A part of the course will survey several
topics in geometric modeling, concentrating
on computer representations of surfaces used in computer graphics applications.
The following topics will be covered: Spline curves and surfaces,
subdivision.
In the second part we will discuss applications of machine learning
in computer graphics.
Many things that we would like to render or animate can be extremely difficult or impossible to model explicitly, such as the mapping from images to 3D
models or creativity of a talented artist. Rather than attempt to engineer such complex models, it is often more effective to automatically learn the model
from data. This section of the course will give an overview of some of the ways that learning can be used in computer graphics.
Prerequisites
Mathematics: linear algebra, calculus.
Computer Science: solid programming ability in at least one
language (Java, C++, C). An introductory graduate class in computer graphics
is recommended.
Requirements
The grade
will be primarily based on the course project. A list of suggested
projects will be provided;
however, students may choose their own project. In either case,
the intended project should be discussed with the instructor.
Deadlines
October 10 A detailed description of the proposed project.
November 11 Progress report.
Finals week Projects due.
Outline
Geometry
Spline curves and surfaces
Subdivision curves and surfaces
Multiresolution representations of surfaces.
Mesh simplification.
Mesh reparameterization.
Machine learning for Computer Graphics
Texture synthesis. Image and animation texture.
Basic probability and statistics. Bayes' rule, Gaussian densities, PCA. Application to facial modeling.
September 12: Geometry overview. Machine learning for CG overview September 19: Geometry. September 26: Probabilities, Bayes' rule, Gaussian density
estimation, PCA, Gaussian mixture models, facial modeling. October 3: Splines. October 10: Subdivision. October 17: Subdivision parameterization and evaluation.
Texture
synthesis. Project ideas. October 24: October 31: November 7: November 14: November 21: November 28: Hidden Markov Models December 5: HMMs for Animation. Motion texture. December 12: