The ‘awesome’ filter for your face

The images on the left are the original snapshots; three different modified styles at right.
Image courtesy of the researchers and MIT-Adobe 5K dataset
The images on the left are the original snapshots; three different modified styles at right.

A few years ago Instagram figured out how to make bad snapshots look reliably stylish, a modest service that netted its founders a cool billion dollars. There’s one kind of photo that’s hard to improve, though: the simple headshot. As a recent article in the MIT Technology Review explains, the app’s signature filters are too blunt for the fine-tuned way we perceive faces: It may work to wash a snapshot of a backyard barbecue in sepia tones, but if you apply the same heavy-handed makeover just to a face, the result appears bizarre.

A team of researchers at MIT is working on a solution: an app that recasts mediocre headshots in the styles of famous portrait photographers like Richard Avedon and Diane Arbus. You take a picture of your face, and then choose from a variety of portrait styles, like “low-key and high contrast” or “warm and soft lighting.” Then an algorithm scours a database of high-quality portraits in that style, looking for faces with similar features to your own, and makes “local” adjustments to your original image—subtle changes in light and shading in different places on your face.

In their sample photos, it seems to work: Headshots that looked like they came from a police lineup now appear to depict a stud entrepreneur or an exceptionally deep young actress. The results aren’t perfect, but definitely improvements. And the subtlety and difficulty of the process reveal just how sensitive our brains are to even tiny shifts in how we see a human face.

An auction you win by caring


It’s the kind of scene you’ll find in a movie. An auction is underway, for, say, an exquisite violin or a rare set of books, and two bidders square off: One has the money; the other has the deeper claim on the object—maybe she’s a talented musician, or the books used to be in her family. Of course, the money wins.

Get Arguable in your inbox:
Jeff Jacoby on everything from politics to pet peeves to the passions of the day.
Thank you for signing up! Sign up for more newsletters here

That seeming unfairness explains why a unique kind of auction that recently took place in Sweden feels so satisfying. The event, as reported by the BBC, featured $34,000 worth of artwork, but no money actually changed hands. Instead, auction participants competed based on the intensity of their emotional responses to each piece. Participants were brought one at a time into a room and hooked up to sensors that measured their heart rates and galvanic skin response (i.e., the sweatiness of their palms). Then each auction item was revealed; the people who registered the most excitement were the ones who took home the goods.

“It doesn’t really matter if you feel happy about it, or sad or angry, just don’t be monotone, just feeling something,” said a man named Andreas, the inventor of the system, to the BBC. It’s definitely one way to think about art: that it’s the magnitude, not the type, of response that really counts. Then again, there are “cooler” ways to experience art: It can be an intellectual encounter, or its significance may become more apparent over time. Which is to say, even if we were trying to auction art according to who really deserves it most, we wouldn’t want to simply award pieces to the most overwrought among us.