Tingyu Yang

📧 frakenation@gmail.com

Shanghai Jiao Tong University

Shanghai, China

I am an undergraduate student in Biomedical Engineering at Shanghai Jiao Tong University, advised by Prof. Xiaoyun Yuan at the OpticAI Lab, AI Institute, SJTU. You can also call me Theo.

My research interests lie at the intersection of generative AI, computational photography, and low-level vision. I am particularly interested in image restoration and enhancement under challenging conditions, including nighttime photography, compression artifacts, and human-centered visual recovery. My work explores how diffusion models, semantic guidance, and perceptual optimization can be used to improve both visual quality and content fidelity.

I have (co-)first-authored papers accepted at NeurIPS 2025, AAAI 2026, and SIGGRAPH 2026, covering topics from night photography enhancement and JPEG artifact removal to human-aware image restoration with one-step diffusion.

My journey into this field began with Jue Gong, who opened its door and showed me the first glimmer of what lay beyond. I am also grateful to Prof. Yulun Zhang for our earlier collaboration, which helped shape the path I continue to follow.

Education

B.Eng. in Biomedical Engineering, Shanghai Jiao Tong University, 2023-present

news

Apr 01, 2026	Paper LUCID accepted to SIGGRAPH 2026 as a Conference Paper (first author).
Dec 01, 2025	Paper SODiff accepted to AAAI 2026 (co-first author). [arXiv]
Sep 01, 2025	Paper HAODiff accepted to NeurIPS 2025 (co-first author). [arXiv]

selected publications

SIGGRAPH

LUCID: Learning Unified Control for Image Deflaring and Exposure Mastery in Nighttime Photography

Tingyu Yang, Yuan Cheng, and Xiaoyun Yuan

In ACM SIGGRAPH 2026 Conference Papers, 2026

Abs arXiv Code Hugging Face Website

We present LUCID, a unified controllable image restoration framework for nighttime photography that jointly handles lens flare removal and exposure adjustment. By combining flare decoupling with Diffusion CFG control, LUCID achieves continuous brightness adjustment through mixed training and supports text-guided light source preservation, enabling multi-mode training and multi-effect inference within a single model. The framework also leverages multi-exposure scene reconstruction for high-quality nighttime HDR synthesis, achieving state-of-the-art IQA metrics on large-scale real-world nighttime datasets.
AAAI

SODiff: Semantic-Oriented Diffusion Model for JPEG Compression Artifacts Removal

Tingyu Yang^*, Jue Gong^*, Jinpei Guo, and 3 more authors

In Proceedings of the AAAI Conference on Artificial Intelligence, 2026

Abs arXiv

We propose SODiff, an image restoration method for JPEG compression artifact removal that aligns image features with semantic guidance to better exploit pretrained generative priors. By bridging compressed image features with semantic information, SODiff improves texture reconstruction naturalness and overall perceptual quality under high compression ratios.
NeurIPS

HAODiff: Human-Aware One-Step Diffusion via Dual-Prompt Guidance

Jue Gong^*, Tingyu Yang^*, Jingkai Wang, and 5 more authors

In Advances in Neural Information Processing Systems, 2025

Abs arXiv Website

HAODiff addresses image restoration for human subjects under complex degradation conditions. By integrating human-aware priors with conditional guidance mechanisms, it enhances the recovery of motion-blurred body regions, maintaining perceptual quality and structural fidelity in challenging real-world degradation scenarios.