Color Management
February 2002

William DeMarco
Worldwide Product Line Manager - Proofing
Kodak Polychrome Graphics

Executive Summary
Although the theory and many of the tools of color management are somewhat complex, the concept itself is straightforward. Color management in its broadest sense is a system of automating and controlling the transformation of a color image from one color space to another, with the goal of maximizing color accuracy and consistency throughout the production process. This is not exactly a new idea; color professionals for many years have been transforming color images from one color space to another, and they have also been calibrating the tools used in the various parts of the production system, either to one another or to some agreed-upon standard. Their goal in doing this has always been to deliver a high level of color accuracy and consistency--from proof to press or from one press to another or both. "Color management," then, refers to automation tools--some in existence, some still theoretical--designed to help establish and maintain color.

Areas of Controversy
Given the amount of arcane and sometimes rather lively discussion this topic has engendered over the past decade, the reader is entitled to wonder if perhaps there isn't a bit more to it than that. Much of the seeming dissension among color management theorists and practitioners is traceable to a lack of agreement over three basic issues.

• The chicken-and-egg dilemma. In order to calibrate an instrument, there must be some accepted standard to with which it is to be brought into agreement. There is by no means universal agreement as to what that standard or those standards should be.
• Scope. There exists a very broad concept of color management as a kind of solar system in which the needs and concerns of the printing industry are confined to one planet, not necessarily the largest or most interesting one. There also exists a narrower concept in which color management is primarily a matter of applying statistical process control to color printing. Clearly, there are some incompatibilities between these points of view.
• Degree of automation. Some observers look forward to a day when establishing and maintaining color will be a software function requiring little or no human intervention. Others believe that the human eye--the trained, skilled, experienced human eye--is irreplaceable in maintaining control of color.

Variables and Decision Points
It is not necessary at the moment (and it may not be possible) to resolve these issues, but it is necessary for printers to be aware of them. Most printers today are in the process of making (or have already made) a transition to digital prepress production. In making this transition, printers find that there are more variables in a digital system than there are in traditional analog prepress.

On the one hand, this is a good thing--the greater the number of variables in a system, the greater the freedom that exists to fine-tune and control it. On the other hand, each variable creates a decision point, and in implementing a system, printers often make decisions without being aware of it and without being aware of the effect those decisions may have on other parts of the process.

The Big Pieces First
Based on experience gained from working with thousands of printers, Kodak Polychrome Graphics (KPG) advocates making certain basic decisions and adjustments first, and then, having stabilized a couple of key components, calibrating the rest of the system to them. In KPG's experience, the only really successful way to do this is to start with the press and work backwards. There are tools that can measure the printing characteristics of an output device and determine what its color behavior is going to be across the entire gamut it can reproduce. (The Press Aim tools from Kodak Polychrome Graphics are specifically tailored to do just this. In their earlier incarnation were the precise tools that were used to measure the presses while the SWOP standards were being developed.) From that information, it is possible to determine the optimum prepress information--what should be on a set of plates for that particular press to get it to print a particular color as accurately as possible.

The press data and the prepress data are married in the digital proofing system. With a verifiable digital proofer like KPG's Approval system, the printer can take density and dot gain readings from a proof and make certain that the proofer is achieving its aim points. This will verify that it is faithfully showing what that particular press is going to print from a set of plates made with that particular prepress data. At that point, the printer has the core essentials of a calibrated system.

As noted, there are additional tools that can help streamline and automate parts of a color system. These tools are based on what is called ICC (International Color Consortium) profiling, which we will describe in general terms shortly. Before getting into that, however, let's back up and take a look at what can happen when a decision point is passed without being recognized.

Missing Link
A senior KPG color consultant was having a conversation recently with a print buyer and his printer. The printer, who had just installed an Approval system, spoke enthusiastically about it, focusing particularly on how much the pressmen liked proofs from the system because they could match them so easily on press.

The print buyer, on the other hand--the printer's customer--was somewhat more guarded in his evaluation. Yes, he said, it predicts well. Now that you're using the Approval, once we sign off on a proof, we feel very comfortable that you'll match it. The problem is that now we're having to go through two or three rounds of proofs to get one that matches the original.

It turned out that while the new Approval had been carefully calibrated to the press, the scanner had not. The image capture part of the equation was missing. Once the scanner was calibrated to the press, the entire original-scan-proof-print was tightened significantly, and the entire system became more efficient. There were fewer rounds of proofs overall, because prepress was able to get very close on the first scan. Then they were able to have very efficient press runs, because the press was very easily able to match the Approval.

The point of this story is that the different components of a digital color production system cannot operate in isolation from one another. This is important to remember when evaluating and implementing color management tools, which are, as noted, ways of automating the process we have been describing. Color management tools must be implemented in a manner that will not allow the components of a color reproduction system to operate in isolation. Users must not assume that the color management tools have somehow "taken care of" the need for calibration and a systemic approach. They can greatly assist skilled, careful operation, but they cannot replace it.

