulated in this old (Chinese?) proverb:

'Give a man a fish and you feed him for a day. Teach a man to fish and you feed him for a lifetime.' "

===

Happy New Year to you all!

Best,

Stephen Marsh.

___________________________________________________________________________

Thanks, Stephen! That was the last nail needed to pin this thing down once and for all. I read all the archive posts you referenced, but your description of a real problem, one that could be visualized by someone who did not understand all the technical aspects involved, cleared up the whole thing for me. Lacking practical experience in certain areas is a big block to understanding. Sometimes a simple illustration is all that's needed to make sense of the whole thing.

My response has been delayed because your post contained so much useful information that it took awhile to sort all of it out.

Maybe Dan could simplify this sort of thing for all of us by setting up a FAQ's section that would make it a little quicker to find specific answers in lieu of reading through post after post trying to gather enough details to make a comprehensive picture. A simple question about not using tags, followed by a concise answer, would take care of the problem of newcomers and novices asking this same tired old question over and over. Detailed information tailored to the needs of someone with experience in the commercial printing industry would be completely unnecessary. An expert would already know all of that, and a novice would have little need for it.

Granted that there are many subjects too fluid to be covered with a simple answer, there are quite a few where readers just want a better understanding and not an excess of detail or of specific recommendations. Photoshop books should explain these things in underestandable terms, but even some of the better known authors seem to lack the technical knowledge to do so. A classic example in an older Photoshop book comes to mind. The author said something to this effect: "Convert to Profile changes file numbers but not colors. Assign Profile changes colors but not file numbers. Go figure." Does that sound like something you would expect from an award-winning author? Not to me it doesn't. For things that require decision making, I agree completely with Dan's position that spoon-feeding will discourage thinking. But something like "What advantage does Convert to Profile have over Assign Profile, and vice-versa?" does not require a discussion of underlying theory, just an understandable answer.

For those who are now wondering what the advantage may be, Assign Profile moves file numbers to a different color space where the fact that the original numbers are interpreted differently will naturally produce a difference in image colors. Convert to Profile translates file numbers as it moves them to a different color space, a translation that changes the numbers so that they will produce the same colors as before even though they are now interpreted differently. Is that hand holding or just an understandable explanation? No answer would be sufficient to list all the advantages and disadvantages that the poster seemed to be requesting. But if he couldn't figure that out once he understood what was going on, that's a problem with no solution. Unfortunately-and I have been guilty of this myself-the answers supplied to posts like that in my example are often either too cryptic or they are little more than reworded textbook answers like the one referred to above, answers that do not always contribute to understanding. An expert with years of experience in the printing industry would understand just about everything about the advantages and disadvantages of Convert to Profile vs. Assign Profile, but someone lacking that experience would need to know more than the fact that one changes the numbers and the other changes the colors. Knowing why one changes numbers without changing colors and why the other changes colors without changing the numbers goes a long way toward understanding how to use those tools. We have a very difficult time trying to evaluate unfamiliar facts without being able to refer them to things with which we are familiar. Its like trying to explain music to someone who is congenitally deaf. That was my problem with tagging vs. no tagging. Without real world experience in the commercial printing industry, it would never have occurred to me that the problems could arise from using software or equipment about which I had only the sketchiest knowledge. The ones who become impatient with such questions are those who likely have forgotten that the facts and understanding that are second knowledge to them were not acquired overnight, nor was it a simple matter for them to understand everything they were learning.

Stephen, it is not your knowledge alone that is valued here. It's at least as much as the fact that you have not forgotten what you had to go through to acquire that knowledge, making you a little more tolerant of those of us who are still asking dumb questions while we're still struggling to get there.

Howard Smith

___________________________________________________________________________

Howard Smith writes,

There is no mention of embedding CMYK profiles in Makeready or the first edition of Professional Photoshop (1994) because it was not possible to embed profiles at that time. There is no mention of it in Photoshop LAB Color because it has no relevance to LAB correction and retouching technique.

In Professional Photoshop 5 (1998), there is an entire section (pp. 155-158) entitled "Should We Use CMYK Profiles?" It states certain reasons for using them but adds, "an embedded profile suggests that you are open to the possibility of having your file altered automatically somewhere down the line. For most people, that's a bad decision." It does not mention the possibility of error by the next person because the technology was new and there was no track record of such errors.

