Event Location: NSH 1507
Bio: Jim is a 3rd-year PhD student in the graphics lab. His advisor is Nancy
Pollard. He is interested in systems and interfaces that operate in
real-time and build user intuition.

Abstract: In standard image editing programs like GIMP, images are represented by
storing, for each pixel, an RGB color value. Graphics researchers have
demonstrated that if one, instead, represents an image by storing the
local differences in image intensity, many more interesting editing
operations (seamless cloning, color manipulation) become possible. This
talk will cover the background of gradient-domain editing, as well as my
recent work on speeding up the process of computing an intensity image
from a gradient image using a numerically-optimized multigrid solver
running on the GPU. Thanks to this work, it is, for the first time ever,
possible to paint with gradient-editing brushes in the same way one is
accustomed to painting with intensity-editing brushes.

Presented in Partial Fulfillment of the CSD Speaking Skills Requirement.