 |
January 2003
Monday January 13th, 2003: 6p.m.- 8p.m.
IBM, 590 Madison Avenue at 57th street, rooms 609/610
Vision, vision, who has the vision? One prediction is easy in the software/IT industry: our industry will change, and new technologies and business
practices will be introduced. Of course, predicting WHAT those changes will be and WHEN they will occur is a little more difficult.
This month's Monthly Meeting will feature homegrown New York City "Big Picture" visionaries. The discussion will cover both short
term and long term. What is the prognosis for Wi-Fi? Bluetooth? What about "The Semantic Web"? Web services? Pervasive computing?
This event is free to NYSIA members, $20 for non-members.
|
 |
 |
 |
 |
January 2003
Friday January 24, 2003 11:30 a.m. Room 1302 WWH 251 Mercer Street New York, NY 10012-1185
Speaker: John Moody, Department of Computer Science, Oregon Health & Science University
Title: Learning to Trade via Direct Reinforcement
Abstract:
A paradigm shift is underway in reinforcement learning (RL) research. The dominant approach to
RL over the past 50 years has been based on dynamic programming, whereby RL agents learn an
abstract value function. An alternative approach, Direct Reinforcement (DR), has recently been
revisited, wherein DR agents learn strategies to solve problems directly. DR can enable a
simpler problem representation, avoid Bellman's curse of dimensionality, and offer compelling
advantages in efficiency.
We review and contrast the major approaches to RL, present Direct Reinforcement, and
describe its application to asset management with transaction costs. In this very
challenging and uncertain domain, DR agents seek to discover strategies that maximize
profit, economic utility or risk adjusted returns. The potential powers of DR are
illustrated through an asset allocator and an intradaily foreign exchange trader.
Other promising applications of DR may be found in robotics, autonomous vehicles,
industrial control, telecommunications, data mining, adaptive software agents, and
decision support.
Refreshments will be served at 11:15 a.m. in room 1302 Warren Weaver Hall.
|
 |
 |
 |
 |
January 2003
Friday January 31, 2003 11:30 a.m. Room 1302 WWH 251 Mercer Street New York, NY 10012-1185
Speaker: Yali Amit, University of Chicago (Statistics and Computer Science)
Title: An Integrated Network for Invariant Visual Detection and Recognition
Abstract: I will describe an architecture for invariant visual detection and recognition. Learning is performed in a
single central module. The architecture makes use of a replica module consisting of copies of retinotopic layers of
local features, with a particular design of inputs and outputs, that allows them to be primed either to attend to a
particular location, or to attend to a particular object representation. In the former case the data at a selected location
can be classified in the central module. In the latter case all instances of the selected object are detected in the
field of view. The architecture is used to explain a number of psychophysical and physiological observations: object
based attention, the different response time slopes of target detection among distractors, and observed attentional
modulation of neuronal responses. We hypothesize that the organization of visual cortex in columns of neurons responding
to the same feature at the same location may provide the copying architecture needed for translation invariance.
Refreshments will be served at 11:15 a.m. in room 1302 Warren Weaver Hall.
|
 |
 |
 |
 |
February 2003
Friday, February 7th
11:30 am
Room 1302 WWH
251 Mercer Street
New York, NY 10012-1185
Speaker: Dr. Laurent Balmelli, IBM Research
Title: Space-Optimized Texture Maps And Extension To Volume Warping
Abstract:
Texture mapping is a common operation to increase the realism of
three-dimensional meshes at low cost. We propose a new texture
optimization algorithm based on the reduction of the physical space
allotted to the texture image. Our algorithm optimizes the use of
texture space by computing a warping function for the image and new
texture coordinates. Neither the mesh geometry nor its connectivity
are modified by the optimization. Our method uniformly distributes
frequency content of the image in the spatial domain. In other words, the image
is stretched in high frequency areas, whereas low frequency regions are
shrunk. We also take into account distortions introduced by the mapping
onto the model geometry in this process.
The resulting image can be resampled at lower rate while preserving its
original details. The unwarping is performed by the texture mapping
function. Hence, the space-optimized texture is stored as-is in texture
memory and is fully supported by current graphics hardware. We present
several examples showing that our method significantly decreases texture
memory usage without noticeable loss in visual quality.
We recently extended our warping technique [Vis2002] to intensity
function defined on the cube, e.g. volume dataset. We use our method to
perform automatic adaptive isosurface extraction without the use of any
hierarchical datastructure such as octrees, etc. An isosurface is
extracted from a warped volume, then the resulting set of vertices is
unwarped such that the surface is rescaled to its initial proportions.
We demonstrate the usability of the technique with several well known models.
Refreshments will be served at 11:15 a.m. in room 1302 Warren Weaver Hall.
|
 |
 |
 |
 |
February 2003
Friday, February 14
11:30 a.m.
Room 1302 WWH 251 Mercer Street New York, NY 10012-1185
Speaker: Ron Goldman, Rice University
Title: Fractals Everywhere: The Turtle vs. Dr. Barnsley
Abstract:
Turtle competitions have appeared in many guises in literature,
philosophy, and mathematics. Aesop (500BC) tells a tale about a
celebrated race between the tortoise and the hare. Zeno (450BC)
presents a famous paradox, central our modern understanding of
convergence, concerning a race between Achilles and the tortoise.
