The visualization of human brain fibers is becoming a new challenge in the computer
graphics field. Nowadays, with the aid of new technologies such as DTI, acquisition of this
information has become available and the generation of complex geometric models that
represent these brain structures is possible. This data is mainly stored in massive sets
of polygonal lines that despite the fact that they are simple geometry, end up in visual
clutter on screen if rendered with naive methods, avoiding the viewer to perceive their real
shape, depth and details.
By taking advantage of modern GPUs, we can execute complex processes at vertex,
primitive and pixel level thanks to shader programs. With these tools at hand, it is now
possible to develop rendering techniques that give a speci c visual appearance to our
model. These techniques consist in providing extra visible cues that help us to notice
these fiber's features and make the clutter disappear, thus revealing its shape and letting
us perceive their orientation, position in space and relative positions with each other.
With this Master Thesis, our intention is to compare two di erent families of methods
that aim to reach these objectives by means of generating halos around fibers: geometri-
cally based and screen-space based approaches. Each of them will try to make that bers
that are far from each other are easily distinguishable, while keeping a compact repre-
sentation for those ones grouped in dense sets. Even so, each one of them will have its
own advantages and disadvantages. This document describes the process of implementing,
testing and comparing them.
