Basic Information Current Project Recent Publications CSC Funded NCCA PhD
Retargeting Animated Human Characters

Retargeting Animated Human Characters

Realistic and visually accurate animation is still a labour intensive
task. In this project, we propose a set of techniques to reuse
character animation by transfering it to different human characters.
We have developed a novel anatomy-based technique to retarget animation from one humanoid character to another. Our approach significantly reduces production time by allowing animators to
transfer existing animations onto new models automatically.

Hyper-Twist: Form of Art Based on Deformation Tequniques

Hyper-Twist: Form of Art Based on Deformation Tequniques

Hyper-twist is a distortion technique for the production of abstract artistic
forms. It is developed by deforming an infinite space filled with a hyper-
elastic media. Any object placed within this space can change its shape as
a result of the distortion of the space. By linking this representation with
time, temporal effects can be produced. Our work incorporates physics and
digital creation, and represents a new form of algorithmic art.

Capturing Dynamics with Cosserat Rods

Capturing Dynamics with Cosserat Rods

The Cosserat Rod model is applied to an arbitrary three-dimensi-
onal object for fast simulation of its vibration behaviours. This
method offers realism traditionally only afforded by the finite
element methods while providing more sophistication than the
mass-spring method.

Animation of Car Crashes

Animation of Car Crashes

In this project we investigate the deformation of engines to efficiently
model and render the damaged structure of vehicles in crash scenarios.
This is particularly challenging as it requires the simulation of large plastic deformations which result from collisions.

Human Muscular-skeleton Structure Modelling

Human Muscular-skeleton Structure Modelling

Graphical simulations of human muscle motion and deformation
are of great interest in artistic and medical applications. This
project combines physics and geometry to produce fast and
accurate simulations of muscle deformations.

Virtual Surgery Simulation

Virtual Surgery Simulation

This project aims to develop a framework of modelling biological tissues
at both global and local scales. The most essential element of a virtual
surgery training system is to model the behaviour of the tissues realistically
in order to produce convincing visual and haptic feedback to the user.

The Bystander Project

The Bystander Project

The Bystander Project is a long term project exploring the effect
of visual fidelity on the bystander phenomenon with respect to
violent emergencies in immersive virtual environments. This is a collaborative EPSRC funded project, involving researchers from University College of London, University of Lancaster and Bournemouth University.

PDE-based geometric modelling and computer animation

PDE-based geometric modelling and computer animation

Partial differential equation (PDE) based modelling have some unique
advantages over other geometric modelling techniques. Our research team
has been researching this field for many years, and have developed various techniques for modeling curves, surfaces and solids. Our group has produced
a significant number of high quality publications in this field.

High-fidelity motion synthesis for humanoid characters

High-fidelity motion synthesis for humanoid characters

Motion is a fundamental and active research area in computer
animation. Generating high-fidelity motions of human characters
is time consuming and expensive. The focus of this project is on
the synthesis of physically feasible motion data for animated characters.

Simulation of human hand grasping on arbitrary

Simulation of human hand grasping on arbitrary
geometric shapes

Grasping objects is as much a physical and visual task as it is a cognitive
task. The goal of this project is to implement a grasp simulation system by
looking into various aspects of the grasping process. We plan to implement
an easy- to-use grasping simulation technique for animation, computer games
and other graphical applications.

Real-time facial expression and animation

Real-time facial expression and animation

Facial expressions are probably the most important visual method
for conveying emotion, and is an important method of non-verbal
communication between humans. They are thus very important for
understanding human behaviour, the perceptual user interface and
interactive graphical applications. The objective of this research is
to develop an animator-controlled facial animation technique
allowing various realistic expressions to be produced easily.

Physics-based skin deformation for computer animated characters

Physics-based skin deformation for computer animated characters

Realistic skin deformation is essential for the animation of human and
creature characters. In this project our objective is to empower the animator
by giving them overall control of the skin deformation modelling process
while reducing unnecessary manual preparation. This project has been
reported by many news organisations, including NewScientist, MIT
Technology Review and LiveScience.

Physics-based deformation modelling for animated objects

Physics-based deformation modelling for animated objects

Physics-based deformation modelling has attracted a great deal of attention because it enhances the realism in animated scenes. Our research team has investigated this problem and developed one of
the first mesh-free deformation techniques for computer animation. The objective is to reduce the amount of unnecessary human work
and improve the level of realism in animating deformable objects.

 

Foldover-free Constrained Texture Mapping

 Texture mapping has long been used in computer graphics to enhance the realism of virtual scenes. However, in order to match the 3D model feature points with the corresponding pixels in a texture image, a surface parameterization must satisfy specific positional constraints. Despite much existing related work, the construction of a mathematically robust, foldover-free parameterization that satisfies positional constraints continues to be a challenge. We address this foldover problem by developing a smooth constrained mesh-free radial basis function (RBF)-based reparameterization.