Lewis Carroll (1870) wrote a dialogue between Achilles and the tortoise,
where the tortoise sets out to convince Achilles that much like Achilles
can never catch the tortoise in a race, he can never get to the end of
any proof in mathematics. Hofstadter (1980) in his book on Godel,
Escher, and Bach also has several dialogues between Achilles and the
ubiquitous tortoise.
Today the turtle has gone high tech. Reincarnated inside the computer,
the turtle has his own programming language (LOGO) for studying
geometry. Fractals are generated in LOGO via recursive programs. The
turtle's modern nemesis is no longer the fleet footed, slow witted
Achilles, but rather the greedy capitalist Dr. Barnsley, author of the
book Fractals Everywhere. Dr. Barnsley generates fractals using
iterated affine transformations, and he plans to take over the world of
data compression with this approach. Here then we have two methods for
generating fractals: recursive turtle programs and iterated affine
transformations. Given two models of computation, it is natural to ask
which one is more powerful? The purpose of this talk is to explore this
question from the turtle's perspective.
|
 |
 |
 |
 |
February 2003
Thursday, February 20
3 pm
719 Broadway, 12th Floor Large Conference Room
Physically Based Modeling of Soft Tissue
for Craniofacial Surgery Simulations
Evgeny Gladilin
Zuse Institute Berlin
In craniofacial surgery, the realistic prediction of the patient's postoperative appearance is of
utmost importance. Patients with facial deformities or paralysis are severely limited in their
interpersonal communication abilities. In such cases, the reestablishment of aesthetical appearance
and normal facial expressions is the primary concern of the corrective surgical impact.
Meanwhile, modern medical imaging techniques, such as computer tomography (CT) and magnetic resonance
imaging (MRI), are widely-used for diagnostic and visualization purposes, enabling the derivation of
useful 3D models of individual anatomy. On the other hand, state-of-the-art numerical techniques,
such as the finite element method (FEM), provide a modeling platform for the realistic simulation
of soft tissue behavior. This talk presents an approach that takes advantage of both the correct
geometrical modeling of individual anatomy and the consistent physical modeling of soft tissue and
muscles on the basis of underlying biomechanical laws.
|
 |
 |
 |
 |
March 2003
Wednesday, March 19th
7-9 pm
719 Broadway, 12th Floor Large Conference Room
NEW MEDIA: Foundations and Futures
Ken Perlin and Christiane Paul in discussion with Noah Wardrip-Fruin and Nick Montfort
Six years ago an NYU CAT-sponsored project began to create a resource for understanding new media's
foundations - aimed at educators and students, scientists and artists, critics and journalists.
Announcing a celebration of The New Media Reader (MIT Press, 2003) featuring two advisors to the project:
- Ken Perlin, scientist, designer, and artist
- Christiane Paul, curator, publisher, and author
- in discussion with Noah Wardrip-Fruin and Nick Montfort, editors of The New Media Reader
(for sale at a 1/3 event discount: $30)
The New Media Reader collects articles, essays, images, and other documents about the development of the computer
as a means of expression - as well as functioning new media objects (from Eliza to Missile Command) and digitized video
(from Engelbart's 1968 demonstration to Lynn Hershman's Lorna)
Wednesday's event will include discussion of The New Media Reader, presentations of new projects from Ken Perlin and
Christiane Paul, conversation about how new media's past can inform its future, and light refreshments.
Ken Perlin is co-director of both the NYU CAT and NYU's Media Research Laboratory; as well as
a major figure in computer graphics, animation, and human-computer interaction; and winner of a technical Academy Award for
"Perlin Noise."
Christiane Paul is the Whitney Museum of American Art's adjunct curator for new media; as well as
the publisher of the new media journal Intelligent Agent; and author of Digital Art forthcoming from Thames and Hudson).
More information about The New Media Reader, including excerpts: http://www.newmediareader.com
|
 |
 |
 |
 |
April, 2003
Wednesday, April 9th
2pm
719 Broadway, 12th Floor Large Conference Room
Frederic Leymarie, Brown University
A Directed graph representation for the 3D Medial Axis and Shape Computational aspects and applications
Abstract: The medial axis (MA) representation has been explored mainly for 2D problems since the 1960's in pattern recognition and image analysis.
The MA makes explicit certain symmetries of an object, corresponding to the shocks of waves initiated at the input samples, but is itself
difficult to directly use for recognition tasks and applications. Based on previous work on the 2D problem, we propose a new
representation in 3D which is derived from the MA, producing a directed graph we call the shock scaffold. The nodes of this graph are
defined to be certain singularities of the shock flow along the MA. This graph can represent exactly the MA --- and the original inputs
--- or approximate it, leading to a hierarchical description of shapes. We develop accurate and efficient algorithms to compute for 3D
unorganized clouds of points the shock scaffold, and thus the MA, as well as its close cousin the Voronoi diagram. Our computational
methods relies on clustering and visibility constraints. We then propose a method of splitting the shock scaffold in two sub-graphs,
one of which is related to the (a priori unknown) surface of the object under scrutiny. This allows us to simplify the shock scaffold
making more explicit coarse scale object symmetries, while at the same time providing an original method for the surface interpolation of
complex datasets.
http://www.lems.brown.edu/~leymarie/phd/
http://www.lems.brown.edu/vision/researchAreas/Shocks3D/
** In collaboration with Professor Benjamin Kimia, Brown University.
|
 |
 |
 |
 |
April, 2003
Wednesday, April 9th
6-8 pm
719 Broadway, 12th Floor
Gallatin Arts Festival 2003
Rebecca Ross, CAT Research Scientist and Students
A forum for Gallatin artists to show their work A way to build community within Gallatin
and among Gallatin artists; A celebration of the creative process; And a two week festival, April 4th-13th.
