Skip to content

A year at a glance: Computational Light Laboratory bridges student potential into scientific success with industrial partners

Written by Kaan Akşit, 1 December 2025

Assessing Learned Methods for Hologram Compression

Assessing Learned Models for Phase-only Hologram Compression

Presented by: Zicong Peng, Yicheng Zhan, Josef Spjut, and Kaan Akşit

Location: Convention Centre, Vancouver, BC, Canada

Time: Monday-Thursday, 10-14 August 2025

Session: Poster Track

Zicong Peng demonstrated exceptional motivation and outstanding implementation skills throughout his master’s thesis work. Zicong dove into a topic entirely new to him: "Deep Learning based Compression methods for Holographic Displays". These displays are an emerging field with transformative potential for next-generation augmented reality glasses, virtual reality headsets, and desktop 3D displays. Firstly, Zicong quickly mastered the replication of established hologram simulation algorithms involving light propagation techniques in free space, and learned to simulate holograms as if displayed using a holographic display. Through a rigorous series of experiments, Zicong generated structured guidance on the performance of learned hologram compression algorithms. In this context, Zicong is focused on Variational Autoencoder and Neural Implicit Representation structures. Zicong's well prepared documentation and assessments formulated the basis of a submission to the prestigious SIGGRAPH 2025 Poster Track.

We evaluate the performance of four common learned models utilizing Implicit Neural Representation (INR) and Variational Autoencoder (VAE) structures for compressing phase-only holograms in holographic displays..

Zicong further elevated the quality of his work by collaborating with industrial and academic partners, including Josef Spjut of NVIDIA, Yicheng Zhan of UCL and Kaan Akşit. As of 10 August 2025, Zicogn Peng's work was accepted and presented at SIGGRAPH 2025. Zicong’s journey serves as an exemplary model for beginners in any scientific field, proving that motivation, attentive mentorship, and active engagement with the collaborators can lead to remarkable success. Throughout this process, he developed a robust technical skillset, a deep understanding of scientific methodology, and improved communication skills—opening doors to future Ph.D. studentship opportunities in Koç University under the guidance of Professor Hakan Urey.

Implicit Neural Representations for Optical Raytracing

Efficient Proxy Raytracer for Optical Systems using Implicit Neural Representations

Presented by: Shive Sinaei, Chuanjun Zhengi, Kaan Akşit, and Daisuke Iwai.

Location: Convention Centre, Vancouver, BC, Canada

Time: Monday-Thursday, 10-14 August 2025

Session: Poster Track

Chuanjun Zheng concluded his remote internship in our laboratory, collaborating with external partners from the University of Osaka. During this collaboration, Chuanjun mentored first year Ph.D. student Shiva Sinaei by sharing insights from his earlier experimtns on implicit neural representations. Together, Chuanjun and Shiva identified an implicit neural representation capable of modeling optical lenses as a learned component. Their work demonstrated that this learned lens representation could accurately raytrace optical beams for simple imaging tasks, an interesting starting point for a more complete investigation in the future. As Chuanjun wraps up his internship, Chuanjun is set to begin a new chapter at the University of Hawai'i at Mānoa, where Chuanjun will work under the supervision of Huaijin (George) Chen.

Outreach

We host a Slack group with more than 250 members. This Slack group focuses on the topics of rendering, perception, displays and cameras. The group is open to public and you can become a member by following this link.

Contact Us

Warning

Please reach us through email to provide your feedback and comments.