A new technique in computer vision may enhance our three-dimensional understanding of two-dimensional images.