GAF began eleven years ago as a Gallatin graduate thesis. The original idea behind the festival was to build community through the arts.
GAF has happened every year since then, and has grown and changed with each iteration. GAF is and has been a constantly evolving entity.
See www.gallatinartsfestival.org
|
 |
 |
 |
 |
April 2003
Thursday, April 10th
2pm
719 Broadway, 12th Floor Large Conference Room
James Davis Honda Research Institute, USA
A Sketching Interface for Articulated Figure Animation
Keyframe animation in computer graphics is currently dominated by IK
tools such as Poser and Maya, however these tools are often difficult for novices to use effectively. This talk will introduce a new
interface for rapidly creating 3D articulated figure animation, from 2D sketches of the character in the desired key frame poses. Since the
exact 3D pose corresponding to a 2D drawing is ambiguous, we first reconstruct a set of possible 3D configurations and then apply a set
of constraints and assumptions to present the user with the most likely 3D pose. The user can refine this candidate pose by choosing
among alternate poses proposed by the system. This interface is supported by pose reconstruction and optimization methods specifically
chosen to work with imprecise hand drawn figures. In addition to discussing the interface in more detail, I'll show a number
of results produced by artists who range from novice to expert.
James Davis is currently a senior research scientist at Honda Research
Institute USA, working on range sensing and motion recovery for human
biomechanics applications. Dr. Davis obtained his PhD from Stanford
University in 2002. While there he occasionally held consulting
positions with Pacific Bell, Apple Computer and Presenter Inc. His
work on image mosaicing was commercialized as part of Sony's
PictureGear, and earlier work on multimedia interfaces was eventually
released as Prentice Hall's "Masterworks: A Musical Discovery".
|
 |
 |
 |
 |
April 2003
Thursday, April 10th
5-7pm
719 Broadway, 12th Floor
New York University has been chosen by Sun Microsystems, Inc. to be a Sun Center of Excellence for Digital Libraries.
NYU and Sun Microsystems are joined in this partnership by Ex-Libris, a leading provider of library
automation software. Sun has provided NYU with state of the art equipment, enabling us to create the technical
infrastructure to accomplish in the electronic realm what libraries always have done in the physical world:
store, distribute, and preserve scholarly knowledge. On Thursday, April 10th from 5:00 p.m. to 7:00 p.m. at the
NYU Center for Advanced Technology at 719 Broadway (12th floor), there will be a reception to celebrate the new
NYU Sun Center of Excellence. Guests will preview some of the exciting digital projects that are being made
possible by this new venture. Special guests will include Kim Jones, Vice President for Global Education and
Research of Sun Microsystems; Robert Berne, NYU Senior Vice President for Health; and Michael Kaplan, Director
of Product Management and Operations of Ex-Libris.
|
 |
 |
 |
 |
April 2003
Friday, April 11th
3pm
719 Broadway, 12th Floor Large Conference Room
Mosaicing New Views: The Crossed-Slits Projection
Assaf Zomet, Doron Feldman, Shmuel Peleg, and Daphna Weinshall
We introduce a new kind of mosaicing, where the position of the sampling strip varies as a function of the input camera location.
The new images that are generated this way correspond to a new projection model defined by two slits, termed here the Crossed-Slits
(X-Slits) projection. In this projection model, every 3D point is projected by a ray defined as the line that passes through that point
and intersects the two slits. The intersection of the projection rays with the imaging surface defines the image. X-Slits mosaicing provides
two benefits. First, the generated mosaics are closer to perspective images than traditional pushbroom mosaics. Second, by simple
manipulations of the strip sampling function, we can change the location of one of the virtual slits, providing a virtual walkthrough of a
X-slits camera; all this can be done without recovering any 3D geometry and without calibration. New views are generated having realistic
changes in parallax, reflections, and occlusions.
|
 |
 |
 |
 |
April 2003
Tuesday April 15th
4-6pm
719 Broadway, 12th Floor Large Conference Room
Napster Audio and Video: Innovations in the Network
Live webcast at http://cat.nyu.edu/meaow/glocal3.ram
What are the possibilities for internet based distribution and production of video and audio?
Napster, Gnutella and their descendant have demonstrated famously the sheer scale of p2p filesharing systems, and
the difficulties of exploiting this for the benefit of traditional entertainment products under traditional
intellectual property regimes. However, less attention has been paid to the
emerging audio and video products and the new genres of cultural product
that exploit netbased distribution and production. This panel will survey
different experiments and projects in this realm, specifically, projects
that are designed to promote and sustain diverse cultural resources, generating demonstrable social value.
Christian Nold: is the author of the Author of Mobile Vulgus, a
controversial book about politically activated crowd dynamics. He is
currently at the Royal College of Art where he is developing the Community Edit system.
Pit Schultz lives and works in Berlin. Currently involved into radio
projects he is the cofounder of bootlab.org, klubradio.de, nettime.org, mikro.org.
