Target Audience and Section Objectives

There is a wealth of color management information elsewhere on the web. The intention here is to provide a fresh perspective derived from actual experience and review of material you will find in some of the links given later. Target audience is the newcomer to color management but others may also find something useful.

Some of the topics and definitions presented will be repeated in one form or another elsewhere, as a convenience and to stress their importance.

My  hope is that more photographers will take CM seriously and, thereby, obtain more satisfaction from their hobby.

Overview of Color Spaces

Owing to the psychological foundation of all perception, the appreciation of color is subject to personal factors and uncertainties arising from experience, motivation, likes and dislikes, gender and even culture. What someone describes as pure red might be identified as a shade of orange by someone else. No two people see exactly the same thing. This poses considerable challenges when it comes to designing devices and systems for the purpose of color display where a high degree of consistency is required.

 In 1931 scientists and regulators formed the unpronounceable  Commission internationale de l'éclairage (CIE) with the purpose of laying a foundation for standardized ways of dealing with color. They subsequently developed a mathematical model of color incorporating characteristics of the human visual mechanism.  These characteristics were determined from numerous careful experiments (some pre-dating the CIE) conducted using human subjects judged to have normal color vision. The result was an internationally recognized color model or color space defining the limits of normal human color perception in terms of actual wavelengths of light and assigning numerical values (or ways of calculating them) to all colors within this space. The color model itself, prepared using standardized inks, is available in various printed formats from many sources. These are expensive and not needed for most photographic work. Color charts downloaded and displayed on monitors or printed are obviously unreliable because their appearance is dependent upon the presentation accuracy of those devices. Such charts are useful mainly for comparing the outputs of various units and have limited meaning in absolute terms.

This standard CIE 1931 XYZ color space, incorporating all the colors most of us can see, is the foundation for practically all color rendering systems we have today, including television, motion pictures, color printers and monitors. You can read more here but it isn't necessary for understanding how color management works. In photography, the standard color space is called the Profile Connection Space (PCS) which is adapted from the CIE model. You will learn more about how this is used later.

The term "color space" dominates any discussion of color management. A color space is a description of the total range or gamut of colors a device can display or an image could contain. In the case of the CIE color space, the device is the human eye, so this is the largest color space of all, when considering visible wavelengths of light. Color spaces are not some arcane technical concept. Rembrandt's palette covered in splotches of paint had a color space every bit as much as an LCD monitor. In Rembrandt's case, the selection of oils defined the gamut of his space and could have been been formalized with modern calibration techniques.

A color space associates perceived colors with their numerical specification. In the computer world, using the additive model of color rendition where all colors are represented as various combinations of pure red, green and blue (RGB) light, we typically represent a color with a trio or triplet of numbers, one for each of the primary colors. For example, with 0/255/0 we mean a fully saturated "green" and nothing else. Just how "green" that triplet is going to look to the eye will depend on the color space we assign. That is, the triplet has no real meaning until we locate it within a color space. Suppose we locate that triplet within the sRGB color space. That space is rather confined in the green direction so, visually, the triplet will be represented by a green with a bit of red mixed in, giving a somewhat yellowish appearance. If we use a larger color space such as aRGB  that same triplet is located farther from the red end of a line drawn from maximum red to maximum green. Accordingly, it will appear more purely green even though the numerical value is the same.

The key concept to remember is that the total range (or gamut) of all colors a device such as a printer, monitor or camera could present or render or which an image could contain is a subset of the Profile Connection Space (PCS) and that through the process of device calibration we can describe that subset mathematically and determine where it fits within the PCS.

The Limitations of Technology and the Need for Color Management

Throughout the history of image preparation, people have been looking for materials that better exploit our abilities to perceive color and which will prove long-lasting. The Great Masters of art were always seeking out such materials with which to populate their palettes (color spaces in their own right) - with varying success. It is no different today with printer manufacturers frequently announcing new and better dyes, pigments and inks. Practically all physical substances deteriorate with time. Ultra-violet light, atmospheric gases, contaminants and humidity induce chemical changes that alter colors. Even so, progress had been good and we now have "paint" that will retain its appearance for many decades while colors are more realistic than ever.

This brings us to the need for color management.  With so many ways to produce color, there is considerable variation amongst image presentation systems in the way colors actually depart from their intended appearance. Depending on technologies chosen and even the biases of product designers there is no guarantee that straight-out-of-the-box a Viewsonic monitor (for example) will present images with the same color balance, tonality and shades as an NEC or an Apple, despite built-in presets promising otherwise. In the days of Rembrandt an image was unique. If you wanted to see his latest painting, you went to a gallery. Now, with the Internet, images come to the viewer as numerical facsimiles of the originals and must be rendered into corresponding colors. It is as if, in the past, you had chosen not to go to the gallery to see the latest Rembrandt but hired someone to come to your home to paint you a copy from memory or a written description. Depending on whom you chose, the result could range from very good to terrible. Similarly, in the digital world, the photographer seriously risks losing control of the final appearance of his work. It could be rendered very differently from what he intended, depending on the equipment in use and its adjustments at the other end.

