Motion Cues for Illustration of
Master thesis project, 2006-2007
Skeletal Motion Capture Data
As spatial data is transferred to an image, it loses
much of its initial information.
Artists use visual cues, non-photorealistic identicators,
to add supplemental information to the image.
To enhance temporal information, they use motion cues,
a category of visual cues that specifically illustrate
the movements in the image.
For some time, techniques such as cartoon illustration
and storyboard design use motion cues to convey temporal
information to the viewer.
This master thesis presents an interactive system that synthesizes
temporal information of an animated character using
artist-inspired motion cues derived from skeletal motion
We propose the combined use of motion arrows, noise waves, and
stroboscopic motion to emphasize movements in a still image.
We let the user manually select a pose of motion
capture data, determine the view point, and add desired
motion cues to the image.
The system automatically adjusts the motion cue
appearance to the view point and allows the
user to fine-tune illustration parameters for greater
expressiveness in real time.
Master thesis [PDF] (french)
NPAR 2007 Article [PDF]
NPAR 2007 Conference slides [PDF]
SIGGRAPH 2007 Reprise slides [PDF]
Some more images
Filtering Structures for the Linux Trace Toolkit Viewer
Kernel traces indicate precisely how the operating system distributes
the execution of the many processes running on a machine at the same time.
While small traces can be easily handled by LTTV,
larger traces, reaching gigabytes of data, cannot be treated as is.
We propose filtering structures that allow the user to choose
what kind of information he want displayed. The user may
specify a filtering expression on the command line or interactively
add filtering conditions using the user interface of LTTV.
Report [PDF] (french)
LTTV web site