Untitled Document

Backcasts and Forecasts – or "Never Say Never"?

Sometimes it's instructive to look back to the past and measure progress against the forecasts of yester-year. Colour management is one such instance. My hypothesis is that the science has made remarkable progress relative to some of the prognosis less than a decade ago. Of course forecasting dynamic stuff like the weather, the stock market and technological change is fraught with risk, so we tend to be very forgiving of our forecasters. We may recall that back in 1999 Dan didn't hold out too much promise for colour management and instrument-generated ICC profiles in pre-press (cf. "How Color Management Failed", Dan Margulis, Electronic Publishing, July 1999).

Well, what's been happening since? It is instructive to read what Michael Reichmann says about the technologies he used to produce his recent book "Bangladesh – First Impressions". That story is here: http://www.luminous-landscape.com:80/essays/100-books.shtml.

Michael showed me how he worked-up this material, I have a copy of the book and I had an opportunity to compare quite a few of his inkjet prints with the book reproductions - the proximity is uncanny, making normal allowance for the inherent gamut differences between an inkjet print and an offset reproduction; he markets the book for 30 bucks.

Readers can refer to the article hyperlinked above. For convenience, here is a very brief selective bullet-point summary showing how he used contemporary colour management solutions:

· Load the image in RGB 16-bit mode;

· Soft-proof using the printer profile supplied by the press that will be printing the book;

· Correct colours in soft-proof mode with gamut-warning toggled on; use adjustment layers to fine-tune saturation for press conditions;

· When the on-screen image is satisfactory, flatten, convert to 8-bit, convert to CMYK;

· Make an ink-jet hard proof using the proofer profile and Epson Semi-matte 250.

The book was printed on a press in China. There was no luxury of traveling to China for hand-holding the press-men. Reichmann concludes: <With these techniques one can quite accurately control and proof the book's images, both on screen and on paper. But …being able to do this assumes that you have the right equipment….The critical pieces include a good quality monitor that is properly calibrated and profiled (*). ….You will also need a proper print viewing station (or D50 light source). Put all of this together, and one can actually do a very decent job of creating CMYK files for offset printing.(* As told on Luminous-Landscape, Michael uses a colorimeter for calibrating and profiling his monitor.)

More recently, my wife, who designs and makes jewelry using exotic beads from ancient cultures, purchased a book published by "Beadazzled" in Washington DC. Beadazzled is a family-owned retailer of fine ethnic jewelry. The book, "Beadazzled Where Beads and Inspiration Meet" also sells for 30 bucks and as befitting its subject matter is a real gem. The quality of the photography and reproduction is superb, as one would expect from the person who did it: William Allen, formerly Editor-in-Chief of National Geographic Magazine. Bill undertook this project in his private capacity after retiring from National Geographic.

Bill explained to me how he did the photography and the colour management for the press run:

<All the photos were made with the Canon 20D. I had the cameras linked real-time to my computer to get an instant read on the photos using the proprietary Canon Digital Photo Professional image capture program. ………I worked with the printer to get the profiles of the specific press that was to be used to print the book. As you know, each press runs things a bit differently; so to get the absolutely most accurate color, I wanted to match the color separations to the specific press. I then made the final color separations through Photoshop CS using those profiles plus some tweaks to the specifications that I knew from my years at National Geographic……….I then went to Wisconsin to run the job on press and reviewed all proofs and press pulls. Because of all the work done in advance, the color proofs were remarkably on target. We only had to make changes in three photos in the entire book. (By the way, on two of the three that were missed… I sent the wrong file)….. Believe me, that is quite remarkable on such a project…… It's all well and good to run a job with numerous proofs if money is no object. However, I find that money is virtually always an object; so why not get to the same destination with less expense, fewer frayed nerves, and save a few trees……

The importance of these examples is two-fold: firstly, it demonstrates how using a scientifically colour-managed workflow including 21st century hardware, software and profiles can yield a high degree of accuracy at relatively modest cost regardless of the distance between the press and the photographer. Secondly, because of this, it has democratized the whole process of publishing fine-art books by putting this technology into the hands of individual photographers and small enterprises who can then produce very high quality output for sale at truly modest prices.

Have things changed between 1999 and 2007?

Mark Segal

June 18, 2007

Note: Both Michael Reichmann and Bill Allen have seen this material, know it is being submitted to the ACTL and agreed to be quoted.

___________________________________________________________________________

Last week on The Online Photographer, Ctein described the methods used to publish a book by Bill Atkinson and printed by Vanfu. Bill waved his check book at the printer who agreed to profile a press specifically to print his book. The process was long and costly but at the end Bill had a book which Ctein says has the best colour reproduction of any book he has every seen. Vanfu was impressed enough to convert their regular production over to this method and have reaped considerable reward since doing so. Obviously the truism that an offset press can't be calibrated or profiled may not be true much longer.

On the other hand. My last experience with a printer, last month, occurred when an artist whose work I photographed phoned me and said her printer was having trouble with the files I gave her and that the colours on the proofs were not even close and the printer was racking up hours of time trying to correct them in CMYK. After describing what the proofs looked like I told her to tell the printer to change his Photoshop pref's so that it would honour embedded RGB profiles, which she did, and he did, and the files printed perfectly with minimum adjustments.

