SIGGRAPH Authors Seminar Series at I2R
Seminar 10
Title: Matting
and Compositing of Transparent and Refractive Objects
Speaker: Dr. Sai-Kit Yeung, Singapore University of Technology and Design
Chaired by: Dr Huang Zhiyong
Venue: Franklin @ 11S, I2R, Fusionopolis
Time: 14:00-14:40, 16 April 2012, Monday
Abstract
We present a new approach for matting and compositing transparent
and refractive objects in photographs. The key to our work is an image-based
matting model, termed the attenuation-refraction matte (ARM) that encodes
plausible refractive properties of a transparent object along with its observed
specularities and transmissive
properties. We show that an object's ARM can be extracted directly from a
photograph using simple user markup. Once extracted, the ARM is used to paste
the object onto a new background with a variety of effects, including compound
compositing, Fresnel effect, scene depth, and even caustic shadows.
User-studies find our results favorable to those obtained with Photoshop as
well as perceptually valid in most cases. Our approach allows photo-editing of
transparent and refractive objects in a manner that produces realistic effects
previously only possible via 3D models or environment matting. This work was
presented at SIGGRAPH 2011 in Vancouver.
Bio-data
Dr. Sai-Kit Yeung is currently an Assistant Professor at the Singapore
University of Technology and Design (SUTD). Before joining SUTD, he had been a
Postdoctoral Scholar in the Department of Mathematics, University of
California, Los Angeles (UCLA) in 2010, a Postdoctoral Fellow at the Hong Kong
University of Science and Technology (HKUST) in 2009, a visiting student at the
Image Processing Research Group at UCLA in 2008, and on an overseas PhD
scholarship as a visiting scholar at the Image Sciences Institute, University
Medical Center Utrecht, the Netherlands in 2007. He is an HKUST alumni, graduated with a BEng
degree (First Class Honors) in Computer Engineering in 2003. He received his MPhil degree in Bioengineering from HKUST in 2005, and
completed his PhD degree in Electronic and Computer Engineering from HKUST in
2009.
Seminar 9
Title: Sampling
and Reconstruction of High-Dimensional Visual Appearance
Speaker: Prof Ravi Ramamoorthi, University of California, Berkeley
Chaired
by: Dr Ng Tian
Tsong
Venue:
Time: 10:00-11:00, 25 Jan 2011, Tuesday
Abstract
In this talk, we describe new approaches to many
current and classic problems in computer graphics. This includes the challenge
of creating realistic images at interactive rates, for real-time rendering applications
like games and virtual design applications.
Another important component of visual realism is the realistic modeling
of surface appearance, for objects like clothing, or phenomena like smoke. In
both domains, data-driven methods are increasingly being used, wherein the
properties of a real scene are measured, or the properties of a virtual scene
are simulated, and then reused to create new imagery. Even classical computer
graphics rendering can be viewed in this light, wherein one is making use of
samples of the virtual scene, corresponding to a particular point in space or
an image pixel, and moment in time.
A key challenge in data-driven visual appearance is its high dimensionality.
For example, the appearance of a human face requires understanding the
variation across the surface, all lighting directions, and viewpoints that can
be a 6D space. Similar high-dimensional spaces arise in real-time and offline
rendering, as well as in image acquisition applications. The challenge of
sampling and reconstructing these high-dimensional datasets has been a major
obstacle in computer graphics. In this talk we detail a research program that
seeks to develop new sparser sampling strategies, coupled with novel signal
processing tools for reconstruction. We describe examples in three different
areas: real-time rendering of area lighting, appearance acquisition of
volumetric media, and Monte Carlo rendering of motion blur. I will also briefly
discuss recent efforts in the imaging domain. These examples indicate the
potential of a broad program across computer graphics for fundamentally new
strategies to acquire and exploit visual appearance data.
Bio-data
Ravi Ramamoorthi has been an
associate professor of electrical engineering and computer science at the
University of California, Berkeley, since January 2009. Earlier, he was on the
faculty of the computer science department at Columbia University, which he
joined after receiving his PhD from Stanford University in 2002. His research
interests span many areas of computer graphics rendering and appearance, as
well as related problems in physics-based computer vision. His focus has been
on developing new computational models and signal-processing methods to
understand and make use of complex visual appearance. His work has been
recognized with many honors, including the ACM SIGGRAPH Significant New
Researcher Award, and the white house's Presidential Early Career Award for
Scientists and Engineers, as well as young investigator awards from NSF, ONR
and the Sloan foundation.
