An Introduction to Color Space and Color Calibration

The following text and diagrams were used for our July 2005 seminar on color management, calibration, and theory. Gamut charts used in this document were gathered from various sources on the web. Additional links and illustrations have been added to supplement the material.

Addition and Subtraction

The basics: light is additive. Red, plus green, plus blue equals white. Inks are subtractive: Cyan, plus magenta, plus yellow equals nearly black. Photographic media are based on the three primary colors of light: red, green, blue. Printed media are generally based on a four-color process using cyan, magenta, yellow, and black inks.

The Standard Observer

Physiological studies over the years have shown that the human eye contains three different types of receptors, that respond to short-, mid- and long-length light waves. These responses correspond to the colors we call blue, green, and red.

The graph below reflects the 1931 CIE Standard Observer for a 2-degree field of vision. What does it mean? The lines on the graph show the amount of each primary light color (blue, green, red) necessary for the eye to perceive a pure representation of those colors.

The CIE “Lab” Color Space

Using these response curves, the CIE (or Commission Internationale de l’Eclairage – International Commission on Light) developed the CIE “Lab” space (actually, L*a*b). This is a model that represents the color reponse of the human eye at various hues, saturations, and intensities. Typically, this three-dimensional model is shown as a 2-D chart showing hue and saturation at maximum intensity.

The standard color space, as determined by the “standard observer” model.

Color Gamut: What, and Why?

Gamut is a term used to describe the colorimetric capabilities of any given reproduction or display method. It describes within the context of the CIE Lab space the range of colors that can be accurately represented.

Why is this important, and how do we get into trouble when moving between color spaces? Quite simply, no currently available reproduction method completely encompasses the entire color response of human vision. And every reproduction method expresses a slightly different segment of the Lab space.

For instance, here is the color gamut graph of a typical CRT monitor:

The points of the triangle are the responses of the red, green, and blue phosphors on the screen. Within those vertices lie the colors that the monitor can accurately represent. You can see how limited this system is, especially in the areas of deep blues and reds.

In the next chart, A is the Lab space, B is a representation of a typical RGB device such as a monitor, and C is the representation of a CMYK device such as an inkjet printer.

You can see the difficulty the user experiences here. There are colors that the printer cannot reproduce that the monitor can, and those that the printer can generate that cannot be displayed accurately on the monitor.

For those working strictly digitally, there are additional factors to be considered. There are two predominant color spaces for digital photographic work. The first is sRGB, the second AdobeRGB (1998).

sRGB was derived by Hewlett Packard and Microsoft, and has been widely adopted. It is intended to accurately reflect the color response profile of the average consumer computer monitor. While this is sufficient for applications where the photograph is intended for a web site or other computer display, it is generally considered to be too limited for high-quality reproduction and proofing.

Adobe Systems proposes their own color space: AdobeRGB, which was intended to represent the colorimetric capability of a CMYK device, as displayed on an RGB monitor. It covers a wider gamut than sRGB, partcularly in the green and cyan portions. In the graph below, the white outline is the sRGB space, and the black outline is AdobeRGB.

What space is appropriate depends on the intended use of the photograph. If you’re shooting for online publication, sRGB is sufficient and will probably yield the best results. If you’re shooting for prints, it’s probably best to set your camera to use AdobeRGB (if it has that capability), and then make a copy of the image converted to sRGB if you later decide to use it for online display.

Lost in Translation

Every device that a photographer uses to capture or reproduce an image responds to color differently. Color film reproduces a wider range of colors than digital RGB sensors or film scanners can capture, particularly at low brightness levels. RGB sensors and scanners can usually capture more color data than can be displayed on a monitor, and a printer or photo paper can reproduce a smaller color space still. In addition, converting from film, digital sensor, or scanned image (which is represented by RGB values) to a printed proof (represented by quantities of CMYK inks), requires some translation.

The difference in gamut between the RGB and CMYK methods, the color purity of the inks used, the response properties of the monitor, films, cameras, and sensors used to capture, edit, and display lead to the frustrating exercise of trying to get a print that accurately represents the original photo.

Enter Calibration

The solution to this problem was put forth by the International Color Consortium, or ICC. This body of imaging industry vendors developed a standard method by which color values could be accurately translated from one device to another, and from RGB devices to CMYK printers. This system is known as a Color Management System, or CMS.

For a CMS to work, one must know how the different devices being used respond to color input. Typically, the manufacturer of a piece of equipment will generate a standard “profile” for their device that represents how it will perform. These profiles are averages, though and, especially in the case of printers, are so specific to a certain environment to be nearly useless. For a photographer to produce truly accurate color images, he or she must measure the color response of their own equipment, most importantly the monitor on which the photos are to be edited, and the output device to which they are sent.

More Printer Problems

In addition to the inaccuracy of printer driver profiles as provided by the manufacturer, there are several other elements that affect the ability of a photographer to generate accurate print images.

The first is the fragility of printer profiles. The color response of a printer is first dependent on the inks used. The stock profiles assume that you are using the OEM’s inks, and even then, they express an ideal inkset. Ink color purity changes from lot to lot, so over time your profile will drift away as individual color ink cartridges are switched out. This generates a wide range of inconsistencies as different batch numbers are combined.

In addition, the paper that you use will influence the final reproduction. No paper is pure white, and some are less so than others. They also absorb inks differently. Most stock profiles also assume that you are using the manufacturer’s brand of photo paper stock. If you purchase another brand, you may find that it causes color shifts (also called “casts”) because of the variations in reflectivity, brightness, and absorbancy.

