Adalberto Gonzalez may well be one of the best painters of cars in Northern California. He doesn’t work in the eye-popping sparkle-and-shine mode of Cali low-rider culture, and he only rarely finds himself refinishing an Italian exotic. Gonzalez, who goes by the nickname Coco, runs the paint room at Alameda Collision Repair, a high-quality shop that fixes slightly more than 13 cars every day, six days a week. Painting a panel, from a simple ding to something much, much worse is the last stop in a car repair, which makes it a bottleneck. What makes Gonzalez so good is that he's fast. He is an artist at uncorking the bottleneck. But unlike most artists, if you can perceive even the faintest hint of his work, he has made a mistake.
Coco Gonzalez is an ace at repainting a car with different colors—so your eye can't tell the difference.
You’re thinking, big whoop. A car comes in, a 2015 Toyota Camry, let's say, in Ruby Flare Pearl (that’s red) needing a bashed-in door Bondo’d and sanded. You just go to a shelf and take down 2015 Toyota Ruby Flare Pearl, click a canister into an airgun, and swoosh, you’re back on the road, right?
Nope. Car companies have put 50,000 to 60,000 car colors on the road, but even a big body shop like Alameda Collision Repair has just 70 or 80 colors on its shelves. Turns out Gonzalez isn't just a fast painter, he’s a fast matcher. “I get the closest one,” he says, “and then I match the color.”
A rack of possible paints waits to get mixed into a match
At the beginning of the process, a painter tries to match a sprayed-out color on a card to the actual body color.
Gonzalez is making what's called a metamer, a color indistinguishable from a reference even if its physics and chemistry differ. It's not easy. Automotive colors are increasingly sophisticated and complicated. "Harlequins,” for example, show three different colors depending on what angle you’re looking from; new mattes compete with pearls and metallics. But in a deep, philosophical way, none of that matters, because the physics of how light interacts with a paint is less important than the biology and neuroscience of how the eye and brain turn that physics into an idea of color.
Gonzalez begins by consulting a wall-mounted computer touchscreen for basic recipes that replicate colors from the original equipment manufacturers. A given OEM paint might require a recipe of seven, eight, a dozen paints from an aftermarket company like AkzoNobel1 or PPG. (I should disclose here that my cousin's company supplies paint to Alameda Collision and put me onto Gonzalez.)
But even with a recipe in hand, Gonzalez will have to account for “field variance.” That’s a polite way of saying, how messed up is this car? Has it been parked outside for two years? Was it painted at the beginning of a new run, when the assembly line’s paint system might not have been cleaned properly and still contained a little bit of the last color? Is it “darker and dirtier?” A little redder? A little bluer?
The bright lights of the paint room are supposed to show differences rarely apparent to the naked eye.
Clad in a plasticized jumpsuit and a Mickey Mouse baseball cap, Gonzalez works in a walk-in-closet-sized room adjacent to the big, ventilated chamber where he paints the parts. The dozens of individual colors are in white plastic containers on paint-spattered shelves that run from floor to ceiling. Gonzalez pulls a plastic container from a dispenser mounted on the wall, sets it on a scale, and starts pouring colors from the shelf—following the basic recipe. His work table is covered in butcher paper stuck down with yellow tape; the floor is a speckled, unintentional Jackson Pollack painting.
When he’s done mixing, the color still won’t be right.
The floor of a paint shop can become spontaneously self-generating art—that might look a little familiar.
After assembly, cars go through an elaborate automated and industrialized paint process. The so-called "body in white"—the steel-and-aluminum-and-sometimes-plastic-and-carbon-fiber completed car—gets a phosphate dip to clean it and then a bath in an e-coat tank, short for electrophoretic coating, where the resulting electric charge makes a grey-green sludge stick to the car in a thin, homogenous layer. That becomes the surface to which everything else—the colors—will stick.