Natalie Jeremijenko is in the Faculty of Engineering, Yale University,
where she runs the Experimental Product Design program(xproduct)--a program
and courses that explore technological innovation for social progress. She
currently has an exhibition at Art in General that demonstrates several
audio and video systems designed for the notforprofit arts sectors to
promote participatory institutional agendas.
Sal Randolph lives in New York and produces independent art projects
involving gift economies and social architectures, including Free Words,
the Free Biennial and Free Manifesta. She has recently been developing new
work in the areas of open source/copyleft music distribution (Opsound) and
political organization (0pcopy). nb UPDATE
Wolfgang Strauss will also join us having just rtned safely from the
Sharjah Biennial in the UAE to report on his recent streaming projects in
the Middle East. Wolfgang is the founder of thing.net
Neil Seiling--former Executive Producer of PBS television series Alive
From Off Center. A Media Arts Curator since 1978, with an emphasis on
building links between multi-disciplinary artists and their audiences through
media development. Served on inaugural panel for short films at 1995
Sundance, and NEA Film/video Panel.
Alan Toner-Studies collaborativity, and the effect of information
enclosure on cultural production and social life. Native of Dublin, Ireland.
Studied Law at Trinity College Dublin, and NYU Law School. He is currently a
fellow in the Information Law Institute at NYU Law. Member of Autonomedia editorial collective.
Zeljko Blace is a co-founder of [mama], a media lab and culture club in Zagreb. He is presently taking part
in a number of projects: Kultura NOVA, a multimedia institute organized by the European Cultural Foundation & Open
Society Institute. Zeljko has organized and curated a number of new media events: GenArt2002, an annual exhibition,
and recently Reality Check for Digital Utopia, a digital culture encounter.
Mark Davis is an Assistant Professor in the School of Information Management and Systems, UC Berkeley.
His work is focused on creating the technology and applications to enable daily media consumers to become
daily media producers. His research and teaching encompass the theory,
design, and development of digital media systems for creating and using media
metadata to automate media production and reuse. Kate Rich is a sound engineer and activist. She is known to work for
the bureau of inverse technology.
|
 |
 |
 |
 |
April 2003
Thursday April 17th
2 pm
719 Broadway, 12th Floor Large Conference Room
Host: Demetri Terzopoulos
Rahul Swaminathan, Columbia University
Perspective cameras have been the center of all vision systems as well as
algorithms. However, there have been many instances when non-perspective
imaging systems have proved to be more useful in applications such as
surveillance, omnidrectional imaging amongst others. In this talk I shall
investigate the general class of non-perspective imaging systems. We
first develop a simple taxonomy of imaging systems based on its imaging
model. Using this model we analyze non-perspective imaging systems
including wide-angle cameras, camera clusters and generic cameras with
viewpoint loci. We present new models to describe non-single viewpoint
sensors. Using these models we analyze their resolution characteristics as
well as present techniques to calibrate such cameras. Finally we turn to
the effects of multiple viewpoints on images and the distortions
introduced thereof. We call such distortions as caustic distortions and
investigate ways to quantify and reduce these distortions.
|
 |
 |
 |
 |
April 2003
Tuesday April 22nd
3 pm
719 Broadway, 12th Floor Large Conference Room
By Professor T. Theoharis,
University of Athens and University of Houston
Department of Computer Science
A huge amount of research effort has been expended in Computer Graphics to date. As a result, a considerable body of knowledge and
devices have been produced. In fact, one of the most expensive and sophisticated component of the everyday PC is the graphics card. At
the end of the day, a reasonable question to ask is whether all the effort spent toward the goal of realistic and efficient image
generation is good for just that. In this talk we try to show that certain commonly available graphics techniques are of much wider
applicability. It is therefore important that their function is understood by non-specialist users. Suitable mathematical
formalisation can lead to better understanding, by clarifying their essential function. They can then be applied to other
problems. Examples are drawn from recent work of our group in object reconstruction, symmetry detection etc.
Short Biography
Professor Theoharis, received his D.Phil. in computer graphics and parallel processing from the University of Oxford, U.K. in 1988. He
subsequently served as a research fellow (postdoc) at the University of Cambridge, U.K. and is with the University of Athens, Greece since
1993. He is currently on sabbatical leave at the department of Computer Science, University of Houston, Texas. His main research
interests lie in the fields of Computer Graphics, Visualization and Archaeological Reconstruction.
|
 |
 |
 |
 |
April 2003
Friday April 25th
11 am
719 Broadway, 12th Floor Large Conference Room
A joint Gallatin School and NYU-CAT event
|
 |
 |
 |
 |
April 2003
Friday April 25th
4:30 pm
719 Broadway, 12th Floor Large Conference Room
Host: Demetri Terzopoulos
Dr. Yuichi Motai, Purdue University
We are designing a human-computer interaction (HCI) system for teaching a robot to acquire vision models of 3D objects from their 2D
images. The purpose of this study is to establish efficient collaboration between machine and human so that the total performance
of robotic vision system could be improved. Building 3D vision models is carried based on the following two guiding principles of HCI: 1)
Provide the human with as much visual assistance as possible to help the human make a correct input; and 2) Verify each input provided by
the human for its consistency with the priori inputs. In this collaborative approach, for example, epipolar lines superimposed on
the 2D images assist the human to facilitate the 3D feature reconstruction of stereo correspondences by the computer system. The
human can interactively suggest the next-best viewpoint for robot to capture images, examine outputs of different feature extractions with
a menu of low-level segmentators, and choose the one that is perceptually most appropriate to acquire the vision model. The 3D
vision models are then used for the object localization in the robotic manipulation. Online demos and videos validate this study, funded by
Ford Motor Company.
