About

As a Ph.D. student

I am currently a Ph.D. student at the Dept. of Artificial Intelligence, Yonsei University. My research interests are generative AI especially image/video generation and editing. I have developed diffusion-based image editing methods. Recently, my focus on research has been extending video generation and editing including talking head problems.

Previously, I completed my internship at Adobe Research (2023 May-Dec), advised by SeoungWug Oh, JoonYoung Lee, Yang Zhou, Difan Liu, Feng Liu, and other great team members. 

As a YouTuber

I am currently running YouTube channels called Veritasium Korean and 3Blue1Brown Korean, which focus on math and science stories. Since 2021, I have been officially collaborating with the original Veritasium channel. Currently, the channel has nearly 250k and 100k subscribers. 

As a Co-founder

I am currently a co-founder of CineLingo, a startup that just launched in 2024. Based on my years of experience as a translation YouTuber, I established a startup that provides AI services for translating videos. Since 2023, I have been researching and working on this as a side project with Jaeseok Jeong, who has greatly supported me. 

Papers

• 🧡❤️ [2026] "FlowBlending: Stage-Aware Multi-Model Sampling for Fast and High-Fidelity Video Generation" is arxived.

• 🧡 [2026] "Syncphony: Synchronized Audio-to-Video Generation with Diffusion Transformers" was accepted in ICLR2026!

• 🧡 [2025] "Balanced Conic Rectified Flow" was accepted in Neurips2025!

• ❤️ [2025] "JAM-Flow: Joint Audio-Motion Synthesis with Flow Matching" is arxived.

• 🧡 [2025] "TTS-CtrlNet: Time varying emotion aligned text-to-speech generation with controlnet" will be arxived soon.

• ❤️ [2025] "TCFG: Tangential Damping Classifier-free Guidance" was accepted in CVPR2025!

• ❤️ [2024] "Harnessing Text-to-Image Models for Video Generation" was accepted in ECCV2024!

• 🧡 [2024] "Motion Customization Video Diffusion Model" was accepted in ECCV2024!

• ❤️ [2024] "Attribute Based Interpretable Evaluation Metrics for Generative Models" was accepted in ICML2024!

• 🧡 [2024] "Plug-and-Play Diffusion Distillation" was accepted in CVPR2024!

• ❤️ [2024] "Training-free Style Transfer Emerges from h-space in Diffusion models" was accepted in WACV2024!

• ❤️ [2023] "Understanding the Latent Space of Diffusion Models through the Lens of Riemannian Geometry" was accepted in Neurips2023!

• ❤️ [2023] "Unsupervised Discovery of Semantic Latent Directions in Diffusion Models" was arxived.

• ❤️ [2023] "Diffusion Models already have a Semantic Latent Space" was accepted in ICLR2023 as a notable-top-25%!

• 🧡 [2022] "FurryGAN: High Quality Foreground-aware Image Synthesis" was accepted in ECCV2022!

❤️ : First author or co-first author   🧡 : My favorite co-working papers!

Research Experience

• Mar 2021 - Present, Graduate Student Researcher, ViLab (Visual Intelligence Lab), Research Advisor: Prof. Yongjung Uh

• June 2025 - September 2025, Atmanity research internship, working with Yi-Hsuan Tsai, Yumin Suh, and other great team members.

• May 2023 - Dec 2023, Adobe summer internship (full‐time internship extended), advised by SeoungWug Oh, JoonYoung Lee, Yang Zhou, Difan Liu, Feng Liu, and other great team members.

Awards

• Won the Intra-University Merit Paper Award

• Gold prize in Samsung Humantech Paper Award, 2022  (1st prize in the category ‐ Computer Science, $8000 (Authors) + $4000 (Corresponding Author)

• Won the Best Graduation Project Award across my entire undergraduate university - Received the University President's Award

• Received the Dean's Award from the College of Engineering at Korea University

• Gold Medal at the National Research Council of Thailand (Patent Number: 4‐2019‐084383‐5)

Patents

• P12986-US: MOTION CUSTOMIZATION FOR DIGITAL VIDEOS

• 8828-361-NP1_P12631-US: VIDEO GENERATION USING FRAME-WISE TOKEN EMBEDDINGS

Academic Services

• Reviewer in CVPR, ICLR, Neurips, ICML, ECCV, AAAI, and some journals