“After the e-coat we start applying the things that in a body repair shop we’d be trying to mimic the color of,” says Mike Henry, a longtime color expert at PPG, the biggest paint and coatings company in the world. He’s been there for 35 years, but unlike most of the people in his world of color, he isn’t a chemist—Henry got his MFA in studio painting from Miami University.
The next coats are primers, or primers and paints. Chemistries vary, but all paints are essentially a combination of pigment, which reflects and absorbs light to give a specific color; solvent, which carries the pigment onto a surface; and binder, which keeps the pigment mixed. The pigment might also include ingredients like silica that make metallic or pearlescent flakes look finer or coarser. Those kind of effects make a paint “goniochromatic”—meaning the finish looks different from different angles. (Gonio is Greek for "angle.")
Then the paint robots atomize it. “Atomization is essentially exploding a liquid into the finest particle possible,” Henry says. That cloud, sprayed toward the car body, gets a light electric charge to attract it to the target, cutting down on waste. “It’s a fog that adheres to the car.” A heated “bake” cycle cures that layer, and then the robot machinery of the line sprays another layer on, a protective clear coat with all the same ingredients except the pigment. The color of a car isn’t just a surface. It’s a three-dimensional shell that light penetrates and then reemerges from, wholly changed.
When you get into a car accident or rake your car across a pillar in a parking garage, that’s what you’ve scraped off: the complex photonic pathways created by roughly four layers of electrodeposited chemistry.
That’s not what a repair shop lays back on. A good shop will apply multiple layers of primers, clear coats, and color, not to mention repair work that might include epoxies and resins to smooth out dents. The geometries, at the atomic level, that absorb some wavelengths of light and reflect or refract others, might have little to do with the original.
But done right, they look the same.
Light doesn’t just hit the uppermost surface of the clear coat and bounce to your eye. It penetrates as discrete disruptions of the local electromagnetic field called photons. You probably already know that photons can act like particles and waves; their wavelength is their color, yes, but interactions with the particles and resin of the coating can alter that wavelength and its direction.
Black pigment particles absorb light, reducing the number of photons that swing back out and reach someone’s eye. Whiteness, on the other hand, can be as much a property of multiple surface interactions as a function of a pigment like titanium dioxide, the classic whiter-than-white stuff in almost every human-made color. These and other pigments also scatter light, extending or shortening the path of a photon passing by.
Photons can reflect off a surface, bouncing away at the angle as they arrived. They can also refract, heading off on a different trajectory. Both things happen when light hits an object. But a surface can also diffract light, bending its path—when light grazes past an object. If the pigment particles embedded in the surface are roughly the same size or slightly bigger than the inbound wavelength, different parts of the particle interact with different parts of the electric and magnetic field changes that make up the photon. The waves can interfere, both constructively and destructively, changing their scattering and eventual appearance.
(This is called Mie scattering. It's different from Rayleigh scattering, which applies when particles are smaller than the wavelength of light, as with particles in the atmosphere. The sky, full of water vapor, is blue because of Rayleigh scattering. Clouds, made of water droplets, are white because of Mie scattering.)
All this physics actually matters to how something looks. A white coating made with titanium dioxide and some neutral, light-absorbing pigment will appear bluish if the coating’s particles are small and yellowish if they’re big. The difference is the length of the path light takes while passing into and back out of the surface. (And in reality, Mie’s math only applies to spherical particles. For real color science on coatings like paint you need the Kubelka-Munk equations, which integrate travel time through entire layers. Calculus for colors: It’s a thing.)
All of which should make you shriek, OK, but what color is stuff? Because, sure, you could just ask, What wavelengths of light are coming back from a surface? That’s what a human eye senses, via three kinds of receptors called rods, each most responsive to different wavelengths of light.
But that's not the whole story. In fact, what the brain then does with that information is determine where that color falls on two sets of choices: reddish versus greenish and bluish versus yellowish. In trying to figure out how the brain assembles the colors of the world from primary, basic colors, the German physiologist Karl Ewald Hering (1834-1918) theorized that the most basic you could get were combinations of blue versus yellow and red versus green.
