Subtractive Light Implications

Color printing uses cyan, magenta, and yellow filters in the form of colorants. Cyan colorants absorb photons with “red” wavelengths but pass photons with “green” and “blue” wavelengths. Similarly, magenta colorants absorb photons with “green” wavelengths and pass photons with “red” and “blue” wavelengths. Finally, yellow colorants absorb photons with “blue” wavelengths but pass photons with “red” and “green” wavelengths. The following naive relationships are often suggested:

Hold on a second. What does the statement Cyan colorants absorb photons with “red” wavelengths but pass photons with “green” and “blue” wavelengths actually mean? It simply means cyan colorants (e.g., inks, toners, wax, etc.) tend to absorb photons with wavelengths greater than 580 nm. What percentage of the photons of a specific wavelength a colorant absorbs depends on the particular colorant and its concentration as well as the specific wavelength under discussion. And here is where it gets very ugly, very fast.

The solid line in Figure 9-3 illustrates the effect of an ideal cyan colorant. At any concentration, it correctly allows all wavelengths below 580 nm to pass. The percentage of photons at wavelengths above 580 nm that pass is strictly a linear colorant concentration function. For example, a zero percent concentration (no colorant) passes all red photons. A 50 percent concentration absorbs 50 percent of the photons at all wavelengths above 580 nm. It’s all pretty simple but, unfortunately, equally naive.

Figure 9-3  Ideal and realistic cyan colorant behavior

The dotted line in Figure 9-3 illustrates more realistic cyan colorant behavior. Note that the colorant absorbs photons with wavelengths less than 580 nm when it should not be doing so. And, above 580 nm the colorant is sometimes allowing red photons to escape that it should be absorbing and absorbing some that should be escaping. Worse, the percentage of photons absorbed varies by photon wavelength and is anything but linear. For example, increasing some black colorants to areas actually lightens the areas rather than darkens them because the additional colorant develops a sheen (specular reflection) that reflects more light. Combine that with some complications such as:

• Colorant spectral reflectance metamerisms
• Varying paper reflectance and colorant absorption
• Varying ambient lighting spectral power distributions
• Varying human visual system nonlinear sensitivities to colors and color shifts based on photon wavelengths and saturations

and you have entered the bold world of device independent color challenges. The effect is that attempts to use straightforward but unfortunately naive approaches to print color pictures and faithfully render them on CRTs are doomed. As one sad example, one printer company president futilely spent an entire day trying to adjust his image colors on a CRT so his son’s color image would print “as it should” on one of his printers.

Even in the simpler case of CRTs, CRT phosphors vary by manufacturer and model with each phosphor exhibiting a unique spectral power distribution intensity based on use (i.e., fading). And, before viewing important images, be sure to leave your CRT on for a few hours and stabilize the room temperature and lighting. And, of course, don’t touch the contrast, brightness, or hue dials capriciously, etc.

In the case of high quality color printers, expect them to have high-speed color look-up tables that specify precise colorant mixes for each possible color. The tables assume specific colorants and reflect painstaking measurements of all possible colorant combinations (don’t buy a generic low-cost cartridge). This allows sophisticated imaging programs to generate faithful output images predictably until someone inadvertently breaks the fragile rules.

As a final note, one of this book’s demonstration programs shows how to print color pictures on low-cost Hewlett-Packard color ink jet printers. The program uses an algorithm Hewlett-Packard Corp. openly includes in its DeskJet printer documentation that helps correct noticeable characteristic color imbalances occurring when cyan, magenta, and yellow ink amounts are nearly equal (near black, white, and gray values).

This algorithm scales the cyan amount back approximately one-third to produce a more faithful achromatic result. Otherwise, the color is tinged jade green. There is absolutely nothing unusual about this particular situation—it’s simply representative of subtractive color technical challenges.

Conclusion

A well known computing industry dignitary from a large Northwestern software company once addressed color scientists at an internationally renowned color research laboratory suggesting they were “making color too difficult.” After all, there is red, green, and blue for CRTs and cyan, magenta, and yellow for printers—that’s “all” that folks needed to know. For most people, faithful image reproduction and device independent color is difficult if not impossible to achieve, period. But, you be the judge.