Runfeng Li

Hi, I'm a Master's student in Computer Science at Brown University, where I'm fortunate to work with Prof. James Tompkin on reconstruction and imaging at the Brown Visual Computing Group.

Previously, I received my Bachelor's degree in Mathematics and Computer Science from University of Illinois Urbana-Champaign, where I was fortunate to work with Prof. Liangyan Gui on deep learning and human motion modeling.

Email  /  CV  /  Google Scholar  /  Github

profile photo

Research

I'm interested in computer vision and its intersection with computational imaging, computer graphics, and machine learning.

Time of the Flight of the Gaussians: Optimizing Depth Indirectly in Dynamic Radiance Fields
Runfeng Li, Mikhail Okunev, Zixuan Guo, Anh Ha Duong, Christian Richardt, Matthew O'Toole, James Tompkin
CVPR 2025 (Oral)
project page / paper / code (coming soon)

Depth optimization via C-ToF radiance field is indirect and ambiguous. Interestingly, two biases can help escape unwanted local optima. Our GS-based system trains and renders 100x faster, too.

Monocular Dynamic Gaussian Splatting is Fast and Brittle but Smooth Motion Helps
Yiqing Liang, Mikhail Okunev, Mikaela Angelina Uy, Runfeng Li, Leonidas J. Guibas, James Tompkin, Adam Harley
In Review, 2025
project page / paper

We benchmark dynamic GS methods, combining existing datasets and a new synthetic dataset to provide standardized comparisons and identify key factors affecting efficiency and quality.

Projects

More on Github
3D Gaussian Physics Simulation and Material Elasticity Reconstruction
code / presentation

We implement PhysGaussian using Taichi, and reconstruct material elasticity from multi-view videos through the differentiable GS rendering and MPM simulation system.
[← Click to toggle sim/recon.]

Gradient Domain High Dynamic Range Compression in CUDA
code

We implement Poisson PDE solvers for the gradient domain HDR compression method in CUDA/C++. Tonemapping is real-time (100+ Hz) for HD images.
[← Gamma v.s. HDRC, planning to finetune the parameters and refine the result a bit]


Based on Jon Barron's template.