Keywords:
Robot vision, Human-computer interaction, Image-based object model acquisition
Presenter:
Dr. Yuichi Motai, School of Electrical and Computer Engineering,
Purdue University
Bio: Yuichi Motai received a Bachelor of Engineering degree in Instrumentation Engineering from Keio University in 1991, a Master of
Engineering degree in Applied Systems Science from Kyoto University in 1993, and a Ph.D. degree in Electrical and Computer Engineering from
Purdue University in 2002. He was a tenured Research Scientist at the Secom Intelligent Systems Laboratory from 1993 to 1997. His research
interests are in the broad area of computational intelligence; especially in computer vision, human-computer interaction, image
synthesis, ubiquitous computing, personal robotics, and smart sensing.
|
 |
 |
 |
 |
May 2003
Tuesday, May 6th
11 a.m.
Warren Weaver Hall, Room 1302
John Buchanan
Video games are now a part of our culture for good or bad. In general people under 40 tend to play video games while people over 40
do not. Next year the dividing line will be 41. In this talk I will present an overview of the video game industry, in particular I will
talk about the incredible growth rate of the Video game industry compared to other sectors of the entertainment industry. I explain
where the growth areas of our industry are. Having set the stage I will talk a little about Electronic Arts
(after all they are paying for this trip). The talk will then turn drastically into a technical talk (for a brief period) about how
animations are put into a game and where we get the animations from. After this brief technical interlude (Do not worry, there is not a
single mathematical equation present) I will meander into the world of story telling, in particular interactive story telling. What you will
realize is that interactive story telling is a rich and un-explored domain where we are just starting to understand the problem. So you
will hopefully be challenged to think about how to tell an interactive story.
And of course, if you would like to derail the talk by asking interesting questions, then we can go there. But I can tell you that
derailing my talks is way more fun over beer where we can build our own interactive story, and if you are buying the beer, then you will be
the hero.
These are John Buchanan's career highlights. In 1976 he and several students in boarding school chip in to buy Pong.
It is promptly confiscated by the Masters and made communal property of the school. In May of 1981, after being successfully employed as a janitor for 6
months, John Buchanan was caught once again setting the high score for Galaga at the arcade in the mall where he worked. This ended his
janitorial aspirations.
In May of 1983, he counseled a junior student at the University of Windsor that a career in the video game industry was a mistake. That
programmer now writes oracle database code. In the summer of 1985 working as a research assistant he used the
department of Mathematics' computers to write his own shareware game. This game grossed 10$. Unsatisfied with his first attempt to break
into the video game market he proceeded to do his MSc at the University of Toronto under the supervision of Dr. Ron Baecker.
Then a small trip west and north ended in Vancouver where John pursued his PhD under the supervision of the late Dr. Alain Fournier.
In 1993 realizing that he had traveled as far west as he could without swimming he headed back east and north to Edmonton where he spent five
years as an Assistant Professor at the University of Alberta. There he investigated the use of computers to generate Non Photorealistic
Images. After this he took a leave to visit Radical Entertainment, a video game company, in Vancouver. Two months later he joined Electronic Arts
Canada as the senior research scientist of a three person research group. After 4 years of being Director, Advanced Technology for the
Canadian studio and University Research Liaison Officer for the company world wide. He is now full time building research relationships with
Universities. There are darker stories, but these must be pried out of me with copious amounts of beer.
|
 |
 |
 |
 |
May 2003
Tuesday, May 6th
2 p.m.
719 Broadway, 12th Floor Large Conference Room
Title: Shape and Materials by Example: A Photometric Stereo Approach
I will present a technique for computing the geometry of objects with general reflectance properties from images. For surfaces with varying
material properties, a full segmentation into different material types is also computed. It is assumed that the camera viewpoint is fixed, but
the illumination varies over the input sequence. It is also assumed that one or more example objects with similar materials and known
geometry are imaged under the same illumination conditions. Unlike most previous work in shape reconstruction, this technique can handle
objects with arbitrary and spatially-varying BRDFs. Furthermore, the approach works for arbitrary distant and unknown lighting environments.
Finally, almost no calibration is needed, making the approach exceptionally simple to apply.
Joint work with Steve Seitz.
|
 |
 |
 |
 |
May 2003
Thursday, May 8th
3 p.m.
719 Broadway, 12th Floor Small Conference Room
Title: Multi-resolution Simulation Using Adaptive Basis Refinement
Many natural phenomena, such as galaxies, turbulent flows, and folding of
proteins exhibit dynamic behavior at multiple spatial and temporal scales.
Adaptive solvers can reduce the high computational cost of simulating such
phenomena, however implementing such solvers can be a daunting task, especially in three and higher dimensions.
