Jiayi(Eris) Zhang

Jiayi(Eris) Zhang

Undergrad @UofT
she/her/hers

About me

Hi! I’m a senior undergraduate student at the University of Toronto, majoring in Computer Science and Mathematics. In parallel, I’ve been working as a research assistant at the Dynamic Graphics Project Lab, advised by Prof. Alec Jacobson. Currently I’m doing a summer internship at Adobe Research under Dr. Qi Sun and Dr. Xin Sun.

I’m broadly interested in physics-based animation, geometry processing, computational fabrication, interactive graphics and novel interfaces that augment creativity and productivity.

Previously, I’ve explored different research areas with Prof. Ken Jackson on numerical analysis and with Prof. Fanny Chevalier on HCI / visualization. I also did research in geometric modeling with Prof. Marc Alexa and physics-based animation with Prof. David Levin and Prof. Alec Jacobson. It’s my great honour to be selected as a finalist for CRA Outstanding Undergraduate Researcher Award 2020 and an awardee for Adobe Research Women-in Technology Scholarship 2020.


Publications

  Fast Updates for Least-Squares Rotational Alignment
  Jiayi Eris Zhang, Alec Jacobson, Marc Alexa
  In Submission


  Complementary Dynamics
  Jiayi Eris Zhang, Seungbae Bang, David I.W. Levin, Alec Jacobson
  ACM SIGGRAPH ASIA 2020
  Paper | Paper (low res) | Project Page
  *Featured in the Technical Papers Trailer


  DataQuilt: Extracting Visual Elements from Images to Craft Pictorial
  Visualizations
  Jiayi Eris Zhang, Nicole Sultanum, Anastasia Bezerianos, Fanny Chevalier
  ACM Conference on Human Factors in Computing Systems (CHI), 2020
  Paper | Project Page


Selected Projects

Fast Updates for Least-Squares Rotational Alignment
Across computer graphics, vision, robotics and simulation, many applications rely on determining the 3D rotation that aligns two objects or sets of points. The de facto solution is to use singular value decomposition (SVD), where the optimal rotation is recovered as the product of the singular vectors. While correct, we observe that SVD computes much more information than necessary to build the optimal rotation. In this project, we propose a novel way for efficiently computing updates for least-squares rotational alignment problems. We show that a single Newton-like step is sufficient for small-rotation settings and further optimize implementation using AVX vectorization.

Fast Support Reduction
In layer-based 3D fabrication, supporting structures are fabricated to support overhanging regions yet discarded later. Reducing supports saves both time and material cost. In this project, we propose a real-time skinning-based method to slim down the supporting structure while maintaining a detailed-preserved and semantically meaningful geometry. We achieve this by optimizing a set of performance objectives and searching globally in the subspace spanned by the joint handles. Artifacts e.g. self-intersection can be effectively avoided. Our method is implemented via OpenGL shaders and has potential to be employed as a structural prototyping tool that facilitates model design and fabrication.


Contact

jiayieris.zhang (at) mail.utoronto.ca