When the printers I have to deal with become more like Vanfu I will stop calibrating my monitor with Adobe Gamma and buy something better. ;-)

For those who might want to read the story:

http: //theonlinephotographer.com/the_online_photographer/blog_index.html

Regards,

John Denniston

www.johndenniston.ca

www.dirtbikephoto.com

___________________________________________________________________________

John,

Yes, as I mentioned in another thread just today, Bill Atkinson's book was published in 2004 and he himself described the process for getting it printed to his standards on pages 178 and 179. I agree with you - it also demonstrates that a press can be calibrated and profiled. I think what sets Bill's effort apart from the two that I described in "Backcasts and Forecasts" is that the latter were inexpensive and didn't involve the heavy overlays of intervention with printers and ink suppliers that Bill's - truly impressive - effort required.

OK - while there are printshops and there are printshops, you really do deserve something better than Adobe Gamma in this day and age!

Mark Segal

___________________________________________________________________________

Mark Segal writes,

It absolutely is. This is one reason that I post so many of my writings, and why we keep archives of this list.

Anyone who recalls the article saying *that* may wish to consider a checkup for incipient Alzheimer's ;-). But there is no need to recall anything at all, as it is posted at

http://www.ledet.com/margulis/How_CM_failed.pdf

Since there is some question as to whether the article was actually read before the above paragraph was posted, permit me to post some quotations.

The description of the article at the above site reads, "A year later [after Photoshop 5], an analysis of why the technology was not (and would not) be adopted by service providers." The article discusses not calibration, but the failure of the concept of embedded profiles (the "ICC workflow") to establish a reliable way of color communications, contrary to the prediction of AFAIK every authority in the field slightly more than a year before. At that time, the debate had been whether it was "a universal language of graphics" (Adobe's term) that would be a boon to every user, however ignorant, or would be valuable only to "a small minority of disciplined users" (my term).

The theme of "universal language" appears in the subhead of the column, and throughout. The opening paragraph likens it to the artificial language Esperanto, which was heavily promoted in the first half of the twentieth century as the universal language, to be learned and spoken as a second language by all educated persons in the world. I found the analogy to be quite close and continued it throughout the column. I pointed out that while Esperanto is not dead--it's spoken by more people than Hebrew or Lithuanian, for example--it certainly got nowhere near the universal adoption that its propounders planned.

I wrote in 1999, "The *ICC workflow* is an imprecise term. Many use it to signify the color equivalent of free love, with all files bearing profiles and subject to conversion by strangers. An ICC partisan, to me, is one who now or in the past has advocated this sweeping approach....These people, in my view, are in denial: the ICC workflow has gone the way of Esperanto. The legacy of the first PS 5 release is a minority who will never, ever, voluntarily embed profiles. The group is big enough to forever ensure that the overall concept will fail." That prediction looks like a pretty valid one from a 2007 perspective.

What Mark offers as an example of "modern" practice consists of the projects of two different photographers who desired to print books privately, using separate commercial printers. Each was given an output profile by the printer or the printer's representative. Using it, the two were able to get accurate monitor-to-print agreement and were also able to set up their personal inkjet printers to produce accurate proofs.

In this, the key piece is the printer's supplied profile. If it was good, the rest is easy. If it was bad, the photographers would have had to figure out how to make their own systems match it based on incomplete information, which they may or may not have been able to do.

The point of Mark's initial sentence appears to be that his examples are some kind of validation of instrument measurement to create profiles--and yet there is no indication that instruments were even used to produce either of these profiles, and if so to what extent they were hand-tweaked afterward. As far as the end-users were concerned, they fell out of the sky. Furthermore, one of these people explicitly went away from the machine-generated profile (if that's in fact what it was). He saw proofs, wasn't quite satisfied with the appearance, and proceeded exactly as he would have in the 1980s: he summoned an experienced colleague (human being, not a machine) to adjust the settings manually.

Personally, I don't care how the profiles were made. I care whether they are good. This is consistent with what I wrote in 1999: "Most ICC partisans nevertheless take the wholly illogical position that profiles are only valid if a machine approves of them. As noted earlier, color management boils down to making things look alike--not measure alike. The success of a calibration is not determined by how much time and money is thrown at it."

One participant in a recent thread was criticized for using Adobe Gamma to calibrate his monitor. When I revealed some months ago that I produced the ICC profile that governs my own monitor without the dubious assistance of a machine, it was stated that it proved I was against ICC color management. In view of these comments, I would have to say that my 1999 statement holds up well.

Also, both these gentlemen appear to have approached the problem in a serious, conscientious way, running tests, asking questions, and responding to difficulties intelligently. The first, if I read his explanation correctly, was not experienced with CMYK. He understood in advance that there would be gamut problems, but he was surprised to see how severe they were. When he found out, he did not become frustrated, or demand that the printer match his out-of-CMYK-gamut colors, or complain of an international conspiracy against photographers. He decided to confront the problem directly and make the best of the situation.

The second also was careful with his testing. When he encountered a problem, as noted above, he sought expert assistance. The stories of both projects are impressive. The question is, are the practices of these two consistent with "a small minority of disciplined users", or of "a universal language of color", accessible to great and small?

It looks to me like the former, and that both put a substantial amount of effort into their book, which no doubt was reflected in the books' quality. They did not just calibrate their monitors, and presto, beautiful color. As I wrote in 1999: "More than that, they understand that such ridiculous claims are one of the major reasons for the failure of the concept to date. Chris Murphy, a consultant who advocates the use of ICC profiles, summed it up in April: 'The problem is that too many people expect this to be pushbutton technology instead of a commitment. There are a lot of people...that will tell you about their pain to get to success but that they actually did make it to success, and it works for them very well. This is not magic box technology.' It's that way with traditional methods, too."

The claims here are three: 1) a calibrated monitor; 2) a means of calibrating an inkjet printer to a press in a foreign country; 3) a printer who is willing to provide an accurate guide for separation setup.