Karl Ewald Hering diagrammed opponent colors as perceived by the human brain.
Those “opponent” colors leave afterimages of each other in the eye, Hering pointed out in 1878. And weirder still, even though you see yellowish greens or bluish reds, you never see a red-green or a blue-yellow—hence "opponent." Nobody’s really sure if so-called Hering opponency is actually neurobiological; scientists have been hunting for opponent structures in the brain for decades with limited success. But as a theory, it’s pretty great, because if you turn those two opponent gradients into four quadrants around two perpendicular axes, you can get pretty much any color the human eye can see.
Because we also want to know how light or dark the color is—how much of the so-called achromatic colors white and black it has—you can build another axis, perpendicular again to the other two, a z sticking right through the center, for lightness.
Now you have a three-dimensional colorspace: lightness, and the two color-opponent axes. Add concentric circles coming off of the lightness axis out toward the edges of the color plane to represent saturation—more pastel-like toward the polar axis and more vivid toward the edge—and you can capture almost every perceptible color. This is the idea behind some commercial color systems, like the Swedish Natural Color System. Another one, the CIELAB colorspace (CIE being the Commision Internationale de l’Eclairage, or International Commission on Illumination) also roughly approximates Hering geometry; L, a, and b are the axes, and the color differences in it are supposed to map, geometrically, to actual differences a human being would perceive.
Wrap your mind around this for a second: a metaphoric, three-dimensional space in which any point is a specific color. The distance between any two of those points is the quantified perceptual difference between those colors.
To be fair, CIELAB is actually better for colored lights (emitted light) than for pigments (which absorb and reflect), and it's only approximately uniform. All of the various colorspaces only work for certain lighting conditions, certain fields of view, idealized observers … and most of them fray a bit at the edges, when the quantitative jumps from color to color don’t accurately represent the colors people see.
You could map colorspace in a lot of ways, of course. You could use hue and saturation for the color plane, for example, and then value—lightness or dark—for the z-axis and get a whole other universe. The field that studies that problem, psychophysics, has many, many books on that subject it would like you to read, and color researchers are working on ever more accurate spaces. But they tend to be computationally complex, and quite a bit less elegant than the clean Euclidean geometries of Hering-derived spaces.
You’ll remember, too, that I made kind of a big deal out of metallic and pearlescent shine. They make things even more complicated.
In 2000, a quantum chemist named Eric Kirchner decided to switch jobs. He’d been working for AkzoNobel,1 the second-biggest paint company in the world, and there was an opening in the company’s Color Research Laboratory. “I thought, color,” he says. “That is a great topic.” So he became a researcher there, along with thousands of others. One of the things that lab does is create the software that helps painters like Gonzalez find those recipes.
Kirchner had a more specific topic in mind: sparkle. “The ‘effect coatings’ like metallic and pearl, they change color depending on the angle on which you look at it. That was well known,” Kirchner says. “We thought, well, it’s very nice to be able to measure from different angles, but we also see texture. It’s not a uniform color.” Those goniochromatic effects have a grain, a sparkle that affects how they appear.
“It depends a lot on the lighting conditions. If you have direct sunlight, you really get a sparkling effect. If you have an overcast sky, the sparkling effect disappears and you get a coarseness,” Kirchner says. “It sounds trivial now, but in the literature before 2000, it is never taught like this.”
So his team started testing. They showed people samples of different coatings. The one with no reported sparkle effect at all, they called that a zero. The sparkliest they could make, they assigned a value of 8. From there they could make intermediate versions and ask people to put them in order. “We created a set of 56 samples that were all the same color with different sparkling,” Kirchner says. “That took a long time. We worked on it for months.”
Eventually they had a set of eight panels that their tests determined had not only increasing sparkle values but were exactly the same distance from one another, sparklingly speaking. Which is to say, now they had a metric.
