Course material for a grad-level course in geometry processing.
Course designed by Prof. Alec Jacobson, University of Toronto, with assistance from Klint Qinami and Prof. Eitan Grinspun, Columbia University.
In general, the materials in this course assume that students should have already taken Linear Algebra and Calculus.
Students should have already taken Introduction to Computer Science and should be proficient in computer programming (in any language) and should feel comfortable programming in C++.
All course assignments are conducted in C++, however none rely on nitty gritty memory management or complicated object-oriented data-structures.
While knowledge of Partial Differential Equations is not required, it will certainly be very handy for derivations. Similarly, previous experience with Computer Graphics is not required but recommended.
All coding assignments make use of Eigen, an open-source linear algebra library; and libigl, an open-source geometry processing library. Each assigment is built using CMake.
The original run of this course structured weekly assignments in the following order:
- Introduction
- Mesh Reconstruction
- Surface Registration
- Smoothing
- Parameterization
- Deformation
- Curvature
Besides the introduction, there is no strict ordering to these topics.
Each topic has its own git repository. Inside each, there is a README.md
file
contains background information necessary for understanding the topic's coding
assignment.
The
Introduction
README.md
contains detailed information about compilation, file layout and
assignment protocols.
The background materials link heavily to Wikipedia articles. Sometimes the wikipedia articles relating to geomtry processing are less informative than they could be. Edit them!
In university offerings of this course, 5% credit has been awarded to the entire class for collaboratively improving Wikipedia's entries on geometry processing topics.
There are instructor repositories for all of the assignments above. If you're an instructor for a geometry processing course, send an email to [email protected] for an invitation.
For my reference, I can add a new instructor with github id [githubid] to all solution repos using:
github-add-user -u alecjacobson -r $(echo alecjacobson/geometry-processing-{introduction,mesh-reconstruction,registration,smoothing,parameterization,deformation,curvature}-solution | tr ' ' ',') [githubid]
- Introduction
- Mesh Reconstruction
- Surface Registration
- Smoothing
- Parameterization
- Deformation
- Curvature
When used for a formal course, it is intended that students fork each assignments repository, commit their solutions to their own forks, and then submit their assignment via pull request to the public repo of the assignment.
More details on this structure are found on Alec's weblog.
Since pull requests are public, students will be able to see each other's completed solutions as soon as their posted. Students will not cheat because they are honorable 👍. Actually, since most of the assignments are standard algorithms there are many implmentations online already. In particular, libigl contains ready-made implementations using Eigen with similar function APIs. Why cheat from other students when you can cheat from the professor? But, really, why cheat?