The average person has little or no experience with color work nor the time or desire to be trained in its use. This is one of the big color computing problems that our profession has to face: because even though the computer brings to the average individual powerful color capabilities, it is unlikely that such a person will know how to use or will learn how to use color wisely. One of the original charters of Canon Information Systems was to develop a better way to do computer based color work. In our attempt to fulfill this charter we started by doing some basic research. Based on this research we developed some concepts, performed a number of experiments, and from the gathered data developed the Canon Color Advisor. The Color Advisor was included in every Canon Bubble Jet Printer Driver for over two years.Our fundamental research explored how people in various fields using non-computerized methods select and use colors in their work. From this study we isolated the idea that color selection and use was based on a projects' context, its' intended audience, and its' expected outcome. In other words, colors are not chosen and used in some random, ambiguous fashion but with a method and a goal. The method was to analyze the nature and goals of the current project and from that develop a small palette of colors.From these empirical results of our research we developed a two-stage experiment geared at gathering data about potential relationships between color and descriptive terminology. The results of this study identified a list of 31 terms that were at once descriptive of a document project and colorful. Then in order to associate each of those 31 terms to a number of defined colors we ran a trial experiment on a group of subjects. In it a test subject rank a defined set of 75 colors according to each of our 31 terms.By averaging the results from these trial tests we created a set of “color relevancy curves” for each term. Then the 31 terms were grouped into four categories: audience, occasion, style and setting. The idea was that by picking one term per category a user could define the nature and goals of their project and from that point, using the color relevancy curves, our software could create a small palette of colors for them. From our original research we had determined a useful palette to be made up of one dominant color, two coexisting colors, two each supporting colors for each of the dominant and coexisting colors, three highlight colors and black.We formulated a simple algorithm: from the curve of one descriptive term the dominant color could be chosen as the color with the highest relevancy value. From that dominant color two coexisting colors were determined. Supporting colors, two for each of the dominant and coexisting colors were calculated from the color that they were to support—basically one lower in value and one higher in value and lower in chroma than the color that they were supporting. Highlight colors were determined using the dominant color as a starting point. This basic algorithm was expanded so that if more than one term was selected then the relevancy curves were averaged together and the resulting composite curve was used to determine the dominant, coexisting, supporting and highlight colors.The results of this work were productized as the Canon Color Advisor. Unfortunately it was buried into the BJ drivers and most users never saw or used it. We believe that that there are further color use technologies that could be developed from our investigations and work. Among them are, Color Agents, Color Use Profiles, color palette resources, system wide color work management and color work management tools.
Larry Lavendel, Tim Kohler, "The Story of a Color Advisor" in Proc. IS&T 6th Color and Imaging Conf., 1998, pp 228 - 232, https://doi.org/10.2352/CIC.1998.6.1.art00049