By 2007, they’d built an algorithm that understood all this quantitative blingometry and taught it to a spectrophotometer, a device that can look at a colored surface and measure its properties—basically a precise, well-calibrated, high-end digital camera. It’s a standard gadget now, called a BYK-mac. (If you’re wondering, BYK refers to one of the founders of BYK-Gardner, Heinrich Byk; mac is short for “multi-angle color.”)
So I ask Kirchner: Have you added a fourth dimension to colorspace?
He says it’s much worse than that. “Color depends on angle, and typically you need three or five or six angles to characterize a color,” he says. “Six angles times three color dimensions, that equals 18.”
OK so you—
“—and we added a coarseness dimension, but sparkling also depends on angle. So we found that with three different angles for sparkle, we have a good characterization. So that’s 18 plus one plus three, so we have 22 dimensions.”
When I email Kirchner weeks later to double-check this math, he says yes, indeed, that’s what he meant, but he forgot about another metric called “diffuse coarseness,” when you look at the color under perfectly diffuse light. So 23.
A 23-dimensional colorspace. “And if you’re comparing two car coatings with each other, you have to compare those 23 dimensions and find the difference,” Kirchner says. “Then we have an equation that will tell you how different those car coatings are, visually.”
The people creating and mixing the colors can actually control all that in the paints. Every paint has what’s called a gloss angle, the angle at which it sparkles the most. And metallic colors often actually contain metallic particles. “If those particles are not so well aligned, it will give a lot of sparkle outside the gloss angle,” Kirchner says. “If all the flakes are parallel to the coating surface, it is almost like a mirror, so you get a lot of sparkle close to the gloss angle.” Manufacturers add a “disorienter colorant” to keep the particles from aligning. Conversely, more matte, “murdered out” colors, like the ones German carmakers have introduced in recent years, have a mattifying agent that kills off that gloss.
With all that said—with the vastness of a multidimensional colorspace spreading around you like a tesseracting wine-dark sea—you want to hear something really cool?
Coco Gonzalez does not use a spectrophotometer. He matches colors by eye.
Opposite the computer in his work room at Alameda Collision Repair, atop the shelves of paint, Gonzalez has a set of plastic drawers like the ones you might find on any workbench holding screws and nails. Only these are full of color. On pieces of pasteboard slightly smaller than playing cards attached to key rings, are ranges of colors like points plucked out of colorspace. The backs of the cards have notes—what OEM color they correspond to, how to mix them with aftermarket colors, and so on.
But what really helps Gonzalez are cards he made for himself. Like every good painter, whenever Gonzalez comes up with a new mix or recipe to match a given, specific car, he creates what’s called a spray-out. They’re each about the length and width of a steno pad, and he has stacks and stacks of them, all arrayed according to hue.
“Here,” Gonzalez finally says, pulling from a shelf a half-dozen cards that all look white to me. “Let me show you.” We head into the shop proper, where a Toyota MR2 convertible is in a state of lamentable dishabille—top down, hood removed, fenders dented. Gonzalez wipes grime and grit off the door panel, holds up one of his cards, and looks at me. “Same or different?” he asks.
Gonzalez activates a flashlight shaped like a gun and points it at the line between the card and the car. It’s meant to simulate sunlight. He moves it—from directly behind us to angles of about 45 degrees to our left and right. He moves his head to look at the colors from the side. Side-tone is critical. Delightfully, a different color seen from a different angle is called a “flop.”
Finally I go for it. I point at the card. “It’s yellower,” I say.
He smiles. I got it. He takes the card away. “This one?”
“That’s darker?”
Got it again. OK. I can kind of see what I’m looking for here. Spectrophotometry is very good at filling the variables in the equations for how different two colors are from each other, but the human eye is fantastic at being a “null detector,” exquisitely sensitive to the existence of a difference.