Now, understanding that I was working for what was likely the most state-of-the-art facility in the world at the time and so was a few years ahead of most people, I first used a calibrated monitor to eliminate contract proofs in 1981. By the end of that decade the entire industry was on board. Calibration of monitors had been a non-issue for ten years by the time my 1999 column appeared--any competently run facility was not having any waste due to monitor issues.

I first calibrated a digital output device to match output from a foreign press in, I believe, 1986, but possibly 1987, definitely not 1988. There were never any remakes AFAIK due to inadequacies of the proof. I am aware of several firms who were doing similar things in the early 1990s.

I first gave an accurate guide for separation to a client using RGB in 1988. By around 1993 I am aware of at least a dozen firms that were doing the same.

Since then, there have obviously been advances, just as there have been advances between Photoshop 2, which was what we were using in 1993, and Photoshop CS3. The tools are faster and more flexible, and easier for someone who does not have extensive experience in calibrating equipment. They cannot, however, improve on a calibration that had near-zero remakes.

So, if the suggestion of the opening paragraphs is that my 1999 article predicted that such color-matching could never occur, this is impossible, because all of it already *had* occurred, many years prior to the article's publication.

I agree that the process has been considerably democratized. To the extent that they make effective calibration easier, machine-generated profiles presumably made a contribution, exactly as I predicted in 1999. I wrote then, "Professional photographers, being relatively quality-conscious, willing to invest time in learning, and inexperienced in the vagaries of the CMYK colorspace, are the prime beneficiaries. Although Photoshop's calibration features are free and an ICC profile editor tends to cost $1,000 or more, it may be so much easier for a photographer to learn and implement that it will justify the expense."

However, it seems to me that calibration is one of the smaller obstacles to achieving what others were achieving 10 or even 15 years ago. The biggest factor would have been the cost: the inkjet printers that these guys have now are not just vastly better and vastly faster than what would have been needed 10 or 15 years ago, they are vastly cheaper and usable for other purposes besides contract proofing. Plus, that long ago, before the advent of Google and similar tools, it would have been difficult for these guys to hook up with overseas printers who could meet their needs.

Above all, 15 or even 10 years ago, most photographers were just scratching the surface of image manipulation; today many are experts. All the calibration in the world doesn't help the reproduction of bad images. I certainly hope that the people mentioned here produced good ones, and that their books are successful. I close with this thought from 1999: "If you aren't getting accurate color, but want to, you could do worse than starting by learning the ICC color model. With even a partial understanding of it, you'll get good results. If you master it, though, you will understand that all such systems are basically the same, and then you can live with it if it suits you and without it if it doesn't. And you will appreciate that it is no contradiction to say, color management is dead: long live color management."

Dan Margulis

___________________________________________________________________________

On Jun 22, 2007, at 3:28 PM, Dan Margulis wrote:

You can get a minority of people to actually believe the moon is made of cheese. It's not a very interesting prediction to say a minority will never voluntarily embed profiles. It's a vague prediction. It's totally meaningless. Had Photoshop 5's color management instead looked more like Photoshop 6, there'd still be a minority who would not embed profiles.

InDesign CS2 has color management on by default and the embedding of CMYK profiles on by default. And the same for Photoshop CS2 and Illustrator CS2. I haven't heard a whisper of a problem with these defaults. It does in fact allow for the communication of both RGB and CMYK color in the same document across workstations. So any prediction that wide spread profile embedding would never occur is flat out wrong from a 2007 perspective. It has happened.

Chris Murphy

Color Remedies (TM)

New York, NY

___________________________________________________________________________

Chris, are you referring to the locked "safe CMYK" policy introduced in AID/AI CS2, I believe? I can't recall if it is this or another policy that is "default" as I am not in front of it to test.

If I understand things correctly, prior to this, if there was a tag and the output intent of the job was different to the tag, there was a colour conversion. The application was designed that way, to consider profiles more critical than the files numbers, that it was fine to convert between profiles as in theory the colour had the same LAB value but with a different mix of numbers (this being the problem, colour build). When using the "Safe CMYK" policy with the padlock symbol, file values are not changed.

One can keep the CMYK tag for use in Photoshop, knowing that InDesign and Illustrator will ignore it and output will not be screwed up due to the image profile not matching the output profile.

Was not the inclusion of a "safe CMYK" workflow in CS2 the result of negative user feedback? Did Adobe suddenly "see the light", years after Photoshop 5 colour handling? Did they get new consultants with press workflow experience? Compared to QXP, AID is so much better, but at least QXP did not change files values by default. The "safe CMYK" workflow policy of AID CS2 was long overdue. It should have been obvious from the start to anybody with any CMYK press workflow experience, that most print users don't want their files values being messed with unless requested/agreed. The final files numbers are that way for a reason, wanted/required, for right or wrong. Nobody cared that their drop shadow was not "colorimetrically correct", it was numerically correct - which provided the correct print response and the desired visual response in a human observer. This is fine for one off use in a known condition, but AID can use CM to repurpose jobs on the fly, which used to be done as a separate task with separate files. There is no doubt that CM can be of help in such cases, but one may not wish this "help" as a general workflow - thus the "safe CMYK" policy.

If you are referring to a "non safe CMYK" workflow policy (any one that converts or does not have the padlock symbol in colour settings) in AID or AI CS2, and that you have heard of no problems with such a policy - then I would be surprised. This was a big reason to move up from earlier versions, it made things safe and workable for those that do wish to tag their CMYK files but do not wish this to have a negative impact on output due to software design.

