About me

(王子陈)

Hi! My name is Zichen Wang. I'm currently a master's student at University of Windsor, supervised by Prof. Jonathan Wu and co-supervised by Prof. Ning Zhang. Previously, I did my bachelor degree in Electronics Science and Technology from Yunnan University in China.

My research mainly focuses on NeRF/3DGS, inverse rendering, and generative models.

Here is my CV. You can email me at luckily1640@gmail.com

My research focus was mainly on developing generalizable NeRF models that can be applied to diverse scenes without requiring scene-specific training, unlike the original NeRF. Through in-depth analysis and extensive experimentation, I have explored seminal methods such as Pixel-NeRF, GNT, Match-NeRF, and TensoRF. Building on these approaches, I have implemented various deep learning techniques, including novel transformers, advanced CNN architectures, and fine-tuning pre-trained image models.
Furthermore, I explored an alternative approach, namely, inverse-based methods, which involve transferring a pre-trained NeRF model to subnetwork nodes. Although the experimental results were not satisfactory, this investigation broadened my understanding of diverse optimization strategies.
In addition to NeRF, I have a grasp of the theoretical foundations of various generative models, including VAE and the vanilla diffusion model. I am eager to explore the potential of integrating these models with NeRF/3DGS to enhance 3D generation capabilities.
I'm currently delving into differentiable rasterization techniques from 3DGS, focusing on resolving the sparse input view challenge.