Guangyan Cai

Guangyan Cai

Research Engineer

SceniX

Biography

My name is Guangyan Cai (蔡广彦), and I am a research engineer at SceniX, where I develop simulation engines for robotic learning systems.

I earned my Ph.D. in Computer Science from the University of California, Irvine, School of Information and Computer Sciences, under the supervision of Prof. Shuang Zhao. My doctoral research focused on physics-based differentiable rendering and its applications, including inverse rendering.

Previously, I received my B.S. in Computer Science from the University of California, San Diego, where I worked with Prof. Ravi Ramamoorthi.

Education
  • Ph.D. in Computer Science, 2020 - 2025

    University of California, Irvine

  • B.S. in Computer Science, 2016 - 2020

    University of California, San Diego

Work Experience

Research-related

 
 
 
 
 
SceniX
Research Engineer
Aug 2025 – Present New York, NY
  • Developing simulation engines for robotic learning systems.
 
 
 
 
 
Meta Reality Labs
Research Intern
Jun 2022 – Sep 2022 Redmond, WA
  • Investigated the baking artifacts in material reconstruction with inverse rendering and proposed a method to mitigate them.
  • Participated in building a hybrid pipeline that combines NeRF and physics-based differentiable rendering to do high quality 3D reconstruction.
  • Showcased our reconstruction results at Meta Connect 2022 (starting at 1:13:20).
  • Published our work at ICCV 2023 (link).
 
 
 
 
 
Adobe Research
Research Intern
Jun 2023 – Sep 2023 San Jose, CA
  • Developed a novel, cost-effective lighting representation called Envmap++ for accurate reconstruction of glossy objects in indoor environments.
  • Conducted research on improving the fidelity of glossy object reconstruction under complex indoor illumination conditions.
  • Sumitted to arXiv (link)

Publications

Quickly discover relevant content by filtering publications.
(2025). Image-space Adaptive Sampling for Fast Inverse Rendering. Proceedings of the Special Interest Group on Computer Graphics and Interactive Techniques Conference Conference Papers.

Cite DOI URL

(2025). A Survey on Physics-based Differentiable Rendering.

Cite arXiv

(2024). PBIR-NIE: Glossy Object Capture under Non-Distant Lighting. arXiv.

Cite DOI arXiv

(2023). Neural-PBIR Reconstruction of Shape, Material, and Illumination. Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV).

Cite Project Paper

(2022). Physics-Based Inverse Rendering using Combined Implicit and Explicit Geometries. Computer Graphics Forum.

Cite DOI Paper Supplement Code URL

(2021). Differentiable Time-Gated Rendering. ACM Trans. Graph..

Cite DOI Paper Talk Slides Supplement Code URL

(2020). Analytic Spherical Harmonic Gradients for Real-Time Rendering with Many Polygonal Area Lights. ACM Trans. Graph..

Cite DOI Paper Video Code URL