Memory is closely tied to the senses, meaning that we’re often able recall the way something looks, smells, or tastes better than what it’s called. For this reason, a Google search isn’t always helpful for people who are, say, visual learners, or are working in a particularly visual field. You can’t drop a color or texture into a search bar, so our ability to rediscover something like a painting or a photo relies heavily on our ability to conjure up the keywords that might bring us closer to it.
That’s something that Alexandra Chemla is working to change. As the founder of ArtBinder, an app that allows galleries and artists to manage and share their inventory digitally, and a former assistant for gallerist Gavin Brown, Chemla often wrestled with this problem after attending art fairs and exhibitions. “I go to 20 art fairs a year and within one I’ll see maybe 200 artworks. I don’t have the kind of memory that allows me to remember names, but I do remember what I saw,” she says. “For people like me–and art is a visual industry–having a way to search for something by a visual property seemed important.”
To meet that need, Chemla has just released ColorSearch, an addictive new feature within ArtBinder that allows users to search for artwork by a specific color or color combination. Offering over 200 hues and pulling from ArtBinder’s immense database of pieces, the tool will bring up a page of matches based on proximity to the shade that was chosen.
To develop the technology, Chemla says one of her engineers, who has a PhD in mathematics, spent months studying color theory to create a color-detecting algorithm that sorts through hundreds of pixels within each image. “One painting that looks blue to the eye might have a lot of yellow in it, but you can’t see it” Chemla says. “One of the things that was technically challenging in developing the algorithm was distinguishing the perceptual from the imperceptual.”
Ultimately, creating an algorithm that looks at the mathematical value of pixel colors allows for the search to be objective, something that Chemla felt was missing in the art world. “One of the problems we were trying to solve for was keyword searches for artwork, which are so subjective they’re flawed. One person’s cobalt is another person blue. We didn’t want to be contributing more to fueling that subjectivity,” she says.
Art collectors and interior decorators will undoubtedly find this technology useful for matching artwork to home decor, but Chemla sees a broader range of applications for the technology. For starters, it’s a powerful tool for discovery: start off searching for art that uses a particular shade of orange and you might wind up learning about new artists, finding new galleries, then clicking around to see what other artists they represent. Scholars could potentially use it to study the colors used in particular decades or movements. And art history professors could one day employ the technology to show connections between artists: how Joan Mitchell’s use of color, for instance, was influenced by Willem de Kooning, whose 1942 painting Standing Man was in turn inspired by Picasso’s Rose Period, and so on.
Chemla and her team are also exploring how they could expand the search functionality to include other visual qualities. “I’m keeping a close eye on various AI technology–like Google reverse image search—to see at what point we’ll be able to distinguish things like line, density, and materials like watercolor or oil painting,” she says. “Over the next five to 10 years I think the technology for that will become much more sophisticated.”
In the meantime, try out ColorSearch here.