Sincerely,

Stephen Marsh.

___________________________________________________________________________

On Jun 22, 2007, at 6:24 PM, Chris Murphy wrote:

Now who's making vague predictions? <g> To me the bottom line is my color management life would be easier today if Adobe had adopted something closer to the current policies from the start. It's now safe to maintain CMYK profiles on your own internal files as you pass them between Adobe applications. The fact is that after PS5 came out even this tiny piece of color management was not safe unless you really knew what you were doing. I know I didn't, and today I shudder at the damage I caused my own files by trying to follow what I understood of the process.

Compared to other's on this list, I exchange files with only a very small number of people-- all of whom consider themselves imaging professionals. What's amazing to me is that so far in 2007 I have experienced 3 separate situations where I have been given untagged RGB files. These are files from *supposedly* knowledgeable users who have gone to the trouble of *removing* the attached profiles before passing on the files! I have no idea why they think it's the right thing to do, but I do know they have to do extra work to do this and so it's not a mistake. I have even seen retouchers open these kinds of files, think the color looks bad, and start correcting them to bring back the desired appearance. I don't believe these are unusually stupid people. I do think this whole subject is more difficult to understand than most color management professionals acknowledge in their teachings.

If I can have so little confidence when I receive RGB files, what should I think when I receive tagged, or untagged, CMYK files?

Ric Cohn

___________________________________________________________________________

Hello Dan,

I agree with you that learning the ICC color model will help anyone who whishes to get accurate color. I do reproduction of watercolor originals on the very same papers used by the artist, Arches watercolor paper, Fabriano etc., and getting an accurate reproduction is essential. It cannot be done without accurate profiles at all stages of the reproduction; Scanning (scanner profile), Viewing (monitor profile), Printing (printer profile).

You say that "you will understand that all such systems are basically the same" what does that mean ? Are there many such systems and if so why is it important to understand that they are basically the same? I use ICC and don't think that Photoshop would recognize other models.

Then you say: "you can live with it if it suits you and without it if it doesn't" I can't imagine anyone working on an image for hours and *not* wanting to have it reproduced accurately. What do you mean when you say that you can live without it if it doesn't suit you ? I can't imagine anyone wanting inaccurate color.

André Dumas

___________________________________________________________________________

Ric:

Our entire society, much moreso than most of us realize, runs on confidence. We must have confidence in our base assumptions to take risks. I'd like to build some more products, but it's still a challenge to get them to the market looking as good as I'd like.

If they look good, they sell better and I prosper. If they don't, I end up eating Kraft dinner instead of filet. Or worse.

How could we have more confidence in color management? I've asked that question before here and elsewhere and it always sparks cynical responses.

What we have is a competitive marketplace that is endorsed for it's "efficiency". Never mind that it seems pretty inefficient to ruin a job because something went wrong in color management.

Efforts to set universal standards with broad consensus are a no-go in this environment because it's too socialistic. Practically no one is willing to sacrifice their own short term business interests in favor of the overall market progress.

There is no authority to regulate this across the board.

What we have instead is continual attempts by the market leaders to impose their vision on the industry. This, in turn, just happens to coincide with their business plan to completely dominate the industry, and for that reason, and a few others, there is always some resistance to change.

Their spokespeople attack everyone with a contrary opinion, calling them unscientific, out-of-date, uninformed, a dinosaur, etc.

Therefore, we go round and round, and you can be confident that this will continue.

When will this change? Perhaps when we've all joined the Green Party and Dan conducts this list in Esperanto. I'm willing to give it a try if you are.

Cheers,

Ron Kelly

___________________________________________________________________________

On Jun 22, 2007, at 10:29 PM, Stephen Marsh wrote:

I'm referring to "North America General Purpose 2" color settings. Not any one color management policy. Those happen to be the default in CS2 and CS3. In IDCS, color management was not on by default.

Output doesn't get screwed up due to the image profile not matching the output profile. One doesn't inherently lead to the other. It depends on the content of the image and the behavior of the printing process. If rigorous process control is in place and registration is excellent in most instances there would be no problem with such repurposing. The areas where there can be a problem are well known but the application lacks the granularity of control needed to isolate those areas, and treat them differently than other areas. It's not an ICC problem, per se.

Like any rational company, I'm sure Adobe solicited feedback from their customers and incorporated it. Since previous versions of InDesign didn't have color management on by default, it's a big deal that InDesign CS2 does have it on by default and tags each InDesign document with both an RGB and CMYK profile and there has been not a wit of a problem with it. Now both the Document profiles and the color management policy have been embedded in the documents, making it much easier as things go forward to know the intent, the color appearance, of the document.

Neither did InDesign.

That we even still dealing with CMYK at the front end at all is a testimony to how unsophisticated our applications, workflows, and process control really are. Fine art photographic printing is routinely done without access to 6+ channels of color data that will ultimately be used to produce the image. The conversion from RGB to 6 + channel space insulates the end-user entirely from having to deal with linearization, ink limiting, black generation and increasingly also output sharpening. If a print driver for something has complex as 8-color separations can be dealt with in a fine art reproduction context, it can obviously be done with 4-colors. The problem isn't that there is a color management problem, it's that there are realities of process control and registration beyond the ability of applications and drivers to compensate for. Those who compel their presses to behave with superb consistency are in fact able to take advantage of such more complex yet more flexible workflows, and have been doing so for years.

OK well a lot of people did care. In reality most people don't care about the numbers, they care about color appearance. If you talk to designers, ad agencies, any print buyer, they care about color. They don't really care about the numbers it takes to get them what they really want.