Seminar 8
Title: Dynamic Models for Character Animation
Speaker: Dr Yin KangKang,
Chaired by: Dr Huang Zhiyong
Venue:
Time: 14:30-15:10, October 1, Friday, 2010
Abstract
This talk presents three examples of our efforts to
develop more sophisticated dynamic models of human motion. We first show short demos
of a remarkably simple and effective balance mechanism for bipeds, and
continuation methods to generalize basic locomotion skills to more challenging
tasks. We then present our latest work on sampling-based motion control. We
demonstrate fast control reconstruction for a diverse set of captured motions,
including walking, running, and contact-rich tasks such as sideways rolling and
kip-up jumps. The proposed method can also generate physically plausible motion
variations, and perform physically based motion transformation and retargeting.
In addition, we show that sampling is effective for reference-trajectory-free
scenarios, such as idling.
Bio-data
KangKang Yin obtained her PhD from the
Seminar 7
Title: K-set tilable surfaces
Speaker: Dr Fu Chi
Wing, Nanyang Technological University
Chaired by: Dr Huang Zhiyong
Venue:
Time: 14:00-14:40, June 2, Wednesday, 2010
Abstract
In this talk, I will introduce a geometric optimizing
method for tiling on a quad-mesh. Given a quad-based surface, our goal is to
generate a set of K quads whose instances can produce a tiled surface that
approximates the input surface. This research work proposes the K-set tilable surface, which can lead to an effective cost
reduction in the physical construction of the given surface. Rather than
mildewing lots of different building blocks, a K-set tilable
surface requires the construction of K prefabricated components only. At the
end of this talk, I will demonstrate K-set tilable
surfaces on various surfaces, including some that mimic the exteriors of
certain renowned building landmarks.
Bio-data
Chi-Wing FU received his B.Sc. and M.Phil.
degrees in Computer Science and Engineering from the Chinese University of Hong
Kong in 1997 and 1999, respectively, and later his PhD degree in Computer
Science from Indiana University in Bloomington in December, 2003. He held
previous visiting assistant professor positions in Hong Kong University of
Science and Technology and joined School of Computer Engineering in Nanyang Technological University from 2008 as an assistant
professor. His research interests include computer graphics, visualization, and
human-computer interaction.
Seminar 6
Title: Research in Visual Computing
Speaker: Prof Heng Pheng Ann, The
Chaired by: Dr Huang Zhiyong
Venue: Resonance @ 13N, Fusionopolis
Time: 10:30-11:30, November 5, Wednesday, 2008
Abstract
The Virtual Reality, Visualization and
Bio-data
Dr. Heng received his B.Sc in computer science in 1985 from the National
University of Singapore. He received his M.Sc in
computer science, M.A. in applied mathematics, and Ph.D. in computer science
from
He has served as the Director of Virtual Reality,
Visualization and Imaging Research Centre at CUHK since 1999 and as the
Director of Centre for Human-Computer Interaction at Shenzhen Institute of
Advanced Integration Technology,
His current research interests include virtual reality
applications in medicine, visualization, medical imaging, human-computer
interaction, and computer graphics.
Seminar 5
Title: Appearance Manifolds for Modeling Time-Variant Appearance of Materials
Speaker: Dr
Steve Lin, Leader Researcher, Microsoft Research
Chaired by: Dr Huang Zhiyong
Venue: Three star Theatrette
Time: 11-11:40, May 16, Friday, 2008
Abstract
We present a visual simulation technique called
appearance manifolds for modeling the time-variant surface appearance of a material
from data captured at a single instant in time. In modeling time-variant
appearance, our method takes advantage of the key observation that concurrent
variations in appearance over a surface represent different degrees of
weathering. By reorganizing these various appearances in a manner that reveals
their relative order with respect to weathering degree, our method infers
spatial and temporal appearance properties of the material’s weathering process
that can be used to convincingly generate its weathered appearance at different
points in time. Results with natural non-linear reflectance variations are
demonstrated in applications such as visual simulation of weathering on 3D
models, increasing and decreasing the weathering of real objects, and material
transfer with weathering effects.
Bio-data
Steve joined Microsoft Research Asia in
June 2000, and is currently a Lead Researcher in the Internet Graphics group.
