Chapter 9
Device Color Consideration Overview

What This Chapter is About

Previous chapters discussed light and color, and how we perceive them. This chapter provides an overview discussing considerations that complicate faithful color image reproduction for both additive and subtractive devices. Usually referred to as device independent color, the end objective is to render identical output regardless of rendering device. Since, say, advertising agencies demand to know exactly how color pictures will appear in various magazines, it is big business.

James Clerk Maxwell’s Color Reproduction Experiments

In 1861, James Clerk Maxwell produced the first trichromatic (three-color) color photographic images by photographing scenes using black and white film. Actually, he photographed any given scene three times—once through red glass, once through green glass, and once through blue glass. Maxwell then created positive slides, referred to as separations, for each of the three images and projected them through lantern projectors.

The light from the lantern projector displaying the slide image captured through red glass then passed through red glass. Similarly, light emitted from the projector displaying the slide image captured through green glass passed through green glass, and light emitted from the projector displaying the slide image captured through the blue glass passed through blue glass. When the three now-colored slide projections were superimposed on a white wall, a seemingly colorful image resulted that visually contained many more colors than just red, green, and blue. All modern color imaging processes owe their origin to the principles these experiments demonstrated. As amazing as Maxwell’s experiments were in their day, their simplicity belied human perception complexities.

Additive Light Implications

Suppose for a moment Maxwell could use blue glass lantern projector filters that transmitted power at wavelengths within a narrow 1 nm wide range centered at 450 nm and absorbed all other wavelengths. Similarly, suppose he could use an analogous green and red filter centered at 510 nm and 650 nm respectively. Figure 9-1 illustrates the human retinal cone response curves discussed earlier and superimposes these hypothetical red, green, and blue filter ranges. This immediately gives rise to a few problems.

Figure 9-1  Response characteristics of the eye

First, looking at the figure, it is clear that this filter set only allows roughly 1 percent of all visible light to pass (3 nms out of 370 nms). Such low resulting power lengthens exposure times during image capture and darkens images during viewing. Second, light passing through the blue filter essentially only stimulates the blue cones because blue stimulations dominate the red and green cone stimulations at 450 nm. Similarly, light passing through the 650 nm filter essentially stimulates only the red cones. However, light passing through the 510 nm green filter significantly stimulates the M-, L-, and S-cones.

Thus, attempting to reproduce green objects in a scene using this approach introduces unwanted red and blue stimulations which desaturates greens and makes white objects exhibit magenta tones. Adjusting projector light intensities may generate more genuine appearing whites but compromises saturated color fidelity elsewhere. So, additive color reproduction requires tradeoffs when attempting to reproduce a spectrally rich scene using scant spectral resources.

An alternative approach would use filters that transmitted a wider wavelength range, but that exacerbates unwanted stimulation problems because a quick inspection of Figure 9-2 quickly reveals that there is no possible filter set that stimulates only the green cone without stimulating either blue or red cones, and possibly both. The feeble mechanisms you have are simply unable to stimulate the complex nonlinear triangulation computations human vision systems perform. And, it’s worse in color printing.

Figure 9-2  Filter sets