In Professional Photoshop 6 (2000) there is an entire section (pp. 228-232) entitled "A More Adult Way to Play Tag". It notes that the conventional color management wisdom had changed and that it "is now of two minds on the question of CMYK tags." For my own part, I reported, "Embedding [CMYK] tags ourselves continues to be a two-edged sword...you need to ask yourself whether the potential gain of tagging the file outweighs the risk of doing so. Having had two major jobs ruined myself by faulty tag management elsewhere, I can assure you that it's not risk-free. A file saved without a tag is less likely to be converted by mistake." There follows paragraph after paragraph of specific examples of where CMYK tags have caused problems, followed by a recommendation that you tag CMYK files only when they are radically different from traditional CMYK, the example shown being an example prepared for a CMYK laser printer with very high dot gain.

In PP4E (2002) I withdrew that recommendation (pp. 273-278) in light of further specific errors that were described, saying, "For me, this was strike five for embedded CMYK tags. Their limited benefits don't compensate for their liabilities, and I no longer recommend them for any purpose." While noting that the CCMW still was of two minds, I pointed out that it had become clear that printers had not and never would learn enough about embedded profiles to make this a reliable means of transfer. I added, "In real life, [CMYK] files are commonly prepared with incomplete knowledge of the printer's capabilities...The last thing I want is to embed a tag and suggest to someone that they should reseparate my files to some other settings on the theory that the tag represents something highly specific."

In PP5E (2006) I reported (p. 295-296, with further analysis on pp. 314-316 and 321-322) that at this point even the most extreme members of the CCMW now advocated removing CMYK tags before providing them to a printer. Explaining this change, I said, "The CCMW's biggest mistake has always been in assuming that CMYK follows the same rules as RGB. All the arguments in favor of using RGB tags are arguments *against* using them in CMYK. A CMYK tag can add as much as a megabyte to a file. One assumes that CMYK files will be output as is, not translated. Translations out of any CMYK, because of the vagaries of black generation and gamut reduction, are inherently less relable than from RGB. Embedding CMYK tags, expecially ones that didn't originate in Photoshop, has caused glitches in certain RIPs and certain applications. It's not like it would happen every day or even every week, but it occurs enough so that a high-volume user would conclude that there's a reliability problem."

The above recommendations apply only to CMYK tags. I have always advocated embedding RGB tags.

As per above, I have given specific example after specific example after specific example of how damage has taken place, as have several other list members. But in any event, just knowing that "tagged CMYK files are frequently subject to mishandling" should be sufficient for you to make your decision, granted that virtually no printers even know what an embedded profile is, let alone how to use it.

The list archive is absolutely littered with such information; there is probably more on this than anything else other than 16-bit. See, e.g. "The Danger of Embedding Profiles" (2002) http: //www.ledet.com/margulis/ACT_postings/ProfilingandProofing/Profiling.htm

See also the three incredibly long "Embedded Profile" threads from 2003 at http: //www.ledet.com/margulis/ACT_postings/DailyLife/Daily.htm

After that, I simply stopped collecting and posting the threads, because their content never changed.

As you can see from the quotations above from 1998 to the present, I change my position as I learn more or as it becomes more clear what general trends are. I am willing to have an argument here if there appears to be any possibility that the opponent will do likewise: that he will ever change his position under any circumstances or can ever possibly be convinced by evidence no matter how overwhelming. The lack of such an opponent is the reason that I believe it is necessary to stop discussion of this matter.

The merits have been given by others and I see no reason to add to them, other than to summarize the arguments against as saying we shouldn't take risks, even small ones, when there is no reasonable expectation of benefit, and those for as saying that a few errors is a small price to pay in the name of progress, and I don't remember the last 25 examples you pointed out to me of costly errors caused by embedded CMYK profiles, could you repost them, and I want to know what hardware and software you use, and without the embedded profiles are meaningless mystery meat, and in any event if your job is ruined it is somebody else's fault.

It boils down to computing the odds that a random stranger will know what to do with an embedded CMYK profile.

The answer to that was obvious at the turn of the century, but some on this list steadfastly decline to enter this one, or to acknowledge that embedding CMYK and RGB tags are completely different items. Clearly, a stranger is likely to ignore any tag you put in any file. Nevertheless, if the next user is ignorant, an embedded RGB tag can do only good; an embedded CMYK tag can do only harm.

Dan Margulis

P.S. A reminder that list members are limited to one post in this thread. This is mine.