"Numerically correct" applies only to a specific printing condition. As soon as they are using a different magenta, or print on uncoated versus a #1 stock, those numbers have to be different. Why? Because it's not the numbers that matter. It's the color appearance that's important.

I don't know anyone who has ever advocated just flipping a switch and suddenly that gets you full blown device independent color management.

That isn't a default. I do know many people who use that workflow and the problems with respect to it are that it's still dyed in the mold of tradition. It isn't flexible enough. There's only one output profile that you can choose instead of multiple output profiles to account for different amounts of black generation (or a smart black generation model that doesn't depend on ICC profiles).

There is still a negative impact to printing "numbers" as-is to a printing process for which they weren't designed. You just have a different problem than the ones you're alluding to, a problem for which there is a solution that's been around for a long time: get a proof, look at it, color correct the originals, rinse and repeat until the proofing budget dries up or you like the proof. A lot of people don't like such iterations anymore than others don't like black only text in a TIFF turning into four color text.

Chris Murphy

Color Remedies (TM)

New York, NY

___________________________________________________________________________

Andre:

Reproducing something as closely as possible could be fairly considered an exercise in "accuracy".

In my work, for off-set, I am making commercially saleable calendars. They start with photographs of course, and end up as press ready CMYK files. This is vastly different to your situation, but I feel compelled to comment because of the blanket nature of your statement and your suggestion that doing things differently than you do is uncaring, or sloppy.

I manipulate the originals for many reasons, but the most important one is "what sells?" Therefore, I am not making the reproduction "accurate" to what the original was.

I know you are thinking "that is not what I meant".

The other reason why I don't just leave everything up to profiles is, well, . . . they just aren't accurate.

For example, if I don't manipulate my blue skys, I'm going to get purple calendars. I hate purple skys; I don't think I've ever seen one but I've seen a ton of purple skys on printed matter.

I don't have the luxury of a press check for most of my work, and I have to use the off-set company's press profile. Experience has taught me to intervene regularly.

Yes, I have profiles in my workflow for monitor and inkjet proofs and probably other places as well. Maybe my telephone.

It seems to me that there are two camps on this issue:

1. profiles are great, profiles are the future, use profiles for everything everywhere all the time. If you get bad results then switch to another company that honours profiles, has better ones, etc.

2. profiles are fine in many circumstances. Don't trust profiles where experience and common sense suggest they can't be entirely trusted, and don't waste time trying to make the perfect profile - it doesn't exist.

I'm in camp 2.

It's not an all or nothing approach; I would say it's a realist's approach.

Lastly, yes, I think we should all pay attention to developments and adjust accordingly, in the event that things change. Physics and business practices would have to change substantially for me to move to camp 1, but I'll keep watching nevertheless.

Cheers,

Ron Kelly

___________________________________________________________________________

On 6/23/07 11:15 AM, "Ron Kelly" wrote:

Well you©ˆve explained clearly that the press profile you©ˆre using from the press house isn©ˆt accurate since you©ˆre not getting blue skies. Perhaps a profile that provides the CMYK numbers based on what you©ˆre seeing in RGB that are correct for blue?

Did the press change or was the profile incorrectly built from day one is the question I©ˆd be asking.

Andrew Rodney

___________________________________________________________________________

Hello Ron,

In my case "what sells" must be accurate colorwise. Not you?

I know what you mean about purple skies, if you look in the archives of ACTL you will see that a few years ago I had a lot of problems with that, and purple Holsteins cows, purple silos, purple concrete. Thanks to the list and to the calibrationists those problems have been solved.

I am also in camp 2. I was just saying that for accurate watercolor reproduction, profiles are essential and after having learned how to create *good* profiles then I find it that much easier for my other photographic work to produce accurate prints that match what I see on my display.

Nothing in my message that would suggest that doing things differently is uncaring or sloppy.

André Dumas

___________________________________________________________________________

Andrew,

I think this may be one - or two - of the fundamental points at issue here, and it would be good if you could clarify. The sentence which starts with "Perhaps....." seems to have ended-up with something left out of it, so I don't quite understand what your saying. But I'm inferring from the two questions you are asking that you think a press house can build a profile that will - within the limits of the press gamut - accurately provide the CMYK numbers that render the RGB the photographer would see on a soft-proofed monitor using that profile, and if so, I would go on to infer that any up-to-date press house most likely uses current calibration and profiling technology to do so. Could you comment on both these points?

Mark Segal

___________________________________________________________________________

André:

I guess I misunderstood you.

Ron Kelly

___________________________________________________________________________

The poster mentioned that the profile supplied results in blues that shift magenta, then he has to Œfix©ˆ the resulting numbers to get correct blues. The profile is obviously not a good one, it shouldn©ˆt be producing blues that shift. And fixing the numbers is a lot more work and counter productive to just starting out with a good profile that produces blues that look, well blue, not magenta.

This goes back to an old (and silly) post that the U.S. Web Coated (SWOP) v2 profile always produces blues that print magenta. Yes, if you send those numbers to a device the profile wasn©ˆt properly built for, blues, reds, yellows, any color could shift. If you send the right numbers to the right device, they don©ˆt. The U.S. Web Coated (SWOP) v2 profile produces excellent blues when you send the numbers to a device that follows TR001 of which the profile was built to reflect.

This press house may have produced a poor profile (or for that matter, used the wrong ink settings in the Custom CMYK settings in Photoshop) or the profile might have been fine but the print conditions changed. We don©ˆt know. We do know that there©ˆs an issue with blues and the issue is most likely fixed by a better profile or conversion process.