We have developed a framework for adaptive multi-resolution discretization. The benefits of multi-resolution have
already been demonstrated in solvers employing wavelets or hierarchical splines. Our framework unifies these
earlier ideas and generalizes them to a broad setting independent of domain topology, dimension and tesselation, and
approximation accuracy and smoothness. The basic principle of our approach is to refine basis
functions, not mesh elements. This removes a number implementation headaches associated with classical mesh refinement.
We demonstrate the versatility of our new approach through 2D and 3D examples, including medical applications
and thin-shell animations.
Time permitting, I will present ongoing work on a discrete model for thin shells.
|
 |
 |
 |
 |
May 2003
Monday, May 19th
3 p.m.
719 Broadway, 12th Floor Large Conference Room
Title: Control Problems in Computer Animation
The first part of the talk will describe some of our recent efforts in control of both passive and active complex dynamic behaviors. I will
describe a novel framework for control of fluid simulations aimed at keyframing realistic smoke animations. I will also describe an
algorithm for synthesis of realistic bird flight along a specified trajectory. Our framework can generate flight motions for different
birds performing various flight maneuvers including takeoff, cruising, rapid descent, turns, and landing.
In the second part of the talk, I will describe our recent parameterization of the space of all human body shapes, as well as some
interesting explorations of this space.
Time permitting, I will also present an intuitive animation system where complex character motion is created by layering multiple passes of
acting. Since the animator simply acts out all aspects of the motion, the system has a minimal learning curve. Our experiments indicate
that of this system can be effectively used for staging, and rapid prototyping of multi-character animations.
The four results presented in this talk will be published at SIGGRAPH 2003 later this year.
|
 |
 |
 |
 |
May 2003
Monday, June 2nd
1 p.m.
719 Broadway, 12th Floor Large Conference Room
Title: Billboard Clouds for Extreme Model Simplification
We introduce billboard clouds -- a new approach for extreme simplification in the context of real-time rendering. 3D models are
simplified onto a set of planes with texture and transparency maps. We present an optimization approach to build a billboard cloud given a
geometric error threshold. After computing an appropriate density function in plane space, a greedy approach is used to select suitable
representative planes. A good surface approximation is ensured by favoring planes that are ``nearly tangent'' to the model. This method
does not require connectivity information, but instead avoids cracks by projecting primitives onto multiple planes when needed. For extreme
simplification, our approach combines the strengths of mesh decimation and image-based impostors. We demonstrate our technique on a large
class of models, including smooth manifolds and composite objects.
|
 |
 |
 |
 |
June 2003
Monday, July 14, 2003
2 pm
719 Broadway, 12th Floor Large Conference Room
Title: Capturing and Reproducing Photoreal Lighting and Reflectance
ABSTRACT
In this talk I will present a technique for estimating the
spatially-varying reflectance properties of a surface based on its
appearance during a single pass of a linear light
source. By using a linear light rather than a point light source as the illuminant, we
are able to reliably observe and estimate the diffuse color, specular
color, and specular roughness of each point of the surface. The
reflectometry apparatus we use is simple and inexpensive to build,
requiring a single direction of motion for the light
source and a fixed camera viewpoint. Additionally our system records a
per-pixel height map for the object and estimates its per-pixel
translucency. We produce real-time renderings of the captured objects
using a custom hardware shading algorithm. We apply the technique to a
test object exhibiting a variety of materials as well as to an
illuminated manuscript with gold lettering.
I will also present some of our group's recent work in capturing
spatially-varying real-world lighting, an extension of our light probe
techniques, as well as our latest research on the Light Stage project
involving accurately reproducing lighting with complex spectra.
This is joint work with Andrew Gardner, Andreas Wenger, Jonas Unger, Tim
Hawkins, and Chris Tchou. Links to the relevant papers are at:
http://www.ict.usc.edu/~gardner/siggy2003/llsIndex.html
http://www.debevec.org/Research/ILF/
http://www.ict.usc.edu/~wenger/egsr2003/cmlr.ht
About the Speaker:
Paul Debevec received his Ph.D. from UC Berkeley in 1996 where he
worked with C.J. Taylor and Jitendra Malik to produce Facade, an
influential image-based modeling and rendering system for creating
photoreal architectural models from still photographs. His work with
high dynamic range imagery (HDRI) and image-based lighting has been
incorporated into commercial rendering systems such as LightWave and
RenderMan and has influenced recent advancements graphics
hardware. Techniques from Debevec's animations "The Campanile Movie",
"Rendering with Natural Light", and "Fiat Lux" have been used in films
including "The Matrix", "X-Men", and "The Time Machine". In 2001 Paul
received ACM SIGGRAPH's first Significant New Researcher award and in
2002 was named one of the world's top 100 young innovators by MIT's
Technology Review Magazine for his work involving the Light Stage.
Debevec leads the computer graphics research group at USC's Institute
for Creative Technologies and is a Research Assistant Professor in
USC's computer science department.
|
 |
 |
 |
 |
July 2003
Monday, July 21, 2003
3 pm
719 Broadway, 12th Floor Large Conference Room
Title: Integral Imaging (Integral Photography) for Three-Dimensional Display
ABSTRACT
The integral imaging, which used to be called integral photography, is a
quite old technology. But recently it became to attract much attention
because it can realize dynamic three-dimensional color 3D display
without any special glasses. The technology can implement 3D display
with full parallax in both horizontal and vertical directions within
some viewing angle (continuous viewing points, not discrete points). But
typically the integral imaging has some limitations in image depth,
viewing angle and resolution. In this talk, the speaker introduces some
technologies developed in Seoul National University to mitigate some
limitations. The technologies include the variable depth integral
imaging that can realize both real and virtual images, reflection type
integral imaging for projection 3D display, moving-barrier integral
imaging for multiple viewing zones, and polarization-switching integral
imaging for enhanced viewing angle.