His research interests lie in the fields of computer
vision and computer graphics. In computer vision, his primary research areas
are photometric analysis and low-level vision. His interests in computer graphics
include reflectance modeling and inverse rendering. He received a B.S.E. in
electrical engineering from
Seminar 4
Title: MoXi Digital Paint
Speaker: Dr. Nelson
Chu, HKUST, Visiting Fellow, NTU
Chaired by: Dr Huang Zhiyong
Venue: Big-One
Time: 11-11:40, April 25, Friday, 2008
Abstract
Chinese painting and calligraphy are among
the oldest continuous art traditions in the world. The expressive brush strokes
and the fascinating ink dispersion contribute a lot to the universal appeal. In
this talk, I will first outline our physically-based methods to model brush
dynamics and ink dispersion. Our goals are to bring the distinct charm of ink
painting and calligraphy to the digital art scene and to further develop the
art traditions. The second part of the talk would be a brief discussion on our
collaboration with the industry, followed by live demo.
Bio-data
Nelson Chu is both a visual artist and a
software engineer. From 2001 to 2007, he focused on the research and
development of a novel digital paint system, which redefined "natural-media
painting" in the field of Computer Graphics. The resultant system
attracted industrial giants Adobe and Sony, who licensed the technology in 2006
and 2007 respectively. Nelson was born in Hong Kong and raised in
Seminar 3
Title: ShapePalettes: a novel approach for 3D markup
Speaker: Dr
Michael S. Brown, Sung Kah Kay Assistant
Professor, SOC, NUS
Chaired by: Dr Huang Zhiyong
Venue: Three star Theatrette
Time: 11:30-12:10, March 28, Friday, 2008
Abstract
This talk overviews a simple interactive
approach to specify 3D shape in a single view using "shape palettes".
The interaction is as follows: draw a 2D primitive in the 2D view and then
specify its 3D orientation by drawing a corresponding primitive on a shape
palette. The shape palette is presented as an image of some familiar shape
whose local 3D orientation is readily understood and can be easily marked over.
The 3D orientation from the shape palette is transferred to the 2D primitive
based on the markup -- only sparse markup is needed to generate expressive and
detailed 3D surfaces. This markup approach can be used to model freehand 3D
surfaces drawn in a single view, or combined with image-snapping tools to
quickly extract surfaces from images and photographs.
The talk will be followed by a short discussion on how the ShapePalette
idea emerged and its road to acceptance as a full paper at SIGGRAPH'07.
Bio-data
Michael S. Brown received his BS (1995)
and PhD (2001) from the
Seminar 2
Title: Image-based Tree Modeling
Speaker: Dr
Tan Ping, Assistant Professor, ECE, NUS
Chaired by: Dr Huang Zhiyong
Venue: Big-One
Time: 11-11:40, March 7, Friday, 2008
Abstract
In this
talk, we present a technique for generating 3D texture mapped tree models from
images. From these images, a set of 3D points and camera poses are computed
with existing technique. Our method will compute a texture mapped triangle mesh
model from these recovered 3D points and images. To faithfully model different
trees with large and small leaves, we designed different approaches. For trees
with relatively large leaves, segmentation is performed in both image and 3D
spaces. Using the segmented image and 3D data, the geometry of each individual
leaf is then automatically recovered from the multiple views by fitting a
deformable generic leaf model. For trees with relatively small leaves, we do
not model each leaf directly from images due to the large leaf count, small
image footprint, and widespread occlusions. Instead, we populate the tree with
leaf replicas from segmented source images to reconstruct the overall tree
shape. In addition, we use the shape patterns of visible branches to predict
those of obscured branches. We demonstrate our approach on a variety of trees.
Bio-data
Ping Tan received the B.S. degree in
Applied Mathematics from
Seminar 1
Title: Towards
High Quality 3D Modeling
Speaker: Dr Zheng
Jianmin, Assistant Professor, SCE, NTU
Chaired by: Dr Susanto Rahardja,
Director and Department Head
Venue: Big-One
Time: 5-5:40, February 26, Tuesday, 2008
Abstract
As an important component of digital 3D content,
geometric models are nowadays more and more pervasive, from traditional
engineering applications such as computer-aided design and manufacture
(CAD/CAM), robotics and physical simulations to multimedia applications 
including e-commerce, cultural heritage, 3D games, animation and special
effects in motion pictures. However, creating and processing such 3D models are
generally labor-intensive and time-consuming, especially when the shapes are geometrically
and topologically complex. Therefore new theoretical insights and practical
algorithms for efficient, flexible and intuitive 3D shape modeling and
processing are highly required. In this talk, I will share some of my thoughts
on high quality 3D modeling and present some of our research work in this area.
Bio-data
Jianmin Zheng
is an assistant professor in the