Achievement

DeCarlo (Computer Science), Singh (Psychology) and trainee Kevin Sanik conducted a large-scale perceptual study of line drawings made by artists or by computer algorithms. The study demonstrated that for some types of shapes, computer algorithms did just as well in conveying 3D shape as a human artist or a shaded 3D rendering. In other cases, there were clear differences, with artist drawings being more effective than computer drawings (and both less effective than a shaded rendering). Even so, people were quite consistent in what shape they saw when they looked at one of these pictures. These conclusions were based on over a quarter-million measurements of local surface orientation gathered using Amazon Mechanical Turk. The data set as been publicly released to encourage further analyses and additional studies of this type.