Xu Peiyu

Computer Science, UIUC, IL, US
peiyuxu3@illinois.edu

I am a Ph.D. student in Computer Science at the University of Illinois Urbana-Champaign, advised by Prof. Shuang Zhao. My research focuses on differentiable rendering, ray tracing, and neural scene representations, with an emphasis on physically-based, GPU-friendly algorithms.

CV (PDF)  /  Google Scholar  /  GitHub  /  LinkedIn  / 

profile photo

Education

  • Ph.D. in Computer Science, University of Illinois Urbana-Champaign (UIUC), US — Sept. 2025 to Present
    Supervisor: Prof. Shuang Zhao
  • B.Eng. in Computer Science, HKUST, Hong Kong, China — Aug. 2018 to Sept. 2022
    Supervisor: Prof. Pedro Sander

Work Experience

  • Research Scientist Intern, Adobe Research, San Jose, US — June 2025 to Sept. 2025
  • Research Scientist Intern, Nvidia Real-time Rendering Research, Redmond, US — June 2024 to Sept. 2024
  • Research Assistant, UCSD, San Diego, US — June 2021 to Sept. 2021

Research

 Stochastic Ray Tracing for 3D Gaussian Splatting
Peiyu Xu, Xin Sun, Krishna Mullia, Raymond Fei, Iliyan Georgiev, Shuang Zhao,
CVPR 2026
Project / Paper

Efficient and generalizable 3DGS optimization via sorting-free stochastic ray tracing.

Robust and Efficient Differentiable Rendering Through Kernel Density Estimation
Peiyu Xu, Lifan Wu, Benedikt Bitterli, Ravi Ramamoorthi, Shuang Zhao,
ACM Transactions on Graphics (SIGGRAPH 2026)
Project (Coming Soon)

Kernel-density estimation on parameter space for efficient differentiable and inverse rendering.

Langevin Monte Carlo Based Sampling of Visibility Boundaries
Peiyu Xu, Sai Bangaru, Tzu-mao Li, Shuang Zhao,
ACM SIGGRAPH Asia 2024 (conference-track full paper)
Project / Paper

Manifold walk on the geometry silhouette space for efficient sampling of visibility boundaries.

Warped-Area Reparameterization of Differential Path Integrals
Peiyu Xu, Sai Bangaru, Tzu-mao Li, Shuang Zhao,
ACM Transactions on Graphics (SIGGRAPH Asia 2023)
Best Paper Award
Project / Paper

Differentiable path integral with divergence-theorem-based interior estimators.

Miscellaneous

 Camera Lens Simulation

Simulation of realistic camera lenses on the Mitsuba renderer, with support for complex lens geometries, chromatic aberration, physically-based defocus blur and distortion, etc.

Awards

  • SIGGRAPH Asia 2023 Best Paper Award

Skills

  • Languages: C++, Python, PyTorch
  • GPU Programming: Slang.D, CUDA, OptiX, Vulkan

Website template adapted from Jon Barron.