Without running the same set of color patches used to build the original profile and then comparing the differences of newly measured data, we don©ˆt know what the source of the blue problem might be. If you did run the same patches and compared them to the original data, we could at least see if there was device drift (if not, it might have just started life out as a crappy profile). Either way, altering the numbers of blues in Photoshop, while a solution seems like a heck of a lot more work than simply starting out with the right RGB to CMYK conversion instructions (profile).

Andrew Rodney

http://www.digitaldog.net/

___________________________________________________________________________

Thanks for the reply Chris.

Chris Murphy wrote:

OK, thanks for making that clear. Now let me be clear[er than before].

The colour setting that you refer to does use the "safe CMYK policy" by default (now that I have AIDCS2 in front of me, I can confirm this).

In fact, all of the "2" presets use the "ignore profile/honour values" method. It would appear that Adobe agree with my position as a defalt. I am not saying this should never be done (convert values), just that it is not a good default to convert CMYK values just because two profiles don't match (which is my understanding of CS or earlier workflow if enabling preserve CMYK profiles and there was an image and document mode mismatch printing composite CMYK or CMYK seps).

This does not match the results from the test that I have just performed using the colour management policy for "Preserve Embedded Profile". This was the only colour managed colour option before CS2 introduced the "safe CMYK Workflow", when using this policy colours are converted if the image profile does not match the document profile when printing/exporting to PDF.

My black only element (image) tagged with legacy Max GCR was reseparated to SWOP v2 (document profile) when output when using the preserve embedded CMYK profile option.

Please explain this further Chris.

Qualified with a very big if Chris.

Indeed.

This does not help things now Chris.

My point is that it should have been there since v1.0 Chris. My company did not move up to AID from QXP until AID CS2 *had* this "safe CMYK workflow" policy.

I submit that there have been no problems becuase by default these General Purpose 2 colour settings presets *use safe CMYK by default* - so there will be no problems like I am talking of (unwanted reseparation of CMYK elements based on their tag).

Are you saying that it would not be considered a problem if one was using the "preserve embedded (CMYK) profile" policy, imported a legacy Max GCR K only tagged file with K only elements into a document profile of SWOP v2 and printed CMYK separations or composite CMYK?

The solid K and tinted K image also with CMY areas would come out as CMY with 90%K value where it only used to have K.

This does not overcome the issues that I have raised, AFAIK.

I must have misunderstood and taken the marketing hype to heart then.

From Adobe: "Color management improvements Employ a safe CMYK workflow by preserving CMYK color numbers all the way through final output."

So Adobe thought it was an improvement to CM, to have it dumbed down so that it would not change values. It may not have happened by default, but it did happen with the right combination of settings as I have listed above (and again below).

It did convert - if the CMYK document profile did not match the image profile if using the preserve ICC profile option and not "safe CMYK", the image was reseparated to the document destination, using the embedded tag as a source when printing to composite CMYK or separations.

Is having precise control unsophisticated?

This is what working in final device space with no conversions give those that require it - precise control.

This segment of your reply is moving past the point I was making. It can be addressed in a different post if required - I don't wish to muddy things up more than they are at the moment.

Even in such conditions, it can sometimes help to work in CMYK (even if the RIP is CcMmYK). I have had one specific problem with inkjet colour builds for dark coloured areas in one job that I recall and the only thing that helped was a CMY only separation so that the RIP would not lay down any K at all and only use CcMmY for this area (a persons face in shadow under a hat brim).

This segment of your reply is moving past the point I was making. It can be addressed in a different post if required - I don't wish to muddy things up more than they are at the moment.

If you still hold this position, despite my examples above, then we will have to agree to disagree on where the problem lies Chris.

Please reread my post, the numbers with the stock/ink etc are the desired colour appearance they "care" for.

I am/have been some of the above (designer, agency worker) - and I am also prepress too.

They do if a profile or human chooses inappropriate numbers, as in this thread:

http: //tech.groups.yahoo.com/group/colortheory/message/18079

I will agree that sometimes appearance matters most, other times numbers matter most. Both rely on correct or good numbers.

That is all that I am talking of Chris.

Yes, agreed - but this has no bearing on the workflow of a user for a single condition that does not understand why their colour builds are magically being changed at output to print or PDF, because their image has a different profile to the document and they are printing seps/comp 4C with a preserve embedded profile policy.

Really, are you sure Chris?

When I load in the preset for North America General Purpose 2 - guess what is in the CMYK field? It is the safe value locked preserve numbers option, not preserve profile (alter colour values).

This is why it is used, these users just want the same numbers to come out as go in, they are not repurposing the job. If they do, they will dupe it and do it manually as a second file marked for that new condition.

Sometimes one can be so flexible that things get bent out of shape!

Not a real concern, in Photoshop the images are put in their correct GCR etc. All the files are preped for this one condition, they can have any profile and the safe CMYK workflow option will ensure that the numbers remain that way, while one can take advantage of the ICC profile in Photoshop.

This segment of your reply is moving past the point I was making. It can be addressed in a different post if required - I don't wish to muddy things up more than they are at the moment.

This is a distraction, I did not make this point, I am only talking output for a single condition, that the file was originally prepared for.

This segment of your reply is moving past the point I was making. It can be addressed in a different post if required - I don't wish to muddy things up more than they are at the moment.

Agreed Chris.

Sincerely,

Stephen Marsh.

