Holography represents the frontier of visual technology. By encoding the full wavefield of light (capturing both amplitude and phase information), it holds the potential to be the ultimate display system. Unlike traditional stereoscopic screens, holographic displays can perfectly reproduce all natural visual cues, including continuous parallax and physically accurate focus cues.
However, realizing this potential comes with a significant hurdle: computing holograms requires complex wave-optics rendering, making it highly computationally intensive. To make high-end holographic displays practical, particularly for next-generation augmented and virtual reality (AR/VR) systems, the field needs fundamentally novel algorithms and data representations.
Recently, 3D Gaussian mixture models, such as 3D Gaussian splatting, have revolutionized standard scene representation, enabling highly efficient, high-quality rendering. Yet, these existing models are natively designed for ray-based optics and are not directly adapted to the complex, wave-based requirements of holographic rendering.
The primary aim of this thesis is to bridge this gap. The student will develop novel algorithms and data structures using multivariate Gaussian components to create compact, adaptive plenoptic representations specifically tailored for holography. By encoding light fields into these sophisticated mixture models, this research seeks to drastically optimize the computational pipeline, bringing real-time, high-fidelity holographic rendering one step closer to reality.

The first challenge is to bridge the gap between geometric optics and wave optics. The student will develop novel 3D Gaussian mixture representations specifically adapted for holography. This requires moving beyond traditional rendering by accounting for ray-to-wave conversion and incorporating complex-valued amplitudes to accurately model light interference patterns.
Moving from theory to practice, the student will implement an extended rendering framework. This software pipeline will be responsible for translating the newly developed, complex-valued Gaussian mixture representations into holograms.
The developed algorithms will be rigorously evaluated. First, the student will test the framework within a simulated holographic rendering pipeline to benchmark computational efficiency and image quality. Finally, the research culminates in physical validation: the student will deploy their holograms on an in-house holographic display prototype, directly observing the real-world performance and optical characteristics of their plenoptic encoding approach.

Zhan, Yicheng, Dong-Ha Shin, Seung-Hwan Baek, and Kaan Aksit. "Complex-Valued Holographic Radiance Fields." ACM Transactions on Graphics (2025). doi: 10.1145/3804450

David Blinder, Tobias Birnbaum, Tomoyoshi Ito, Tomoyoshi Shimobaba. The state-of-the-art in computer generated holography for 3D display[J]. Light: Advanced Manufacturing 3, 35(2022). doi: 10.37188/lam.2022.035

Strong programming skills (Python, C/C++, GPU programming is a plus)
Familiarity with 3D graphics is recommended
Interest in experimental physical research and evaluation (visual and optical testing)