 |
|
|
SIGGRAPH
2002 Papers Fact Sheet
Conference: 21-26 July 2002
Exhibition: 23-25 July 2002
Henry B. Gonzalez Convention Center
San Antonio, Texas USA
www.siggraph.org/s2002
The SIGGRAPH 2002 Papers program has established itself
as the worldís premier forum for the presentation of new
and exciting results in computer graphics and interactive
techniques. SIGGRAPH 2002 continues the tradition with
groundbreaking papers in many areas. From a record number
of 358 submissions, 67 papers were accepted for presentation
at the conference. John F. Hughes, Brown University, is
the SIGGRAPH 2002 Papers chair.
PAPERS PROGRAM HIGHLIGHTS:
Motion Synthesis from Motion Capture Data
Five SIGGRAPH 2002 papers address the single topic of
motion synthesis from motion capture data from multiple
perspectives. With recent developments in motion-capture
devices, large collections of motion-capture clips are
now available. Just as a single image can be created by
piecing together bits of other images, new character motion
can be created by splicing together pieces of recorded
motion. The hard work is in determining how to break a
recorded motion into clips, and then make the joints between
clips appear seamless.
Each paper addresses this issue in different ways, and
each presents a different application of the general idea.
Motion Graphs shows how to apply this idea to creating
natural walking motions along arbitrary paths under user
control.
Interactive Control of Avatars Animated with Human
Motion Data uses it to control synthetic characters
that represent the userís persona in a virtual world.
The user can control the avatar by making desired motions
in front of a single video camera, for instance, and the
system converts these into plausible 3D motions.
Motion Textures: A Two-Level Statistical Model for
Character Motion Synthesis uses captured motion to
create new but similar motion, so that a virtual character
can dance in a non-repeating way after only a modest amount
of motion-capture data has been recorded.
Motion Capture Assisted Animation: Texturing and Synthesis
allows animators to use traditional keyframe tools to
roughly sketch out an animation, and then can fill in
the details of the motion from the captured data.
Interactive Motion Generation From Examples concentrates
on generating new motions from old in real time, and even
on generating motion for multiple characters at once.
New Hardware
Two papers concentrate on new hardware. SAGE Graphics
Architecture describes in detail the design of a new
graphics system capable of rendering 80 million high-quality
triangles per second. Of particular interest is the support
for high-quality antialiasing of triangles, removing the
jagged-edged artifacts that are especially apparent on
thin polygons and lines. By contrast, Real-time Ray
Tracing on Programmable Graphics Hardware shows how
ray tracing, once thought to be too expensive for any
real-time application, may be competitive with traditional
triangle-based rendering as new graphics hardware evolves.
Ideas for Future Use in Commercial Software
Four image-based papers describe ideas that are likely
to rapidly become commonplace in commercial software.
In Self-Similarity Based Texture Editing, the authors
describe how a user may mark up a portion of an image
(coloring one roof-shingle brown, for instance); the system
then finds all similar portions of the image (all the
other shingles) and modifies them similarly, and the house
gets a brown roof. Instead of changing the color of a
part of an image, one can instead change its geometry.
This means you can enlarge all parts of an image that
are similar to the one beneath the cursor and shrink the
others, so you can make a brick wall have larger or smaller
mortar seams, for example.
Transferring Color to Greyscale Images does just
what its title promises. If you have a black-and-white
image, and a somewhat similar color image, the system
will produce a colored version of your black-and-white
image. For instance, if you have a B/W forest scene, and
a color image of some trees, the system can color in the
forest, make the sky blue, etc. ‚ all without user intervention.
In Object-Based Image Editing, the authors describe
how to edit a photograph in which distinct parts are reasonably
evident and distinguished by color ‚ a photo of a national
flag, or a traffic light, for instance. In such an image,
the user can manipulate regions of fairly constant color
as if they were objects, moving them around, distorting
them, etc., so that one could, for instance, turn all
the round lights of the traffic signal into square ones,
or shuffle their positions.
Finally, in Stylization and Abstraction of Photographs,
the authors extend previous non-photorealistic rendering
work in a particularly innovative way. There are many
algorithms for turning images into digital paintings,
but the results donít often resemble the painting that
an artist would make of the same scene, because artists
tend to concentrate their strokes in the important parts
of an image, and itís difficult for an algorithm to know
which parts these are. In this work, the original image
is presented to a human viewer, whose eye-motions are
tracked. Then paint strokes are applied in regions where
the eye spent more time, giving a much better result.
Paper Titles and Authors
Interactive Control of Avatars Animated With Human
Motion Data
Jehee Lee
Jinxiang Chai
Carnegie Mellon University
Paul S. A. Reitsma
Brown University
Jessica K. Hodgins
Carnegie Mellon University
Nancy S. Pollard
Brown University
Interactive Motion Generation From Examples
Okan Arikan
D.A. Forsyth
University of California, Berkeley
Motion Capture Assisted Animation: Texturing and Synthesis
Katherine Pullen
Christoph Bregler
Stanford University
Motion Graphs
Lucas Kovar
Michael Gleicher
University of Wisconsin-Madison
Fred Pighin
USC Institute for Creative Technologies
Motion Textures: A Two-Level Statistical Model for
Character Motion Synthesis
Yan Li
Tianshu Wang
Heung-Yeung Shum
Microsoft Research Asia
Object-Based Image Editing
William Barrett
Alan Cheney
Brigham Young University
Ray Tracing on Programmable Graphics Hardware
Timothy J. Purcell
Ian Buck
Stanford University
William R. Mark
Stanford University (now at NVIDIA Corporation)
Pat Hanrahan
Stanford University
SAGE Graphics Architecture
Michael F. Deering
David Naegle
Sun Microsystems, Inc.
Self-Similarity Based Texture Editing
Stephen Brooks
Neil Dodgson
University of Cambridge
Stylization and Abstraction of Photographs
Doug DeCarlo
Anthony Santella
Rutgers University
Transferring Color to Greyscale Images
Tomihisa Welsh
Michael Ashikhmin
and Klaus Mueller
Stony Brook University
|
|
|