About the speaker:
Dr. Lee received his B.S. and M.S. degrees in 1987 and 1989,
respectively, from Seoul National University in Electronics Engineering.
He received his Ph.D. degree in 1993 from University of California at Berkeley in EECS.
In 1994 he joined the School of Electrical Engineering, Seoul National
University as a faculty member, where he is currently an associate professor.
In 1999 his laboratory was honored as a National Research Laboratory (in
holography technologies) by the Ministry of Science and Technology of Korea.
He became a Senior Member of IEEE in 2000 and a Fellow of SPIE in 2002.
In 2002, he received the Young Scientist Award from the President of Korea.
His research fields are three-dimensional display, holography
applications and optical fiber grating devices for sensors and optical
communications. He has published about 100 international journal papers
and about 160 international conference papers including 18 invited papers.
|
 |
 |
 |
 |
September 2003
Wednesday, September 24, 2003
10 am
719 Broadway, 12th Floor Large Conference Room
Title: Mapping conversational social cyberspaces
Abstract: Got threads? Participation in online conversational threads is
a widespread phenomena. Conversational social cyberspaces are
repositories of messages organized into chains of turns and replies.
These spaces are among the most popular aspects of online usage. Online
conversation spaces like email lists, newsgroups, and web boards, are
rich social environments that are increasingly important spaces for
civic discourse. Most conversational social cyberspaces suffer from the
problem of "too much" and are vulnerable to disruption by a small
minority. Sociological studies of these spaces are hampered by the
limits of existing tools and interfaces. Social cyberspaces are in their
early development and still lack many elements of the infrastructure of
physical interaction spaces.
Information about basic social properties of online environments, their
size, activity, and composition of their populations, for example, are
entirely missing or difficult to construct with existing tools. Mutual
awareness is a necessary component of most social institutions. In its
absence many social cyberspaces become noisy conflictual spaces of
limited value. Tools for navigating and evaluating the content found in
social cyberspaces that are based on the analysis of social history can
help support social institutions by encouraging accountability and
highlighting the future value of identity and reputation. Built on
mutual awareness, the resulting institutions may be more resistant to
invasion and disruption. These tools serve the additional function of
providing analysts with empirical data that covers a broad scope of
social cyberspaces, allowing the formation of maps and measures that can
answer basic questions about the dynamics, structure and variation of
these novel interaction environments.
This talk will review recent work from the Microsoft Research Community
Technologies Group, including the Netscan project that implements
social accounting tools for the complete public Usenet newsgroup space.
BIO:
Marc A. Smith
www.research.microsoft.com/~masmith
http://netscan.research.microsoft.com
Marc Smith is a research sociologist at Microsoft Research specializing
in the social organization of online communities. He leads the Community
Technologies Group at MSR. He is the co-editor of _Communities in Cyberspace_ (Routledge), a
collection of essays exploring the ways identity, interaction and social
order develop in online groups.
Smith's research focuses on the ways group dynamics change when they
take place in social cyberspaces. Many groups in cyberspace produce
public goods and organize themselves in the form of a commons (for
related papers see: http://www.research.microsoft.com/~masmith). Smith's
goal is to visualize these social cyberspaces, mapping and measuring
their structure, dynamics and life cycles. He has developed a web
interface http://netscan.research.microsoft.com) to the "Netscan" engine
that allows researchers studying Usenet newsgroups to get reports on the
rates of posting, posters, crossposting, thread length and frequency distributions of activity.
This research offers a means to gather historical data on the
development of social cyberspaces and can be used to highlight the ways
these groups differ from, or are similar to, face-to-face groups. Smith
is applying this work to the development of a generalized community
platform for Microsoft, providing a web based system for groups of all
sizes to discuss and publish their material to the web.
Smith received a B.S. in International Area Studies from Drexel
University in Philadelphia in 1988, an M.Phil. in social theory from
Cambridge University in 1990, and a Ph.D. in Sociology from UCLA in 2001.
|
 |
 |
 |
 |
October 2003
Thursday October 2nd, 2003
1 pm
719 Broadway, 12th Floor Large Conference Room
Title: Robots, MIDI, Pyrotechnics and Assorted Mayhem
Eric Singer will present a history of his work in music, art and
technology. A former researcher at the CAT, Eric has gone on to found
both the Madagascar Institute, a Brooklyn guerilla arts group, and
LEMUR, an art/tech group creating musical robots. Along the way, he
has created numerous projects in the areas of alternative
controllers, interactive performance systems, integrated
music/graphics systems, networked multimedia environments and
computer-controlled pyrotechnics.
|
 |
 |
 |
 |
October 2003
Friday October 17th, 2003
3 pm
719 Broadway, 12th Floor Large Conference Room
Finding Beauty in an Image Stack
Digital photography provides a new freedom over
traditional film. Images are almost free. Pixels
within images are quickly becoming more plentiful. But
aside from increasing the chances that at least one
snapshot of scene will be a good one, few tools have
been built to help the photographer find the beauty
lying in the whole stack of images.