At least for some variables. “Our three cone receptors are sensitive to short wavelengths, middle wavelengths, and long wavelengths, so you can say we have blue, green, and red receptors,” says Henry, the PPG color expert. But he knows that characterization of color receptors doesn’t tell the whole story. Not only do the receptors’ peak wavelengths overlap unevenly—thanks to the evolutionary history of “trichromacy,” in which primates like us lost the ability to see three colors and then gained it back through a lucky mutation—but we don’t have equal numbers of them. Humans have more red-sensitive cones than the others. “So we don’t align with blue and red-blue opponency very well in terms of the Euclidean colorspace,” Henry says.
What’s that mean in real life? “When I was working with spectrophotometers I realized I had a problem. The scale was off in different colorspaces,” he says. “I can see down to a very small numerical value in light colors like white, but then, when you get into a highly chromatic color, especially a chromatic blue, I was having difficulties.” In other words, if Gonzalez had shown me a blue car instead of a white one, I might not have been as good at the game “From a psychophysical point of view, I’m not a very good sensor,” Henry says.
That’s part of the reason that Gonzalez keeps every spray-out he makes. But it’s not the only reason. On the back of each card are more notes, the exact recipe for each one. So the real argument for making them is speed. Instead of having to cook up batches of color from the official, company cards and the software in his workroom, Gonzalez can eyeball a car, grab the right handful of spray-outs, and hold each one up next to it to figure out an even closer mixture.
When he gets a car that’s in between two, he can make them both and combine the colors, or tint up or down with white or black, and make a new spray-out to try. He even jury-rigged something to further speed up the process, an inverted cardboard box with a little pizza-oven-type flap cut out of the bottom and a hole on top to dock a blow-dryer. He mixes a new color, sprays a new card, and slides it under the dryer to get it ready super-fast. Remember, the paint room is the bottleneck in the repair shop. If Alameda Collision Repair is going to move 13 cars a day, the paint must flow.
So, I ask Gonzalez, do you share those cards with other painters?
“No way,” he says, drawing his stack closer to his body protectively. “Those are mine.”
A century of spectrophotometry and colorimetry and a history of color theory dating back to Aristotle still can’t beat Gonzalez’s eye. It’s not for everyone. “I’m a physicist, and physicists have a hard time doing it like this, because a physicist first wants to understand how a material is composed and, based on that, to match it,” Kirchner says. “But what do we see in this coating? And based on that, how can we match? That’s a much better strategy.”
Of course Gonzalez knows more tricks. He paints like a dancer dances, fluidly and precisely. That’s how he avoids “tiger-stripes” caused by overlapping layers of clear coat. He knows how to lay down clear primer alongside the area he’s painting to help blend the new with the old. He moves like an artist.
In fact, watching him work makes me think I’ve undercounted the dimensions of colorspace yet again. Photons don’t move from object surface to eye instantaneously. Light has a speed, a distance-per-time. It’s just shy of 300 million meters per second in a vacuum; different media slow it down. So figure, if you’re looking at a car from arm’s length, your brain is actually processing whatever color it was 100 millionths of a second ago. Not nothing. Almost nothing, but not nothing. It could have changed. You don’t know.
Go further. When Gonzalez is done with that MR-2, it will once again be the white it used to be. That is, not the original color it was five years ago when it rolled off the line, but the color it was two weeks or so ago, just before that unfortunate collision.
So make it 24 dimensions of colorspace. Because in controlling for all the variables he can see, Gonzalez is controlling one other thing: time.
1 UPDATE 7/27/18 9:35 AM Fixed spelling of AkzoNobel
More Great WIRED Stories
- How Google's Safe Browsing led to a more secure web
- PHOTO ESSAY: The most exquisite pigeons you'll ever see
- Scientists found 12 new moons around Jupiter. Here's how
- How Americans wound up on Twitter's list of Russian bots
- Beyond Elon's drama, Tesla's cars are thrilling drivers
- Get even more of our inside scoops with our weekly Backchannel newsletter
