Computational Portraiture

I began the course with a theory of parallel histories of photography. One is the familiar consumer history: cameras presented as humanistic devices for art, portraiture, and daily life. Running beside it is another history of cameras built for mapping, measurement, computer vision, astronomy, microscopy, and other technical forms of looking. Treating only the first history as creatively available produces an artificial narrowing.

The course asked: who says we are not allowed to use those other cameras? Portraiture gave us a deliberately human and expressive form through which to test them. We were particularly interested in environmental portraits. A thermal camera, lidar scanner, or microscope might each make a different portrait of a person and the place around them.

James George and I taught the class across four consecutive semesters at NYU ITP. It was an open-entry survey: no prior technical skills were required, and students began with the cameras, phones, laptops, and tripods they already had. We looked at artworks made with photogrammetry, 3D scanning, and computer vision, along with the research and tools behind them. The assignments were very different from one another: students traced a person's signature gesture, rebuilt a face from one photograph, and carried GoPros onto the subway.