Event Details

I will describe two ongoing projects that seek to produce
easily-perceived imagery using new theories in cognitive science
that relate artistic images to realistic images through our
understanding of human visual perception.  Much of this talk will
focus on a new computational approach for making convincing
line-drawings of three-dimensional shapes.  It goes beyond occluding
contours (which include the silhouette) and creases, and relies on a
new class of lines: suggestive contours.  Suggestive contours are
lines on the surface that become true contours in nearby viewpoints;
they turn out to be located at certain view-dependent inflections of
the surface.  Suggestive contours convey much richer information
about shape than contours can alone, yet they do so in a way that
harmonizes with the contours.  Supported by a range of examples,
I'll discuss the mathematical properties of suggestive contours,
their perceptual implications, and their computational realization.