About Me/Planning
Navigating the Academic Landscape as a Newcomer
My Academic Path
I completed my undergraduate studies at the University of Macau, majoring in Computer Science. During my master’s program of Visual Computing at Universität des Saarlandes, I worked on extending research in Regression-based Monte Carlo Integration, under the supervision of Qinqin Hua and Pascal Grittmann. Currently, I return to the University of Macau to pursue my PhD, supervised by Prof. Xiaodong Cun and Prof. Chi-Man PUN. My research now focuses on 3D Vision (3DV) and its applications in the field of AIGC.
Research Interests
My research interests bridge Physically Based Rendering (PBR) and AI-Generated Content (AIGC) within Computer Vision, focusing on the fundamental properties of visual representations. While PBR seeks objective realism through physical principles, AIGC offers powerful, data-driven but often less explainable results. I aim to understand how various representation methods—from NeRF and 3D Gaussian Splatting to PBR, across latent, vector, and pixel spaces—influence interpretability and editability, and to develop strategies for their conversion and utilization.
Specifically, my work is structured around three areas:
Visual Representation Interplay: Investigating the characteristics of diverse visual representations (e.g., NeRF, 3D Gaussian Splatting, PBR, across latent, vector, and pixel spaces) and explore effective methods for their conversion and utilization.
PBR with Large Models: Extending traditional PBR by integrating large models to address its limitations in real-world scene modeling, implementation complexity, scene constraints, and hardware utilization.
3D Vision on AIGC: Exploring how 3D Vision advancements, particularly scaling laws, improve the capture of real-world properties. This progress offers significant advantages for AIGC applications such as video generation and 3D asset creation, with the full potential of this synergy still being explored.
Evolving Research Perspectives
As time has passed, I’ve gradually formed coarse yet meaningful principles regarding research and learning, which now guide approach to my work:
Consistency: Scarce yet cultivable, it offers the most sustainable path to avoiding friction and leveraging energy long-term..
Time Management: Effective time management, rooted in holistic project understanding, delivers both efficient completion and the satisfaction of learning.
Idea > Achievement: While engineering is valuable, unique ideas are the true differentiator in research.
Learning by Doing: Ideas and practice are synergistic.
Designing Experiments: Effective experiments are simple, decomposable, and efficient.
Knowing When to “Think Hard”: Concentrate deep thinking on experiment design and result analysis. Avoid overcomplicating the experiment itself.