Hyunsoo Lee, an undergraduate student in the Department of Electrical and Computer Engineering at Seoul National University’s College of Engineering, has presented research in generative visual computing at leading international conferences, including NeurIPS, CVPR, and ECCV. His work spans image editing and 3D content generation, leveraging pretrained generative models to align outputs with user intent and maintain structural consistency. Lee began with text-conditioned image editing, co-authoring "Conditional Score Guidance for Text-Driven Image-to-Image Translation," accepted by NeurIPS 2023, which introduced a method that modifies text-specified regions while preserving source image structure. He extended this to high-resolution editing in "Low-Resolution Editing is All You Need for High-Resolution Editing," accepted by CVPR 2026, which uses a low-resolution edit as a semantic reference and enables editing at resolutions above 1K while preserving global semantics and local detail. Lee also investigated diffusion synchronization in "SyncSDE: A Probabilistic Framework for Diffusion Synchronization," presented at CVPR 2025, which identifies where correlations should be introduced among diffusion trajectories and provides a unified framework for collaborative generation across images, human motion, and 3D content.