Chen Zhaoxi Wins 2026 CCDS Outstanding PhD Thesis Award; Wang Jianyi and Dong Junhao Recognised as Runner-Up and Honourable Mention

A jury of faculty in CCDS has conferred the college’s top internal honour for PhD research on Chen Zhaoxi by naming him the 2026 winner of the Outstanding PhD Thesis Award (OPTA).

The OPTA is CCDS’s highest internal honour for PhD research, presented annually to graduates whose dissertations demonstrate originality, rigour, and far-reaching impact. For the 2026 cycle (1 April 2025 to 31 March 2026), the jury comprised Assoc Prof Qian Kemao (Chair), Assoc Prof Cheong Kang Hao, Assoc Prof Lin Guosheng, and Assoc Prof Luu Anh Tuan.

Chen, who had also previously received the NTU PhD Medal of Honour, received the OPTA for his thesis, Multimodal Neural 3D Asset Synthesis. The award was given for making “seminal contributions to efficient, generalisable, and controllable neural asset synthesis of digital humans, scenes and objects, that can power the commercial production of next-generation digital content.”

The jury wrote that Chen’s thesis “formulates a unifying primitive-centric representation and introduces six seminal systems—Relighting4D, PrimDiffusion, URHand, Text2Light, SceneDreamer, and 3DTopia-XL—that collectively redefine the production pipeline for geometry, reflectance, and radiance. These demonstrate annotation-free 4D human relighting, diffusion-based human generation with dense correspondence, universal hand relighting, zero-shot text-to-HDR panorama creation, unbounded scene generation from unlabeled images, and scalable physically-based object modeling. [...] The thesis represents a milestone in the convergence of graphics, vision, and artificial intelligence, empowering next-generation interactive digital worlds everywhere.”

Asset synthesis has important applications in allowing physical AI systems to better model and interact with the world around them. This is a field Chen is currently exploring as the CEO and co-founder of Ropedia. Ropedia, which recently raised US$30 million to scale data infrastructure for physical AI, is an NTU spin-off developing real-world data infrastructure for robotics, world models and embodied intelligence.

Chen was advised by Professor Liu Ziwei and was a member of the S-Lab for Advanced Intelligence.

The 2026 OPTA runner-up, Wang Jianyi, was recognized for “seminal contributions to research on image quality assessment and technical leadership in image and video restoration” in his thesis, Harnessing Foundation Models for Visual Content Restoration and Quality Assessment.

Wang, who is currently a Senior Researcher at Apple Machine Learning Research, was a member of MMLab@NTU and advised by Professor Chen Change Loy.

In the citation for the runner-up award, the jury wrote, “Jianyi’s research focuses on harnessing large vision foundation models in real-world scenarios, leading to significant improvement in both image and video restoration quality in practical applications. [...] His works, such as CLIP-IQA and StableSR, have drawn broad interests from the research community, receiving some of the highest citations in their corresponding years. Furthermore, these works have also been translated into commercial products, extending its impact beyond academia.”

CLIP-IQA is integrated into large commercial platforms such as Alibaba Cloud and LIFULL (one of Japan’s largest online real estate platforms) to automatically evaluate videos and images, while StableSR is integrated into the open-source AI/ML community HuggingFace.

Dong Junhao, also a previous winner of the NTU PhD Medal of Honour, received an honourable mention for his thesis, Towards Generalizing Adversarial Robustness in Deep Learning. Dong was advised by Professor Ong Yew Soon.

Chen and Wang will be presenting their achievements and research at the 2026 OPTA Symposium on Friday, 9 October 2026. The symposium will be open to all CCDS research students, researchers, and faculty.