Skip to main content

These ghostly portraits are algorithmically-mashed faces from Hollywood movies

These ghostly portraits are algorithmically-mashed faces from Hollywood movies

Share this story

As police look toward a future where face and iris recognition lets them profile anyone walking down the street, two South Korean artists are using it for something less sinister, though still a bit spooky. Shin Seung Back and Kim Yong Hun's new project, called Portrait, calculates the mathematical average of every face found in Hollywood movies like The Matrix, Black Swan, and Avatar, resulting in a ghostly visage that sometimes seems reminiscent of the film's character.

The artists explain to Wired UK that they wrote an algorithm which preserves a film's "identity" by recording color information from the surrounding frame. The facial recognition software captures all faces found in every 24 frames of footage, then generates the final image by finding the mathematical average of the entire movie.

The duo's previous work also plays with the technique: a picture frame installation called Memory averages the faces of everyone who has looked at it; another piece called A Million Seasons compiles Flickr photos of landscapes to create machine-generated digital tapestries. "Our works experiment with various aspects of the technology: what it means that computers see, how it is related to human vision and how it will affect human life," they said. "This project deals with how computer vision can extend human vision. In other words, how it can help us see things that we can hardly or cannot see and view the world from different perspectives."