Desaturate: As Good as Convert to Grayscale?

By Dick Dickerson & Silvia Zawadzki Back to

Dick_Zawad_MA_2008_1

In perusing the countless color digital images we seem to accumulate, our eyes are always open for one that would work well in monochrome. We have a tendency, when reviewing a batch of color, to stop every now and then and hit Desaturate or Convert to Grayscale, just to get a quick idea of the black-and-white possibilities. We’ve been using the two commands almost interchange- ably for this, and only recently took a serious look at them—and will never again use Desaturate for quick peeks.

The color image shown here, quickly assembled from items in Silvia’s kitchen, is what turned the light bulb on. Consider the two lower images. The lower left one is a desaturated version of the color image, and the one on the lower right has been converted to grayscale. There is no hard and fast rule about which shade of gray a par- ticular color should translate to, even in the film world. But the translation does need to be visually credible. Compare the items in the pan individually in both Desaturate and Grayscale to the original color. Items such as the orange pepper on the pan’s right and the lemon really jump out as seriously wrong. Most of the other items fail the visual credibility test to a lesser but certainly significant degree.

The Desaturate command is reputed to “remove color” while “maintaining relative lightness.” Clearly it does not, particularly with anything containing an element of yellow. Wondering how it could be so wrong, we explored the command in more depth and were amazed to learn how it works. Every shade of every color in the original has Red, Green, and Blue values between 0 and 255 assigned

to it. For example, a smooth spot on the orange pepper is assigned 183R 121G 1B. The only thing Desaturate does is to select the highest and lowest of these three figures, regardless of what colors they represent, and average them. The third (middle) color value is ignored. (We have confirmed this with dozens of other colors). In the case of the spot on the orange pepper this means De- saturate converts it to a gray value of (183 + 1) / 2 = 92. Converting to Gray- scale, on the other hand, renders the same spot as a much lighter—and much more visually credible—gray value of 135.

So why does Convert to Grayscale work so much better? We have been unable to fathom the exact algorithm that drives it, but we do agree with a pop- ular opinion found on the Web and in many Photoshop books that claims it works in part by invoking a Channel Mixer routine that assigns to gray a value equal to 30% of the original’s Red, 59% of its Green, and 11% of its Blue channel. In the case of our pepper spot (the upper right image in the illustration), this means the corresponding gray is 126, very close to, but not the same as, grayscale’s 135.

The green-peaking sensitivity distribution of the grayscale command or 30/59/11 Channel Mixer is not unlike the color sensitiv- ity of our eyes. Comparable too is the spectral sensitivity of panchromatic films. They also peak in the green, though they often do have relatively more blue sen- sitivity than Convert to Grayscale invokes. Noteworthy too is the fact that the 30:59:11 ratio is a standard built into color TVs set for black-and-white broadcasts.

So what do we conclude? Converting a color image to grayscale or summing 30%R, 59%G, and 11%B produces shades of gray that certainly honor what we have termed visual credibility. But the Desaturate command, which effectively sums 50% each of the darkest and lightest, whichever primaries they represent? We have to think any similarity between the resultant gray shade and what our visual system tells us to expect is, at best, rather coincidental.


About the Author

Dick Dickerson & Silvia Zawadzki
Contributor
Dick Dickerson and Silvia Zawadzki are retired Kodak black-and-white product builders who have authored numerous articles for PT. They can be contacted at querybw1@aol.com. Dick and Silvia reside in Rochester, NY.