Yunfan Zeng
|
About MeI am an undergraduate student in the Department of Computer Science and Technology at Tsinghua University. My research interests include physically based rendering, differentiable rendering, 3D reconstruction. |
Education
-
Sep. 2020 - Jul. 2024 (Expected) : Bachelor of Engineering in Computer Science and Technology, Tsinghua University | Beijing, China
Publications
-
Wenyuan Zhang, Ruofan Xing, Yunfan Zeng, Yu-Shen Liu, Kanle Shi, Zhizhong Han, “Fast Learning Radiance Fields by Shooting Much Fewer Rays”, in IEEE Transactions on Image Processing, vol. 32, pp. 2703-2718, 2023.
Research Experience
-
Differentiable Rendering Benchmark (June 2023 - now)
Advisor: Shuang Zhao, Associate Professor of Computer Science at University of California, Irvine
Proposed a benchmark for various differentiable renderers including physics-based renderers and rasterizers, aiming to test and compare the renderers’ capability of solving inverse problems. Developed an inverse rendering toolkit as a universal platform for various differentiable renderers. Currently conducting experiments and preparing a comprehensive dataset.
-
LUISA: A High-Performance Rendering Framework (June 2022 - Sep 2022, Sep 2023 - now)
Advisor: Kun Xu, Associate Professor in Department of CS&T at Tsinghua University
Learned Luisa, a rendering framework focusing on high-performance physics-based rendering. Implemented a simple ray tracing renderer 'Nori', and learned GPU programming with Cuda for parallel rendering. Developed a path-tracing renderer for Luisa, achieving a 50x to 100x acceleration compared to its CPU-based counterpart. Currently implementing differentiable rendering algorithms on Luisa.
-
3D Synthesis with Diffusion (Apr 2023 - June 2023)
Advisor: Tai-Jiang Mu, Assistant Professor in Department of CS&T at Tsinghua University
Collected and analyzed latest studies on 3D diffusion, including LION: Latent Point Diffusion Models for 3D Shape Generation and Magic3D: High-Resolution Text-to-3D Content Creation, etc. Tried to reproduce the results in the papers and write a survey.
-
3D Reconstruction from Sparse Views (Nov 2021 - July 2022)
Advisor: Yu-Shen Liu, Associate Professor in School of Software at Tsinghua University
Contributed in proposing a method to accelerate and improve the efficiency of the Neural Radiance Field (NeRF) algorithm for the problem of 3D reconstruction from sparse views. The key idea was to reduce the redundancy by shooting much fewer rays in the multi-view volume rendering procedure, which is the base for almost all radiance fields-based methods. Adjusted the algorithm and improved its efficiency by 20% to 50% by testing the algorithm in numerous datasets.