ICC Profiling
Most color management being done today is based on specifications defined by the International Color Consortium (ICC). The ICC approach begins by creating a profile of the color behavior of every input device (digital camera, scanner) and output device (color monitor, proofer, press) in a system. (The system can be as small as one printer's prepress, proofer, and press, as described above, and, in theory, at least, as large as a multi-plant publication printing operation with many customer or agency input sources, many presses, and a variety of proofing systems. It could eventually, in the eyes of some theorists, be the entire graphic communications industry.)

The color characteristics of the various devices are translated from their own color spaces (CMYK for the presses and proofers, various approaches to RGB for the cameras and scanners and monitors) into a common tristimulus (three-color; there are several models in use) color space with a gamut wide enough to encompass the entire tonal range of every device in the system. Once this has been accomplished, an incoming image is tagged with information that relates it to that common reference color space. When it is time for the image to be output, a set of calculations based on lookup tables or polynomial equations is applied, which translates the common-color-space value for each pixel of the image into its equivalent value in the color space of whatever output device is being used.

In theory, this enables a workflow in which any color image can be handled as device-independent, and can be rendered successfully by any ICC-compliant output device. A database of color images could exist in a state of unlimited potential, expressed in a device-independent color space, and at the last minute be realized as magazine ads or billboards or web page illustrations or flexible packages or whatever might be required.

This is the broad-scope, high-automation view of color management referred to under "Areas of Controversy" above. It is much too early to say that an omni-purpose, ICC-profile-based color workflow could never be a practical reality in the printing industry--many of the day-to-day tools of the industry are applications of technologies that were dismissed as nonsense not too many years ago. However, it is safe to say that ICC profiling, useful as it unquestionably is, is an adjunct, not a replacement. It can at best get a given process function 80% to 90% of the way to where it needs to be. Limitations in profiling software, operator error (making an ICC profile is not a simple matter) and "noise" (random error caused by other variables) combine to create an irreducible margin of error.

Beyond that point, live oversight is required. The main difficulty with color management today is in automatically creating an ICC profile. Once you've created a profile, you can color tune it on a specific device to be extremely repeatable--but getting from the 90% or 85% level to a higher degree of color match on a particular device is the part that's difficult to automate. Also, a profile created on one machine may not be indicative of what needs to be done on every similar machine. On an ink-jet device, for example, every time a nozzle or a head is changed, it is necessary to re-characterize the device to get back to the optimum color match.

The Black Channel
Another area of concern is that the realities of four-color CMYK process printing create some challenges that, so far, are simply not well dealt with by the ICC profiling process.

For one thing, high-quality color separation algorithms replace tricolor (CMY) gray values largely or entirely with an appropriate density of black (K). There are two related but somewhat different processes by which this is accomplished. Undercolor removal (UCR) is intended to reduce the total amount of overprint coverage. If all four colors are involved in the dark areas of an image, the maximum total possible coverage would be 400%; removal of CMY (and, generally, increasing the percentage of K) lowers the total coverage while maintaining the darkness of those portions of the image. Gray component replacement (GCR) takes the neutral components of an image and, instead of printing them with the complementary colors of the dominant color, prints them with a neutral black. This is intended to reduce the effect of color shifts by making any shift in color less visible as a hue change. The application of UCR and GCR is an important component in producing high-quality printing, and printers tend to be fairly painstaking in the way they go about it.

ICC profiles, however, are created by colorimetric measurement. As far as a colorimeter is concerned, gray is gray. It will be expressed, in the common reference color space, in terms of three values (RGB, L*a*b, etc.); the profile, in effect, is blind to GCR or UCR. A lot of printers have come up with workaround solutions to this problem, but the more an image is accompanied in its journey through the workflow by not only an input profile but specific output data, the farther that workflow deviates from the theoretical ideal of device-independent color. To eliminate this issue, many companies are moving to use of RGB images in their workflow until the final output medium is selected.

Other Printers' Issues
In addition to the black channel challenge, printers have other specific concerns that impact the application of color management. One--a source of friction between printers and graphic designers for many years--has to do with gamut. The (CMYK) color reproduction process for printing has an inherently narrower color gamut than color film, digital cameras, and color monitors. This means that designers and photographers can see--and design with and specify--colors that cannot be reproduced on press. Here, ICC profiling can be of enormous help. It is possible (and should be done as a matter of routine) to apply an ICC press output profile to a color monitor, so that a designer is seeing and working with only printable colors. This type of initial gamut restriction, particularly if the designer's monitor is also (regularly) calibrated, it can create a much tighter link between the initial design and the contract proof. This helps avoid over-promising, so that expectations of final color are realistic.

