Introduction

With the development of color computer devices ranging from color monitors, color scanners, and color printers, as well as software for computer graphics and publishing, the use of color is abundant. However, there are many methods to select colors, varying in human comprehensibility and accuracy. Some methods are based directly on color models used by hardware such as RGB for monitors and CMYK for printers. Other methods are based on the user's perception of color, such as HSV. Even further, a Crayon color selection method tries to present a discrete palette of colors analogous to a familiar box of crayons [4].

Ideally, color selection methods (or "color pickers") should be accurate in allowing the user to specify the precise color that he or she wants. The method should also be intuitive, and therefore be a fairly quick process.

Apple's Mac OS 8.5 includes several color pickers, presenting many options for color selection, from the hardware-based RGB (Red, Green, Blue) and CMYK (Cyan, Magenta, Yellow, Black), to the user-based HSV (Hue, Saturation, Value), and Crayon pickers.

These pickers are ideal for this experiment because they use direct manipulation, along with dynamically updated color representation, to provide an experienced user with potentially fast and accurate color selection with their preferred picker.

Apple's pickers are also interesting because they range from 1 dimensional (crayon Picker) through 4 dimensional (CMYK) color models, with HSV being 2 dimensional, and RGB being a 3 dimensional color model. Each additional dimension has an additional adjustable attribute, and therefore an additional mouse action.

The Crayon Picker (1 Dimension)

The HSV Picker (2 Dimensions)

The RGB Picker (3 Dimensions)

The CMYK Picker (4 Dimensions)

Table 1 describes the characteristics of different color models used thoughout various industries.

Table 1: Some commonly used color models [4].
Acronym	Intended Use	Axis1	Axis2	Axis3	Used in
RGB	Device specific color specification	Intensity of red gun	Intensity of green gun	Intensity of blue gun	Macintosh, WIndows
HSV	Color Mixing	Hue	Saturation	Value	Macintosh
HLS	Color Mixing	Hue	Lightness	Saturation	Windows custom color dialog
LAB	Measuring color distance in human perception	Lightness	red/green	green/blue
CMYK	Color Printing	Cyan	Magenta	Yellow	Macintosh
YIQ	North American Broadcast Television	Luminance	blue-green/orange	yellow-green/magenta
Opponent	Model of human color vision channels	Achromatic	red/green	yellow/blue
TekHVC	Perceptually uniform color selection on electronic displays	Hue	Value	Chroma	Color selection for Tecktronix
*The opponent "model" is in fact a family of related models derived from similar but slightly different theories of color perception.

There are only a few studies that explore the methods of color selection when it comes to measuring the speed and accuracy of different color models. One such study, "A Human Factors Study of Color Notation Systems for Computer Graphics", was performed by Berk, Brownston, and Kaufman [2]. They compared a color naming language of 627 distinct color names versus specifying numerical coordinates in the RGB and HSV color models. They found that the natural language of colors produced the most accurate results, with the HSV coordinates next, followed by the RGB coordinates. Although this experiment relates to color selection methods, it did not involve testing the speed of the color models.

More importantly there are two other studies that relate very closely to the testing of speed and accuracy in using various color models. Schwarz, Cowan, and Beatty performed "An Experimental Comparison of RGB, YIQ, LAB, HSV and Opponent Color Models"[7]. They compared these five color models and found many significant differences between them. They found the RGB model to be the fastest but least accurate, while the HSV was amongst the slowest of the color models, it turned out to be the most accurate. The experiment was conducted by having subjects match a target color appropriately with each of the different color models. The problem with " the Schwarz interfaces is that they are not typical of the direct manipulation interfaces popular today", as claimed by Douglas and Kirkpatrick in their experiment "Do Color Models Really Make a Difference"[4]. Direct Manipulation refers to the continuous display of the object of interest, and rapid, incremental, and reversible actions that have results that are immediately visible [8]. Douglas and Kirkpatrick, looked at the relationship between the color model and the interface which followed the ideas of direct manipulation, based on performing a color matching experiment with two interfaces (HSV and RGB) that offered different levels of feedback. They found no differences in speed and accuracy between the RGB and HSV color models, but found that increasing the feedback led to improved accuracy of matching [4].

We propose that the dimensions of the color selection method will determine how fast and how accurate the subject is with matching target colors. The crayon picker should be the fastest because it is 1-dimensional and there is no color mixing The subject just has to pick the appropriate color, however it might not be accurate. Next in line is the 2-dimensional HSV picker, which represents a color wheel that involves finding a hue on the wheel closest to the target color and then using the slider below it to control the darkness of the color, which adds more accuracy. Then the RGB picker offers three sliders of red, green, and blue, for 3 dimensions, which can be used to match the target color. However the three dimensions of control will slow the color selection process. CMYK offers cyan, magenta, yellow and black sliders to adjust colors which will make it accurate, but it is more complex due to the four dimensions, and therefore it will not be as fast the HSV or Crayon picker.

Department of Computer Sciences  Direct questions and comments to the student editorial team