Bingxuan Li | 李秉轩

Hi! I am a first-year Computer Science Ph.D. student at NYU, advised by Prof. Qi Sun in the Immersive Computing Lab.

My current research lies at the intersection of deep learning and different areas of computer graphics, including neural/physically based rendering and 3d vision. More boardly, I am interested in leveraging advanced machine learning techniques to solve the challenges in building realistic, efficient, and interactive visual computing systems.

Prior to NYU, I received my Bachelor's degree with honors from Turing Class , Peking University . During my undergraduate years, I was honored to be advised by Prof. Sheng Li and Prof. Lingqi Yan.

Email  /  Github  /  CV  /  LinkedIn  /  WeChat  


profile photo

Research
Image-GS: Content-Adaptive Image Representation via 2D Gaussians
Yunxiang Zhang*, Bingxuan Li*, Alexandr Kuznetsov, Akshay Jindal, Stavros Diolatzis, Kenneth Chen. Anton Sochenov, Anton Kaplanyan, Qi Sun
Paper / Code
SIGGRAPH 2025

In this work, we present Image-GS, a content-adaptive neural image representation using anisotropic 2D Gaussians, offering high memory efficiency, fast random access, and flexible fidelity control for real-time graphics and machine perception.


Nano-3D: Metasurface-Based Neural Depth Imaging
Bingxuan Li*, Jiahao Wu*, Yuan Xu*, Yunxiang Zhang, Zezheng Zhu, Nanfang Yu, Qi Sun
Paper
To be presented as a demo at SIGGRAPH 2025 Emerging Technologies.

We built an ultra-compact, metasurface-based neural depth imaging system leveraging nano-optics and deep learning to achieve precise depth estimation from monocular polarized imagery.


Proxy Tracing: Unbiased Reciprocal Estimation for Optimized Sampling in BDPT
Fujia Su*, Bingxuan Li*, Qingyang Yin, Yanchen Zhang, Sheng Li
Paper / Project / Code / Video
ACM Transactions on Graphics (SIGGRAPH 2024)

In this work, we propose a novel sampling method to robustly handle challenging specular/glossy paths in bidirectional path tracing (BDPT), ensuring unbiased rendering and significantly improving sampling efficiency.


Selected Projects
EasyRender  Github

We present EasyRender, a scalable, high-performance ray tracing renderer built upon OptiX 8.0, featuring modular architecture, support for pbrt-v3 scenes and Disney Principled Materials, and implementations of advanced physically based rendering algorithms.

ROMA on Mitsuba  Github

This project explores the further applications of ROMA in rendering, implemented on Mitsuba 0.6. We primaryly focused on enhancing the performance of ray tracing using ROMA, a novel acceleration method.


About Me

I enjoy reading, gaming, photography and other nerdy stuffs outside of work.


This homepage is designed based on Jon Barron 's website and deployed on Github Pages. Last updated: May. 1, 2025
© 2025 Bingxuan Li