|
This project actually consists of two parts that can be
done together or in parts depending on the student(s). Optimally a
physics student and a computing science student could work together on
it.
The first part is an implementation of the Smoothed
Particle Hydrodynamics method for simulation of fluids or other complex
materials, while the second part focuses on the real-time rendering of
these materials, i.e. using state -of-the-art computer graphics
hardware to realistically visualize e.g. a fluid.
Smoothed Particle
Hydrodynamics (SPH)
The SPH method utlizes a novel and innovative form of
local field approximation to differentials and functions making
it possible to solve a differential field equation using an intuitive
and extensible particle method. One could see it as if the particles
span a dynamic mesh which is used to discretize and solve the continuum
equations, but there is actually more to the method than this. The core
of the method is the use of a smoothing
kernel, i.e. a function with certain properties that is used to
sum up an approximative value of the field functions from the
surrounding particles. One part of the project will be to optimize the
use of these smoothing kernels for the specific terms of interest, i.e.
elasticity and plasticity.
Mesh-free methods are generally gaining in popularity
since they are both computationally and numerically efficient, and also
easy to implement and use. Applications lie in e.g. simulation of
fluids (water, blood, etc), soft matter (e.g. skin, organs) and
granular materials (e.g. sand and mud). All of these applications have
considerable industrial impact too, e.g. in the game industry, movie
special effects, medical training simulators, vehicle simulators, etc.
Possibly the impact will be quite big on future computer games -
imagine a character that can change it's form and structure from solid
to fluid a´la T2 and Terminatrix.
For further information, a full specification and
references, contact Kenneth
Holmlund.
Real-time visualization
of complex materials
Optimally, this project is directly connected to the SPH
project, but it can also be done separately, by focusing on physically
based surface reconstruction and hardware based rendering of particle
represented dynamic surfaces. Modern GPU shader programs will be used
and algorithms will be based on physical principles. Real-time soft
shadows of complex materials could be one possible focus of this
project.
For further information, a full specification and
references, contact Kenneth
Holmlund.
|