Artists have always found ways in many media to create
beauty and meaning from their observations. This talk
will draw its inspiration from the art world to aid in
the development of a set of new tools for combining
multiple images. These include combining images of one
or more scenes taken from multiple angles, at multiple
times, under different lighting conditions, or
exposures. Newly developed technologies (high dynamic
range, graph cut, Poisson blending) will be discussed
that provide increasing possibilities both for the
artist and for the casual photographer.
Bio:
Michael F. Cohen , Senior Researcher, joined Microsoft
Research in 1994 from Princeton University where he
served on the faculty of Computer Science. Michael
received The 1998 SIGGRAPH Computer Graphics
Achievement Award for his contributions to the
Radiosity method for image synthesis. Dr. Cohen also
served as paper's chair for SIGGRAPH '98.
Michael received his Ph.D. in 1992 from the University
of Utah. He also holds undergraduate degrees in Art
and Civil Engineering from Beloit College and Rutgers
University respectively, and an M.S. in Computer
Graphics from Cornell. Dr. Cohen also served on the
Architecture faculty at Cornell University and was an
adjunct faculty member at the University of Utah. His
work at the University of Utah focused on spacetime
control for linked figure animation. He is perhaps
better known for his work on the radiosity method for
realistic image synthesis as discussed in his recent
book "Radiosity and Image Synthesis" (co-authored by
John R. Wallace). Michael has published and presented
his work internationally in these areas.
At Microsoft, Dr. Cohen has worked on a number of
projects ranging from image based rendering, to
animation, to camera control, to more artistic
non-photorealistic rendering. One project focuses on
the problem of image based rendering; capturing the
complete flow of light from an object for later
rendering from arbitrary vantage points. This work,
dubbed "The Lumigraph" is analogous to creating a
digital hologram. He has since extended this work
through the construction of "Layered Depth Images"
that allow manipulation on a PC. Michael also is
continuing his work on linked figure animation. He and
colleagues have focusing on means to allow simulated
creatures to portray their emotional state, and to
automatically transition between verbs. Recent work
also includes creating new methods for
non-photorealistic rendering, low bandwidth
teleconferencing, technologies for combining a set of
"image stacks", as well as developing new approaches
to the low level stereo vision.
His work can be found at:
http://www.research.microsoft.com/~cohen
|
 |
 |
 |
 |
November 2003
Friday, 11/21, 2003
1 pm
719 Broadway, 12th Floor *SMALL* Conference Room
President, FullView
Title-
Panoramic Cameras: Designs and Applications
Panoramic photographs have long held a fascination for both
photographers and the public at large, this fascination dating almost
to the very invention of photography in 1827.
Of late, this interest has been reinvigorated by three technological
trends: ever improving digital imaging, which allows the rapid capture
of high-quality digital images; ever improving digital computing,
which allows the rapid creation of and manipulation of digital
panoramic images; and, finally, ever improving digital communication,
which allows the rapid transmittal of digital panoramic images, in
whole or in part, between different physical locations.
Although the optics of the various techniques currently used for
panoramic photography are mature, advances in digital imaging,
computing, and communication have made possible implementations and
applications that were hitherto infeasible. I shall discuss both.
|
 |
 |
 |
 |
November 2003
Monday, November 24th, 2003
12:00 noon
719 Broadway, 12th Floor Large Conference Room
Title:
Phenotropics, or Prospects for Protocol-adverse Computing
Abstract:
Whenever we pass a variable to a function, or send a message to an object,
we're simulating the sending of pulses down a wire. The way that works is
the sender and receiver agree in advance on a format that makes the pulses
interpretable, also known as a protocol. Protocols aren't the only way
information can travel between places, however. When a physical coffee mug
sits on a table, it's possible to imagine that there's a protocol that
exists between the two things, but it's an awkward way to think. And yet
that's what we often do when we try to build scalable simulations of the
world. We can end up with a "coffee mug module" connected to a "table
module" via a protocol. In the early years of computing, many researchers
wished that the world was a little more like a protocol, so that would be
easier to interface computers to it. Early natural language researchers,
for instance, were unhappy to find that it wasn't so. What happened
instead was that processors eventually became powerful enough to run
pattern classification algorithms that could gather information even
though the world didn't agree with us in advance on a format. Some
examples are face recognition and feature tracking, voice recognition,
and scene understanding. The idea of phenotropics is to use similar pattern
recognition techniques to connect software modules together inside the
computer. Hopefully systems built in that way will display more
informative failure modes, and therefore be more amenable to adaptive
improvement. Another potential benefit is that scientific simulations
might not be distorted by protocols (as in the example of the coffee mug
on the table), and might be more easily integrated into a new
iteration of the scientific method in which they could be usefully published,
tested, and reused.
BIO:
Jaron Lanier is a computer scientist best known for coining the term
"Virtual Reality". He is currently Visiting Scientist at SGI, and
recently served as Chief Scientist for Advanced Network and Services, the
engineering office of Internet2, a coalition of American research
universities sharing an experimental next generation network. While there,
he lead the Nation Tele-immersion Initiative, which was responsible
for providing the "driver" applications for Internet2.
www.jaronlanier.com
|
 |
 |
 |
 |
For additional information, contact: info@cat.nyu.edu
|
 |
|
 |
|