Color management is the process by means of which a digitally represented image is rendered as nearly as possible in the way it was intended to appear in its original color space, no matter what display technology is used.

The First Practical Solution - sRGB

In 1996 Microsoft and Hewlett Packard developed a color standard especially congenial to the characteristics of CRT monitors (and hence the Internet). This was sRGB, which defined a color space derived from CRT monitor capabilities. In a way, sRGB simply formalized what was already something of a de facto standard. sRGB proved a huge success because product designers could readily build the standard right into their devices. In fact, they were already making displays that came close. Images created according to this standard could then be exchanged with reasonable assurance they would be displayed as intended without any intermediate software processing required. In effect, color management was built-in. If the image was created in that space and if it was displayed by a device honoring the standard (which was most of them, by default), it would be rendered much as intended. Life was simple. To this day, sRGB is still very much the standard for exchange of images over the Internet when monitor viewing is intended.

Emerging Standards

In 1998 Adobe developed the aRGB standard incorporating a larger color space than sRGB. This standard recognized the improving capabilities of printer technologies. aRGB is not a huge improvement over sRGB but it is better and has been gaining acceptance - especially for printed images and premium monitors. Its superiority is especially noticeable in greens and to some extent reds. There are many standard color spaces out there now, all subsets of the master CIE color space or PCS, occupying various niches within the color production world. Display technologies keep improving which means that the very real limitations of sRGB grow increasingly noticeable. sRGB remains dominant, however, because it is something of a lowest common denominator, able to be fully rendered by practically any display and compatible with most older equipment and software.

The upshot of the situation was that chaos threatened anew as color rendering technologies advanced, going beyond what sRGB could deliver.

The International Color Consortium, Color Management, Device Profiles and Calibration

In 1993 the ICC was formed by Adobe, Agfa, Apple, Kodak, Microsoft, Silicon Graphics, Sun Microsystems, and Taligent. Its purpose was to establish a standardized environment of protocols and algorithms for creating and displaying images that would minimize departures from intended appearance no matter what display technology was used. This involved the idea of the Device Profile.  Through the process of calibration an image rendering device is characterized and adjusted to achieve the following:

1. Determine the color gamut of the device (its native color space and where this fits within the PCS).
2. Measure parameters such as Gamma(primarily mid-tone contrast) and color temperature of the device in its native, precalibrated mode of operation.
3. Finally, calculate corrections, later stored in the device profile, enabling software to correct the the performance of the device so that characteristics determined in Step 2 above conform to the accepted standard (Gamma of 2.2 and color temperature of 6500K, typically) There may be adjustments for reducing irregularities in the progression of brightness from one tone to the next.

What the calibration process ends up with is:

1. A descrption of where the device's native color gamut resides within the PCS.
2. A set of curves and other corrections enabling rendering software to modify device output in conformity with standard expectations. For example, if the device is designed to be an sRGB device then, post calibration, it should more accurately demonstrate compliance than before.

This information then comprises the Device Profile which is directly used by display drivers and the Color Management Module (CMM) when transporting colors from one space to another. The Device Profile is usually embedded in the image file.

In a color managed system, a Color Management Module (CMM) translates or converts an image prepared in one color space (the source color space) to another (the target color space) with minimal departure from intended appearance. It does this by mapping the source color space (probably a color space embedded in an image and obtained from the creating device) into the Profile Connection Space (The Mother of all Color Spaces), then taking that subset color space and mapping it into the target color space. The target could be a device, in which case its color space is usually read from its device profile or - it could be another image file. It doesn't matter. Since the source and target color spaces may not be identical, the CMM has to make some decisions concerning what to do if a color from the source does not fit within the color space of the target. This is done with reference to rendering intent. Depending on user preference, these out-of-gamut colors can be force-fitted in some way that makes sense in terms of the rendering intent. For example, photographs of scenery will likely benefit from specifying perceptual intent whereas saturation intent is more appropriate for graphics. To be sure, this is a compromise but the result is better than trusting to luck or having colors simply disappear (clipping).

If you click here you will find an analogy to color management using a fictional monetary exchange system. It may further clarify some aspects of CM. The analogy concludes with a sample color managed process description.

Note that work of the ICC preceded deployment of sRGB. sRGB essentially recognized what was something of an informal or assumed standard already (the CRT dominated environment). sRGB made what we had already been using part of the emerging color managed display environment.