Color management for digital photography: a primer

by Robert Harrington

It took me a long time to understand the basic ideas of color management, largely, I think, because the explanations I read and I read a lot of them were so, if you'll pardon the expression, opaque. It is almost as if there was a massive conspiracy among those who understood it to keep the rest of us from joining the club.

There are two basic concepts to understand about color management for digital photography: color spaces and color profiles. A color space is a defined set of colors to be represented. A color profile is a transformation required to reproduce the desired colors by a particular device.

That clear? Maybe not. Let's go into concrete examples.

Color Spaces Color is typically represented as three numbers for red, green, and blue (RGB). In the JPEG specification each is one byte, or 8 bits. Together you have 24 bits, or 24-bit color as it is called, by which 256x256x256 = 16,777,216 colors can be represented.

With 8 bits, you can count from zero to 255. Some colors so represented are (0, 0, 0), which is black; (255,255, 255), which is white; and (255, 0, 0), which is "pure red." I put that in quotes because the meaning of "pure red" is not self-evident; in fact, it is arbitrary, and the same is true for all other colors.

To avoid the chaos of everyone having their own definitions of the meaning of the (r,g,b) numbers, some standards have been developed. These standards, called color spaces, define exactly what physical colors the (r,g,b) numbers are to mean.

Thus, when pure red, or (255, 0, 0), for example, in a given color space is displayed, such as on a monitor, it should be possible to measure it with a color-measuring device (colorimeter) and get a definite predetermined result. Real- world monitors are imperfect, of course, and you will normally get a different result (more about that later), but the point is that we know what result the colorimeter should be giving us.

In general, a different color space will have a different definition for pure red (and all other colors), and we will look for a different result from the colorimeter.

You may ask at this point why I said color spaces in the plural. Unfortunately, many different standards, or color spaces, have been developed in response to different needs and applications. They vary mainly in the range, or intensity, of colors they can represent. A color space with a wider range of colors is said to have a wider "gamut," and one with narrower range is said to have a narrower "gamut."

The term "space" is used because the three colors are akin to three physical dimensions (x, y, z). Color spaces are often depicted as a three-dimensional volume in the form of a two- dimensional graph showing color gradations. To my mind this form of illustration serves only to confuse the issue.

A comparatively recent attempt to standardize on one color space was undertaken, resulting in the color space known as sRGB (or standard RGB). It has been widely adopted, and it is the standard for the Internet. It is not the end of the story, however, because as a be-all and end-all color definition, it was designed with compromises to accommodate commonly-used devices, such as computer monitors. Consequently, sRGB does not satisfy those who wish to obtain the dazzling colors that some devices (mainly printers) are capable of. Consequently, other color spaces are still used, a prominent one being Adobe 1998, which has a wider gamut than sRGB.

Most digital cameras produce files intended to be interpreted in the sRGB color space, and this fact is normally designated in the "EXIF" data that is included in the files. In some cases, such as the Nikon D100, it is possible to choose between sRGB and Adobe 1998 capture.

It is possible to convert a given image file from one color space to another. The relative benefit of doing so is a matter of debate.

Color Profiles It is fine and good to have a color point in your computer's memory that, according to the color space you're using, is supposed to represent a specific actual color, but how do you view it? What you get on your monitor or printer, unless they're calibrated, varies widely.

OK, how do you calibrate them? Good question! Consider the monitor.

There are a number of products on the market that allow you to calibrate your monitor. Each uses a colorimeter called a "spyder," a device you place on your screen and connect to your computer via a (probably USB) cable. Accompanying software then generates a range of colors and measures through the spyder what color your monitor actually produces. By noting the difference between what your monitor produces and what it's supposed to be, specified by the color space, the software is able to generate a table of corrections. This table of corrections is the color profile of the monitor.

Software that is "color smart" uses the monitor profile to apply appropriate corrections when displaying colors, causing the monitor to display colors that closely approximate what the color space says they're supposed to be.

I'm not going into calibration of printers for the very good reason that I have no experience with it. Suffice to say that printer profiles vary with manufacturer, model, paper, ink, operator settings, and who-knows-what else. The point is that if you want color consistency, calibration via a profile is necessary.

A final note: Some speak of "color accuracy." In my opinion, that's a will-o'-the-wisp. There are just too many variables, beginning with the light at the time of capture. Whether we admit it or not, photographers actually pursue what we consider pleasing colors, and furthermore that's about the best that can be done. If it were possible to produce a photograph with actual colors, we probably wouldn't like it and couldn't resist improving on it. Does that make color management pointless? No, color management is not about accuracy so much as it is about control and reproducibility.

Comments? Email me at robthrr@cox.net.