Le 15.6.2003 3:30, «?PhotoRoy6@aol.com?» <PhotoRoy6@aol.com> a écrit?: > Color space ? Isn't that a term from the 1950's (g) > > Simply put a color space is the mathematical location of the three > subtractive primaries on a graph with white location being the third > dimension. > > >From what I've just read (in Rene Bouillot "Cours de photographie, theorie et pratique"), the most common standardized way of measuring color spaces was established by the CIE (Commission internationale de l'eclairage = International lighting Commission) in 1937. It is also named the XYZ color system, or the CIE color model. We know that by mixing three primaries colors (indigo, green and red), we can reproduce almost all colors (almost, because some of the pure spectral colors can never be synthesized). In order to allow the graphical representation of colors the CIE has chosen three arbitrary primaries (blue, green and purple) which serve as coordinates. These three primaries colors are also named coefficients of chromatic distribution and their maximum is located at 4500 angstroms on the Z axe (Blue), at 5641 angstroms on the Y axe (Green); there are two values for the X axe (since we are in a three dimensional system) the first is for purple at 6000 Angstroms and the second is at 4500 Angstrom for indigo. So the CIE system defines colors as the the proportion of the three primaries colors allowing the reproduction of a sampled color. A complete graphic representation would need three dimensions, but usually we only take the quality of colors in account. This is possible because the CIE has defined standardized color temperatures; when you work under controlled situations for the color temperature (for instance 5000K), you suppose that X+Y+Z is always equal to one and thus, for a given TC (color temperature), only the proportion of primary colors will change, thus limiting the diagram to two dimensions. This is the chromaticity diagram which takes the form of a left leaning triangle whose height is usually longer than its base. These chromaticity diagrams represent ideal color spaces (like SRGB, Adobe RGB, Color Match, Apple RGB etc.), as well as specific color spaces showing for instance the color gamut of emulsions, or the color gamut which you will be able to get from your ink and paper combinations. (this is a free translation from Rene Bouillot's chapter : "the CIE color system" pp 147-148) All the color calibration problems are caused by the fact that each medium, each tool has its specific chromaticity diagram, its own color profile : your printer is able to print colors your display isn't able to show you and vice versa. Add the cameras and scanners and the complexity grows. This is why color reference spaces have been created : in this ideal world (alas not the real world..) if you calibrate each of your tool for the same color space (SRGB or Adobe RGB, or ..) then you'll get a more or less good match between the input source and the destination. Between what you saw an what you print. Both the RVB and CYMK mode are expressed in the CIE or XY color model, because they make use of the same coordinates system. There are other ways of measuring colors however; one often named is the LAB color model which is used internally by Photoshop to make computations; it's a model including three color channels, like the RVB model, but the first is the the Lightness component (L channel), while the A channel gives the proportion of Red and Green and the B channel the proportion of Blue and Yellow. To see better the differences between these two color models, open a color picture in Photoshop; click on the preferences command and open the display and cursor panel : be sure that the under the display option, the "color channel in color" box is checked (which usually, I prefer to keep unchecked). Then open the channel palette and look at the individual color channels, making them active or inactive alternatively : you will see each of the Red, Green and Blue channel. Now convert your picture to the LAB mode (Image/Mode/Lab Color); if you have the L channel active only, you get a B&W picture, adds the A or B channels separately.. You'll see what it means to be in another color model; when all the channels are active, the picture is the same : one can convert the pictures from one system to the other. The Lab color model is often preferred as being more precise in computation, thus preserving a better quality of the picture. -- Christiane
