SGP 2021 Virtual Field Trip Report
I attended this year’s Symposium on Geometry Processing (SGP), one of the premier conferences for publishing cutting-edge research in geometry processing. Here is my brief resume.
The conference was scheduled to take place in Toronto, Canada. However, due to current circumstances, the conference was virtual only. In any case, the organizers did an awesome job organizing a top notch conference and making it a highly worthwhile experience.
Positive side-effect of the virtual conference: All the graduate school lectures, technical papers, and keynotes are available online. Check out the program, there’s a whole bunch of interesting lectures and presentations. The graduate school lectures from previous years are archived online as well.
Graduate School Highlights
The graduate school is one of the most valuable aspects of SGP. It enables practitioners to stay up to date with the state of the art and to expand their knowledge in concentrated manner.
I particularly enjoyed the following lectures:
- Maps Between Surfaces by Marcel Campen and Patrick Schmidt. Awesome presentation, great content, very instructive. I learned a lot from it, especially since this is a somewhat more recent area of research.
- Shape Approximation and Applications by Tamy Boubekeur. A great overview on shape approximation techniques, including a classification of different methods and their development over time, as well as some recent research highlights. Well done and entertaining presentation definitely worth watching.
- Digital Geometry by David Coeurjolly and Jacques-Oliver Lachaud. Geometry processing on voxel representations! Since I’m dealing a lot with voxel-based modeling in my day job, this was of particular interest to me.
Paper Highlights
The main part of the conference were three packed days of technical paper presentations. Quite a marathon! Zoom fatigue kicked in from time to time, but I kept on watching (almost) all presentations.
Best Paper Awards
There were plenty of interesting papers, most notably those winning this year’s best paper awards:
- Surface Map Homology Inference
- Geodesic Distance Computation via Virtual Source Propagation
- Simulated Jet Engine Bracket Dataset
Papers of Interest
Beyond the above, there were a couple of papers I found very interesting. Without particular order:
- Delaunay Meshing and Repairing of Defect-laden NURBS Models
- Globally Injective Geometry Optimization with Non-Injective Steps
- Simpler Quad Layouts using Relaxed Singularities
- The Diamond Laplace for Polygonal and Polyhedral Meshes
- Frame Field Operators
- On Landmark Distances in Polygons
Test of Time Award
This is a new type of award to recognize papers that have stood the test of time, i.e., papers which are widely recognized throughout the community even after more than ten years. Great idea to highlight research that really sticks and has a long-lasting impact.
This first iteration of the award went to Poisson Surface Reconstruction. That’s certainly a well deserved recognition. The method and the corresponding software has become the de-facto standard for surface reconstruction.
Keynotes
We had three keynote speakers, each offering their unique insights:
- Bradley Rothenberg, CEO, nTopology: “Engineering-driven design: a new foundation”
- Geoffrey Hinton, University of Toronto/Google Research: “How to represent part-whole hierarchies in a neural net”
- Lining Yao, Carnegie Mellon University: “Computing Morphing Matter: the Marriage of Geometry and Hidden Forces”
See the full descriptions for details.
Enjoy the show!