Reference Printing Conditions
Many printers today are moving toward a concept called reference printing conditions. This idea is based on the observation that the color gamut of a printing process such as gravure, lithography, etc., is determined primarily by the densities that can be achieved using that printing process. Once the densities of that process are known, the image data can be adjusted to create an optimal reproduction given that specific color gamut.

Three reference printing conditions in fairly wide use are keyed to printing standards: Specifications for Web Offset Publications (SWOP), Specifications for Newsprint Advertising Production (SNAP), and General Requirements for Applications in Offset Lithography (GRAcOL). These three sets of guidelines are aimed at the magazine publishing, newspaper/nonheatset, and sheetfed commercial printing segments, respectively. These reference printing conditions have different densities. Because of that, their gamuts are different, and the way one achieves a color match--what the data has to be transformed to--is also different.

Use of these print reference conditions raises an interesting proofing issue. Because the print reference conditions have differing color gamuts, they have different density requirements. KPG's Approval is the only laser thermal device that has adjustable density. This will allow the user to accommodate those multiple print reference conditions. Other devices adjust their dot gain to try to emulate what's happening with the density, an approach that simply doesn't work.

The potential for this kind of process-to-process color match can be seen in publications like Reader's Digest and National Geographic, in which different signatures are printed not only on different presses, but with different processes. These publications mix offset- and gravure-printed signatures while maintaining a consistent appearance throughout the magazine. Perhaps the most virtuosic display of color control and consistency in put on by TV Guide, which wraps an offset-printed cover around a gravure-printed body. The publication routinely prints cover two/page one ad spreads that maintain excellent color consistency the spread while mixing not only two processes but two stocks, a coated litho cover stock and a supercalendered gravure body stock.

Plant-to-plant consistency
Another color management issue of major concern to printers is plant-to-plant color matching. A multi-plant printer may for any number of reasons want to divide a job among several facilities; even more commonly, the printer may have several plants doing different jobs for the same customer. For this to work, obviously, work from press A in plant X has to match, within fairly tight tolerances, work from press B in plant Y and press C in plant Z.

Here the situation is different from the tight press-proof-prepress link described above. These presses may differ considerably in optimum printing behavior--the so-called "sweet spot"--and may use different ink sets and other raw materials. An image generated for advertising in an automobile brochure in Detroit, for example, might have to be printed on other continents. In Europe the tendency is to print from positive plates (versus negative plates used in the US), and just that fact alone requires a very different color calibration and color transform. Not only are the plating systems different, but the inks are different. These types of variables lead to the need for color transforms that enable a file to behave as if it were native to the printing system that's being used to print the image.

True Benefits of Color Management
For the graphic arts industry, color management offers the possibility of a tighter and more accurate color production system, fewer proofing cycles, and better overall communication of color. The key to a successful implementation is to work back from the output device--the press--to the proofing and prepress functions. ICC profiles and automation tools are extremely helpful in controlling color better, faster, and across a wide range of devices.

For one thing, they can help prevent errors born of excessive calculation. In a color management system there are input profiles, device-independent color space, and output profiles. At one point, the input profile and the output profile need to be married. There needs to be a calculation from the input profile to the output profile; you want to marry those and calculate a composite profile, but you only want to do it once. A double calculation of the image will create rounding errors, which can cause artifacts such as banding. These are basically shades of gray that are calculated in one transform to come together and rounded to the same number; they can't be pulled apart later in another, subsequent calculation. The way the ICC architecture has been designed, you don't actually calculate the data. You carry the transform with you. This allows you to avoid the number of calculations that cause these quantization errors. You need only calculate the data once, which is one of the benefits of true color management within the ICC profiling infrastructure.

A perfect translation system that enables the automatic, effortless, generation of totally reliable color transforms does not seem likely to appear in the immediate future; however, much worthwhile progress is being made, and will continue to be. Powerful very effective color management is becoming a reality. Kodak Polychrome Graphics (KPG) has access to a significant patent portfolio from both parent corporations. We offer many products and services to characterize printing systems, color match proofing systems, and to establish coordinated process color systems from image capture through output on any number of media. KPG also offers an extremely powerful "tuning" software for adjusting ICC profiles to obtain that final degree of accuracy. The company is aware of the various hurdles to the implementation of color management, and has built tools that assist customers in overcoming them.

Finally, printers need to remember that the very nature of digital color workflows requires them to have an informed awareness of the range of things that can happen when one part of a system communicates color data to another part. It is not effective for the various components of a color production system to operate in isolation from one another. (It may be, in fact, that the organizational and cultural changes mandated by digital technology will ultimately prove to be more challenging than the technology itself.) In a digital workflow, you must make more explicit choices about color. You must understand the choices you make, and you must communicate them to everyone affected by them. The better you do that--and the better the tools you use to help you do it--the more accurate, efficient, and productive your color operation will be.

Download pdf