|Man knows so much and does so little. - Inventor Buckminster Fuller|
A color space is an arbitrary agreed upon way to define color. There is any number of ways to visualize color. Each has its different advantages and disadvantages.
Colors cannot be defined by only two values. They require at least three. That is why the term "color space" is used instead of some two dimensional term like color chart. Color cannot be accurately represented on two dimensions just like a three dimensional globe cannot be accurately represented on a two dimensional map. Of course, it is done all the time, but the map is never completely accurate.
Our eyes perceive color by using color receptors called cones. There are three types of cones in the human eye. Each receives short, medium or long wavelengths, which correspond to blue, green and red, respectively (the primary colors of light).
The three values of a color represent a measurement of how much each cone is being stimulated by the color the eye sees. Note that perception of color varies among people, making this a non-exact science.
There are many different models to define color space. Some of the more common color spaces are:
The Munsell Color Tree
The Munsell color tree is one of the oldest color systems we'll be discussing. It was popularized in the early 1900's. Munsell was reportedly the first color researcher to systematically work with color in a well defined color space that combined hue (or color), value (or brightness), and chroma (or saturation).
It is a great color system to study because it can help create pleasing color combinations that are well balanced.
HSV: Hue Saturation and Value
The HSV color space is one of the easier ones to understand and use. It breaks the color space into hue (or color), saturation, and value (or lightness).
HSL: Hue Saturation and Lightness
The HSL color space is similar to HSV, with the exception that it assumes colors approaching white have a limited saturation and higher color value. Those seem like valid issues, but they complicate the model to a degree and make it less attractive, while not really providing additional useful benefits.
RGB: Red Green and Blue
The RGB color cube. The values of red, green and blue are balanced so that when all values are at 100%, white results. This is the color system of choice for computer displays, as it is easy to use and understand.
The three monochromatic (consisting of a single wavelength) primary colors are defined at standardized wavelengths:
Red = 700 nm
Green = 546.1 nm
Blue = 435.8 nm
CIE XYZ: The Tristimulus Values
The CIE XYZ color space is the gold standard for defining a color. The reason is that it defines the whole human gamut of possible colors, and legal XYZ values will even define colors that we cannot create and observe. Think of it as a superset color space that is especially useful over time, since it doesn't need to change as each new display screen technology becomes available. In fact, it's great for defining the limitations of the various display technology.
To help give you a sense of what's happening, we show the rgb cube of all possible computer colors as just a sub-section of the XYZ color space. This is also known as the 1931 XYZ color space, since future modifications have been made to related color spaces that have other useful properities.
The color bands below represent the limits of human vision. Notice that those limits are not at the defined points, X, Y or Z, meaning it is actually impossible to see those points, or in fact, a lot of the defined color space.
Other Properties of the XYZ color space
Almost every color system has published mathematical functions that derive from the CIE XYZ color space. The X, Y and Z values are roughly related to red, green and blue respectively, and are derived from the Long (L), Medium (M) and Short (S) cone cell excitations of the Standard Observer. The Y value has been specifically set to also represent the Lightness of the color.
Note, although we've drawn this chart over a CIE chart, there is not a direct relationship. In fact, the points X, Y, and Z are usually not drawn in a cube like we have done here. The accepted shape of the space is really quite different an difficult to visualize.
CIE xyY: The Chromaticity Chart
The CIE gamut chart extended to a 3rd dimension. x and y are mathematically related to the CIE XYZ values in such a way that they are always positive for easy calculations. The equal energy white point is represented at x = 1/3 and y = 1/3.
The chromaticity gamut chart is useful because the top curved edge represents pure colors and all of the colors that can be created by mixing two colors will be found along a straight line between those two colors.
The CIELAB, or CIE L*a*b* which is an improvement on the CIE Lab, is a color space that has been specifically designed to accurately map color perception. It is commonly used in color manipulation software since it so accurately stores and aids in processing color relationships, like contrast.
The CIE L*u*v*, or CIELUV as it is know, is another attempt to create a color space that accurately maps color perception. Both CIELAB and CIELUV were proposed and accepted by the International Commission on Illumination when there was support for both systems.
The three parameters that define the space are Luminance, red to green, and blue to yellow.
CIE LCh: Lightness Chromaticity and Hue
LCh, can be thought of as the cylindrical version of CIELUV.
YCrCb: The color space of RGB compression
YCrCb, or its close relatives, Y'UV, YUV, Y'CrCb, and Y'PrPr are designed to be efficient at encoding RGB values so they consume less space while retaining the full perceptual value. Have you ever looked at the size of a bmp or bitmap file in comparison to a jpg file of the same image? The jpg file, which utilizes tye YCrCb encoding, is much smaller. That makes it faster and easier to save or transmit.
Y is the luma, which is different the actual luminance, but very similar. Cr is the component of red to green, and Cb is the component of blue to yellow.
YCrCb is specifically attuned to encoding RGB signals where the source colorants or colors of red, green and blue are known.
Although RGB values are easy to work with, they are somewhat redundant. In other words, the raw RGB values of a picture contain far more information at greater resolution than can be perceived by our eyes. By translating the image to the YCrCb space, only one value needs to be transmitted in detail, the Y or luma, whereas the Cr and Cb values can be compressed without loosing substantial picture quality.
And Many More...
This is only a sampling of the most common color spaces. Additional color spaces include a Color Globe, where the north pole is white, the south pole is black, and the hue varies with longitude. Others include: LAB94, LAB2000.
Color Space Summary
Speed of Light
Additive and Subtractive Colors
CIE 1931 Color Space
Spinning Color Top
Glossary of Color Terms
History of Color Science
Motion After Image
Munsell Color System
Color Optical Illusions
Report A Broken Link Contact Information