___________________________________________________________________________

On Jun 24, 2007, at 7:16 PM, Andrew Rodney wrote:

Andrew,

Obviously yes-- and sometimes no. If a printer tells you to use US Webcoated (or Sheetfed) Swop (V2), as this is what they aim for, I'd say there is a chance that they really do. If you know for sure that this is the case you can *generally* convert and move on.

Unfortunately, unless you know for sure that they really do *and* that they have excellent process control it may make sense to take some precautions. Dan has shown very clearly in his book- with printed examples- and on this list- with CMY numbers- that US Webcoated Swop (V2) puts more magenta into blues in the range of skies than any other profile he's looked at. I confirmed this with 2 SWOP profiles I have access to that are used by 2 different large Ad Agencies for their conversions. These agencies prepare magazine ads for some of the largest, and pickiest, clients in the world. Obviously, they prepare one file for magazine repro and the magazines need to match their proofs or they don't get paid. If the skies come out consistently purple they have huge clients complaining. Why do you think they don't use the canned profile(s) in Photoshop? These internal agency profiles aren't "custom" profiles in the sense that they were prepared to match a particular press. They are custom profiles in the sense that they have been custom built to better match their output needs from presses that conform to SWOP or GRACOL or they don't get paid. My guess is that they have been built with better fudge factors for the realities of presses.

If a slightly CYAN sky would be preferable to a slightly PURPLE sky then why, as a color management consultant, are you so loath to admit that pulling a little magenta out of the skies is a reasonable precaution to recommend? There are some colors which if they change slightly in one direction are rarely a problem (skies going more cyan, skin going more yellow) and if it changes in the other direction is more likely to be a problem (skies going purple, skin going blue). There are also situations where the exact same colors (if they are not skies or skin, for example) will not matter to anybody if they make the same shift. This is where human intervention still makes sense and where I honestly can't understand your resistance to acknowledging what I see as the realities of press reproduction. Please explain.

When I am supplied a true custom profile by a printer I take it much more seriously. However, I still think it makes sense to look at the individual image and see if there are colors that if they vary in one direction are a problem and in the opposite direction would be less of a problem and to adjust accordingly. What are your thoughts on this?

Ric Cohn

___________________________________________________________________________

Andre writes,

If you are using "profiles" in the sense that I do, agreed. If you are intending "profiles" to mean machine-generated ICC profiles only, then it would appear you are saying that this type of work could not have been done before 2002 or whatever arbitrary date you might choose for the time when such profiles became marginally acceptable.

The problem of getting an accurate match to one's screen and/or hard proof is much older than the ICC. The problem did not change because the ICC specification was propounded, the methods of solving it did not change, and the means of judging it did not change. The idea of taking measurements and generating a profile from them was already implemented before the ICC itself was even formed.

There are only two novelties in the ICC approach. First, the presence of an embedded profile that was intended to facilitate communication between strangers; second, the introduction of a file format that can readily be exchanged between devices instead of a large number of proprietary formats. The first was a disastrous failure that caused great damage to the industry. The second was a great success that has empowered many users to achieve matches that previously would have required the services of an expert. The difference between the two should be understood, and we should also understand that the underlying means of calibration never changed.

To answer this all you need do is ask yourself what would happen tomorrow if you woke up and found that your profiles no longer functioned. Further, that those who say that any form of color management that is not ICC color management is not a form of color management had so irritated God that He in His vengeance had disabled all spectrophotometers, except for those used by His righteous servants for the purposes of verifying, not establishing, calibration. What would you do?

This is not a very difficult question. There are many ways to achieve the same thing. Today, most likely one would calibrate the monitor using Apple's internal calibrator, which is quite powerful IMHO. The other conversions could be handled by Photoshop Actions, which have all the power of ICC profiles and then some.

An experienced person who did something like this in 1999, when my column was written, would certainly have gotten a more accurate result than using pure instrument-based profiles, which were, at that time, appallingly poor. Today, they're doubtless much better, although I haven't tested any in several years. There's also the possibility of some hybrid that uses a little of both methods for conversions, or the possibility of adopting, say, eyeball monitor calibration but computer-generated scanner profiling and hand-tweaked output profiling loosely based on machine measurements.

Lots and lots of different ways to achieve the same result. I don't recommend

adopting one based on political correctness, but rather on whether it gets a

good result, how much time it takes to get to that result, and how easy it is

for somebody without much experience in calibration to proceed.

Dan Margulis

___________________________________________________________________________

Seems it's time once again for the bi-annual flogging of "calibrationists" and those who ascribe to the notion that embedding profiles may in fact be a good and helpful thing to those downstream of the image creator.

My position has always been...

* Embedding profiles/tagging RGB images is a must. Period. I can't see how anyone could make a case otherwise.

* Embedding profiles/tagging CMYK images is at least helpful. It is helpful if the image is being handed off to a PROFESSIONAL and is only potentially dangerous if handed off to a non-professional that has their color management policies set incorrectly. Even in this case, I would still contend that this non-pro will find a multitude of OTHER ways to "damage" an image even if the image is untagged. Even so, if their default CMYK working space is correct for their printing process and a "blind" or unintended conversion takes place as a result of a profile tag, this conversion is potentially less "destructive" than blindly sending the original CMYK values straight to press (try sending a "SWOP" separation to a press running newsprint and see what you get!).

But what I really wanted to chat about was the fact that profile conversions at prepress and print shops have been going on for quite some time whether folks on this list realize it or not. And I believe this will become even MORE prevalent in the future for one simple reason: INK SAVINGS. Let me explain...

