Thematically, culturally, the pixelated pattern made sense. But scientifically, it seemed a little off. Why use squares or rectangles, with their unnatural, hard edges? According to the U.S. patent issued for the Marines' MARPAT camouflage, the selection of four-sided pixels had nothing to do with concealment. Rather, the tiny squares were easier to print on fabric. (The same patent concludes, rather meekly, that "camouflage is an art in the process of becoming a science.") The shape doesn't matter, said the textile engineers, so long as the pixels are applied in the correct proportions of color and brightness.
According to O'Neill, who now works as a consultant and does not endorse any particular pattern now under review, any kind of pixel will do so long as it has clearly defined edges. These are necessary for creating a proper texture match with a natural background, he says, even though the details of the pixel pattern are invisible when viewed from any reasonable distance. When I asked about the fact that an enemy soldier would be likely to see just the macro-pattern of blobs and streaks, and not the smaller dots, he chalked up that observation to a BFO—"a Blinding Flash of the Obvious." The pixels have an effect on your visual system, he assured me, even when they're too small to see. It's complicated, he says, and has to do with something in the cortex called "lateral inhibition."
That didn't sound right. Lateral inhibition, which results from the interaction of adjacent neurons in the retina, helps to sharpen our picture of the world by highlighting the edges of objects. But this mechanism applies at the very earliest stages of vision, using the raw data from your eyeballs. If the photoreceptors in the retina can't make out the tiny dots on someone's uniform, then lateral inhibition wouldn't apply. O'Neill had another reason for why the dots are important, though. He pointed out that dots of a few basic colors can be dithered to create new shades, and a richer palette. (Fun fact: In the late-1980s, O'Neill published a novel called Shades of Gray.)
Whatever the merits of the theory, there's only one way to know for sure if digital camouflage works, and that's by testing it. The Canadians had found that their pattern (later adopted by the Marines) performed well in field tests. But the Army's own pixel print flunked its early trials, finishing below the splotchy MultiCam pattern and a feathery design called Desert Brush. For a follow-up study in 2009, researchers posed models in each of 18 different camouflage suits and nine different settings, and asked soldiers to mouse over photographs and click the hidden targets on a computer screen. (Never mind that the screen itself was pixelated.) The Marine Corps uniforms excelled in certain environments, as did the Desert Brush. But if you looked across the full range of backdrops for a single pattern that blended in everywhere, the MultiCam print did better than either one.
The Army now appears ready to make MultiCam into the new standard. But the data suggest that camouflage testing is itself a tricky and ambiguous art, with the big winner from one study fading in the next round. Even the 2009 experiment, which included almost 900 test subjects, came up with some confusing results: The best "universal" pattern of all—even better than the MultiCam—turned out to be an old-fashioned, unpretentious inkblot of khaki, brown and beige. The researchers had labeled this the "Syria" print, though experts now claim that it's a derivative of a German marsh pattern from the 1950s.
Does this mean the testing methods are faulty? Are the latest theories of camouflage misguided? Whatever the problem, the success of the Syria print casts an ominous shadow over the field. For all our advances in the science of camouflage, from the analogue design of the M81 Woodland to the digital MARPAT prints, and then back to the MultiCam, we're still getting lost in the woods.
Correction, July 5, 2012: The photo caption originally identified the soldier pictured as a major. He is a sergeant major. Correction, July 6, 2012: The article originally misspelled "pixelated," as "pixilated." Pixilated means "drunk."