Hsueh-Ti Derek Liu is the recipient of the 2022 Alain Fournier Award for Outstanding Doctoral Dissertation in Computer Graphics. Dr. Liu’s dissertation, titled Algorithms for Data-Driven Geometric Stylization & Acceleration, made outstanding contributions to the field of computer graphics.
Dr. Liu’s dissertation addresses scalable 3D content creation from the angle of deep learning while paying attention to expressiveness. Among his numerous contributions, the first chapters of his dissertation introduce tools to enable image-based editing of 3D shapes. Dr. Liu achieved this by inverting the process of rendering a 3D shape into an image. Stylization in the image domain can be propagated back to the 3D shape. In later chapters, he shows how to learn stylizations for 3D shapes directly from 3D data. While classic subdivision surfaces can be viewed as a smoothness prior, Dr. Liu describes how to train non-linear subdivision operators trained from a self-supervised data prior. Dr. Liu’s machine learning work is buttressed by his contributions to geometry processing, including a robust method for maintaining mathematical bijection before/after simplifications, the key ingredient to creating datasets of low and high resolution 3D shapes. His differentiable rendering approach has also served to the understanding of adversarial attacks in the field of machine learning itself. Finally, Dr. Liu’s dissertation also develops coarsening operators to maintain near-perfectly spectral properties on meshes. There have also been several follow-up projects where he acted as a mentor for other students.
Dr. Liu’s dissertation comprises eight top-tier publications, including six presented at ACM SIGGRAPH and one at the International Conference on Learning Representations (ICLR). The committee felt that Dr. Liu was able to develop a coherent and rigorous narrative for his dissertation, each chapter standing on its own, yet the whole dissertation demonstrating a unifying class of geometry-processing solutions. The committee firmly believes his dissertation will serve as a critical record of his research methodology for future researchers and a springboard to further advances in this field.
Dr. Liu obtained his Bachelor in Engineering Science and Ocean Engineering at the National Taiwan University in 2014 and his Master in Mechanical Engineering at Carnegie Mellon University in 2017. He graduated in 2022 with his Ph.D. in Computer Science from the University of Toronto under the supervision of Prof. Alec Jacobson. During his Ph.D. studies, he also worked as a research intern at Ecole Polytechnique in Palaiseau, Adobe Research in Seattle, NVIDIA AI in Toronto, and consulted for Urus Entertainment. He received fellowships and awards from Okino, Adobe, Mary H. Beatty, and Mitacs Globalink. While early in his career, he has already been involved on conference program committees for Eurographics, Pacific Graphics, Shape Modelling International, and Graphics Interface.