Time: 1pm, June 29th
Title: 3D Content Generation: from Inspired Modeling to Creative Modeling
As a computer graphics researcher by training, I once pondered what computer graphics
should really be about as I always wanted it to be much more than just photorealistic
image synthesis. The first big revelation for me came at the SIGGRAPH 2011 Award
Talk by Jim Kajiya. To keep some suspense before the talk, let me only say that Kajiya's
talk highlighted the importance of content generation, especially of 3D content.
Fast forward to 2021, an era dominated by AI-powered computing, I will show you why
"learning to generate" is at the heart of AI, by unlocking a connection between the
famous Turing Test and the not so well-known Lovelace Test. I would argue that the
ultimate goal of intelligent content generation is originality or creativity.
However, learning to generate creative content is a tall order. In my talk, I will
first take a step back and focus on *inspired* 3D modelling, where the inspiration
can come from a photograph, a sketch, natural language, or a set of exemplars. I
will go over a small sampler of my research on these, highlighting unique challenges
associated with learning to generate *3D* content, how they can be addressed, while
demonstrating the importance of *structured* 3D representations.
Next, I shall tap into computational creativity, showing you its three-tiered goals,
and presenting several of our attempts at creative 3D modelling, both success stories
and on-going pursuits, which include genetic programming, co-creation, and multi-hop
generative adversarial networks. If time permits, I will raise several questions
related to creative content generation to motivate future research.
Hao (Richard) Zhang is a full professor in the School of Computing Science at Simon Fraser University (SFU), Canada, and holds a Distinguished University Professorship (2020-25). He has also been a visitor professor at Stanford University, Shenzhen University, and the Beijing Film Academy. Richard obtained his Ph.D. from the University of Toronto, and MMath and BMath degrees from the University of Waterloo, all in computer science. He directs the GrUVi (Graphics U Vision) lab at SFU. His research is in computer graphics and visual computing with special interests in geometric modeling, shape analysis, 3D vision, geometric deep learning, as well as computational design and fabrication. He has published more than 170 papers on these topics, including 58 articles in SIGGRAPH, SIGGRAPH Asia, and ACM Trans. on Graphics, the most prestigious venue in computer graphics. Methods from three of his papers on geometry processing have been adopted by CGAL, the open-source Computational Geometry Algorithms Library.