For my thesis I considered the automatic generation of hedcut style portraits using
photographs of faces. These renderings mimic those that appear in the Wall Street Journal which are produced by hand with pen and ink. I begin this work by following
Kim et al. who render hedcuts by placing stipples so they follow not only the outline
of facial features but also isophotes (lines with constant illumination). I apply a
variety of image processing techniques to extract facial components and inform stipple
placement and size to illustrate tone and depth of the face. I then expand on the work
of Kim et al. by introducing an interactive tool that allows for a fully parameterizable
version of their approach. Finally I experimented with allowing for areas of negative
space and areas of stipple density variation. In this thesis I report on this tool and
assess the quality of these methods.
Click here to read the final draft of my thesis. All code for my thesis is hosted on
my GitHub.
Live Demo of CLI