Details:
Summary:
One of the problems faced in reproducing an image in a different medium is that the luminance ranges of the original and reproduction are often quite different. While the preferred mapping of the luminance/lightness component is well understood for most images, the mapping of the chromatic components of color images is not. In this paper we examine several alternative mappings, both analytically and experimentally. In the analytical phase of this investigation we postulated that a color reversal transparency was to serve as the original image, and four-color process printing on uncoated paper was to serve as the reproduction medium. We considered several color spaces (XYZ, xyY, and CIELab) and performed compression on the luminance/lightness component while holding the other two coordinates constant. This created three different ways in which to map a color in the original into a color in the reproduction. In addition, we examined the case of compressing the X and Z tristimulus values using the same compression used for Y, which yielded a fourth mapping. Finally, a fifth mapping, with a user-selectable Chroma Compression Ratio (CCR), was examined. We then examined the effects of these five mappings on hue, chroma, saturation, and gray balance. In the analytical phase of this investigation we postulated that a color reversal transparency was to serve as the original image, and four-color process printing on uncoated paper was to serve as the reproduction medium. We considered several color spaces (XYZ, xyY, and CIELab) and performed compression on the luminance/lightness component while holding the other two coordinates constant. This created three different ways in which to map a color in the original into a color in the reproduction. In addition, we examined the case of compressing the X and Z tristimulus values using the same compression used for Y, which yielded a fourth mapping. Finally, a fifth mapping, with a user-selectable Chroma Compression Ratio (CCR), was examined. We then examined the effects of these five mappings on hue, chroma, saturation, and gray balance. The results of our experiments are summarized as follows: Holding X and Z constant while compressing Y resulted in undesirable translation towards yellowish green. The effect is greatest for darker colors. Holding the chromaticity coordinates constant caused an increase in Chroma, while maintaining gray balance and hue. Again, the effect is greatest for darker colors. This is undesirable, because media with limited luminance ranges are typically limited also in Chroma. The remaining three techniques are all related; the last is shown to be a generalization of the other two. They differ in their treatment of Chroma. One leaves Chroma unchanged; it uses a CCR of unity. Another uses a CCR which is the ratio of the L* range of the reproduction to that of the original. The final technique allows the user to set the level of CCR. All three of these methods preserve hue and gray balance. For most originals, the mapping of color coordinates from one domain to another reduces to a problem of determining the CCR. Our experimental evidence indicates a value midway between unity and the L* range ratios of the two media.