It's a fact of life that this has been going on in the newspaper industry for a while. For them it's all about dealing with separations that have been incorrectly separated for "SWOP" print conditions where the total ink amount is way in excess of what they can print reliably. A separation with 300% total ink is simply not going to fly for a process that can't handle much more than 220%. So to combat this, newspaper prepress departments will generally have some sort of profile conversion (typically device links) that not only correct for total ink but almost by definition will perform a profile conversion/re-separation in the process. My guess is that most will assume SWOP TR001 for untagged CMYK which could possibly result in less-than-ideal color as opposed to supplying tagged CMYK. This is assuming of course that their device link/ink reduction conversion has the option of honoring embedded profiles and that they have the option enabled.

This same technology is now finding it's way into publication and even commercial sheetfed printing where ever tightening margins have forced them to find ways to reduce cost. At some medium to large printers, ink cost alone can be anywhere from several hundred thousand to in the millions annually. A (conservative) 15% ink savings can be real money for these people. The icing on the cake for this technology is that it also provides a very simple method of achieving conformance to print standards and specifications such as the relatively new G7 GRACoL and G7 SWOP. At the very least, it's quite a bit more sophisticated than using simply plate/press curves to meet these specifications.

Bottom line, I believe you're going to see MORE in terms of taking customer-supplied images and converting them using either straight ICC profiles, device link profiles or proprietary conversions of one form or another and it's just going to become another part of prepress workflow in the same way that auto-trapping has become over the past 10-15 years. At the end of the day, quality on press is going to be about color and matching a standard proof specification (here in the US, that will likely be "G7" GRACoL and/or SWOP) and not someone's notion of their sacred CMYK values.

Regards,

Terry Wyse

_____________________________

WyseConsul

Color Management Consulting

G7 Certified Expert

704.843.0858

http://www.wyseconsul.com

http://www.colormanagementgroup.com

___________________________________________________________________________

It's fairly meaningless to talk of "SWOP" profiles without some sort of reference to where the measurement data came from. OTHOH, if one took the now dated TR001 characterization data and built a couple of profiles using different profiling software and/or gamut mapping options and then compare these to Photoshop's "USWebCoated(SWOP)v2", then you'd have a meaningful comparison. But to compare profiles built from different "SWOP" data sets is meaningless.

I'm guessing here but I suspect that some of this notion of "too purple" skies from Photoshop's SWOP profile comes from folks that may have been using profiles created from proofing systems (Matchprint, Kodak Approvals, etc.) and not REAL press data like the Photoshop profile created from TR001 press data. Let me explain:

Analog and virtually any (non-color managed) digital dot-proofing system "suffers" from having near-perfect "wet ink trap" characteristics. In other words, overprinting solids "trap" with near perfect efficiency. In the case of C+M overprints, these would have a relatively purple bias since the magenta colorant is going to overprint cyan perfectly and not show the relative loss of trapping efficiency as you'd have on a press with the typical KCMY (magenta over cyan) ink sequence. A profile made from such a proofing system would compensate for this natural purple bias by reducing the amount of magenta ink when overprinting cyan ink during the CMYK conversion. When this goes to press however, this C+M combination would tend to shift towards cyan due to the less efficient wet ink trap of the printing press. By the same token, a profile made from REAL press data such as TR001 would see the less efficient overprint characteristic of magenta over cyan and compensate by adding relatively more magenta in blues. Make sense?

My advice to anybody concerned that this is an issue but still wants to produce "SWOP" separations, they should go to either www.gracol.org or www.swop.org and download the latest "G7" family of profiles. Theirs one flavor of GRACoL for Paper Type 1 (commercial sheetfed on typical gloss coated paper) and there's two flavors of SWOP, one for Paper Type 3 (a neutral white coated publication paper) and one for Paper Type 5 (dirty yellow publication paper, same as TR001). The SWOP2006_Coated3 profile is essentially the new default SWOP profile and should be considered a replacement for Photoshop's supplied SWOP profile. I can tell you that these are VERY GOOD profiles all made from real press data. If you just can't leave well enough alone, you can also download the characterization data that was used to build these profiles and roll your own profile, provided you own a profiling application that will accept standard CGATS data.

Regards,

Terry Wyse

_____________________________

WyseConsul

Color Management Consulting

G7 Certified Expert

704.843.0858

http://www.wyseconsul.com

http://www.colormanagementgroup.com

___________________________________________________________________________

On 6/25/07 7:25 AM, "Ric Cohn" wrote:

One would hope yes.

Such as?

So what, it doesn't make it right (he's wrong, I can tell you this from experience sending data to output devices that DO conform to TR001. That Dan hasn't seen other CMYK profiles that exhibit this behavior doesn't make what he's saying correct because its not. I also don't know that Dan has enormous experience working with all nature of CMYK ICC profiles.

But this is easy to test as I've done. Trick is finding someone's who's press or proofing system does indeed conform to TR001 and send a blue sky to this device. Lots and lots of users have done this without skies going magenta!

Why don't they use the correct press conditions? That is, right being a press condition that the profile was built for.

Maybe but it appears if you're saying they have an issue with blue. I suspect this is a profile issue but it could be user error elsewhere in the process. But the idea that an ICC profile always produces a blue shift is simply not correct and silly. Now if every printed piece shifted to blue, OK, we'd have something to discuss. Clearly this isn't happening.

Andrew Rodney

___________________________________________________________________________

Ric your message reminds me of my battle in past with purple skies and purple galvanized metal silos, etc.

So at one point Stephen Marsh mentioned that a lack of yellow could be "one" of the culprits, and *that* solved many of my problems with blues turning purple.

André Dumas

___________________________________________________________________________