Finally, there is the problem of metamerism. Metamerism is the phenomenon wherein the human eye perceives colors differently in different light. Light sources have their own color biases, and a print viewed in flourescent light looks different under incandescent and sunlight. A proper profiler will calibrate your system to a known color temperature – typically 6500K.

How to Measure

Colorimeters

The easiest method for calibrating a CRT or LCD monitor is the use of a “tristimulus colorimeter”. This is a device that uses photo sensors and filters to accurately measure the colors of light in the same way as the CIE Standard Observer.

The colorimeter and its accompanying software will measure how your monitor reproduces color, and generate a profile telling your video card and display driver how to adjust their output. When the adjustment profile is loaded into the video card’s color lookup table (LUT), the colors displayed on the screen should then be in proper proportion to each other according to the Standard Observer model.

Popular colorimeters are made by Pantone/ColorVision, and Gretag-Macbeth. ColorVision, in particular, makes an inexpensive product for home users called ColorPlus.

Photospectrometer

This device measures the exact wavelengths of light being reflected or emitted by an object. The resultant scan can then be modeled against the Standard Observer and a colorimetric profile can be generated. These devices are typically extremely expensive and are outside the realm of most imaging enthusiasts and professionals. However, generic device profiles provided by equipment vendors are generally created by averaging photospectrometer readings from manufacturing samples.

SpectroCAL is a photospectrometer product from Cambridge Research Systems, and will probably cost you more than your car did.

Patch Reader

An inexpensive device for calibrating printers, the patch reader is essentially a dedicated scanner. Typically, the patch reader is calibrated by running a standard color target through it. From these standard colors, the computer knows how the reader responds to known colors. Then, a color target generated by your printer is read, and the software compares the two. From these two readings, the software calculates how accurately your printer reproduces known colors, and generates a profile that adjusts your printer driver’s output to meet a standardized CMYK gamut.

There are some packages that utilize the user’s existing flatbed scanner as the patch reader. These are generally less accurate, as there is no guarantee that the user’s scanner will produce a scan of sufficient quality.

Profile Prism is a popular scanner-based calibration system, provided you have a good flatbed. ColorVision offers the PrintFIX patch reader.

Spectrocolorimeter

More expensive than a patch reader, the spectrocolorimeter is a pre-calibrated device that uses a light source with known properties to illuminate color patches produced by your printer. It also contains photosensors that measure the reflected light, and can very accurately generate a difference profile. The spectrocolorimeter is typically used when absolute accuracy is necessary, and typically costs four or five times the price of a patch reader.

Both ColorVision and Gretag-Macbeth offer spectrocolorimeters in their SpectroPRO and Eye-One product lines, respectively.

The Magic of CMS

Once an ICC profile has been generated for each device, the CMS can then translate between them. Typically, the source color is translated to its value in the CIE Lab space, and then using the profile for the display and printer, can then be mapped to its nearest equivalent for the appropriate device.

What does this mean in the end? The CMS ensures the most consistent rendering of the color values across devices. However, it cannot guarantee that a color on one device will appear exactly the same on another – that’s impossible given the gamut differences.

Why This Is Still Important for Black and White Images

When reproducing monochrome (B&W) images using a CMYK device, rarely is only black ink utilized. Merely using black ink in various amounts will not produce a smooth gradient of grey tones. Thus, many printer manufacturers use some proportion of color inks to generate grey halftones. And since papers respond differently to inks, you may find your “black and white” prints are really “Mostly black and sorta white with a little bit of cyan or magenta in the greys.”

This is where printer profiling comes in. If you are not using dedicated monochromatic inksets with varying concentrations of black pigment, you need to ensure a uniform color reproduction that generates neutral greys, or as close to them as your system will allow. The way to do this is through profiling. In addition, a good profile generator should allow you to adjust the CMY biases of your neutral tones.

Converting Color to Monochrome

With all of this talk about properly rendering color images, it should be said that color perception must be discussed even when considering black and white photos. This is especially important when converting color digital images into monochrome.

Typically, if your digital camera includes a “black and white” mode, it captures a color image and then removes the color data, leaving only luminance values. This is a straight greyscale conversion, and it is not an accurate representation of how black and white films react to light. Nor does it accurately represent the way that the eye proportinately responds to red, green, and blue. Straight conversions of this type rarely produce acceptable monochromatic prints.

The reason is that even panchromatic black and white films (those that respond to a wide range of color) have a characteristic response curve, which tends more to red and green and less to blue. Just removing the color data (desaturating) assumes that each of the RGB components contributes equally to the final image, and the final result frequently looks “muddy” and flat. The correct way to convert to monochrome involves using the Photoshop channel mixer tool to ensure that each of the RGB channels contributes the proper proportion of brightness to the image.

This is the Photoshop 7 channel mixer, with a good starting profile for converting a color photo. Select the “monochrome” check box to tell PS that you want to tell it how each of the red, green, and blue primary colors contribute to the greyscale values. The curve shown above is roughly based on the chromatagraph for Ilford HP5+. A scene processed with these values will retain the tonal relationships of the original image without luminance shifts.

Summary

Understanding how our eyes and our equipment perceive color is important for both color and black and white photography.

Color management systems are not perfect, in that they cannot always provide WYSIWYG translations from one medium to another. However, they do provide consistency. Once calibrated, your system will be able to reproduce the same results from the same sources time and time again. In addition, a profiled system will provide the greatest proofing accuracy which, while not 100%, will be much closer to the mark than one that is uncalibrated. A profiled system also allows you to adjust for your print media and produce consistent results across output devices and paper stocks, even when printing monochromatic photos.