Lee Hyoseok

I received my master's degree in Artificial Intelligence at a POSTECH, advised by Tae-Hyun Oh.

I work on research problems in computer graphics, vision, and machine learning. My research interests include 3D Reconstruction, Computational Photography, and Generative Model, but not limited to.

CV / Google Scholar / LinkedIn / GitHub / Email

Publications

	Measurement-Consistent Langevin Corrector: A Remedy for Latent Diffusion Inverse Solvers Lee Hyoseok , Sohwi Lim , Eunju Cha , Tae-Hyun Oh arXiv, 2026 arXiv / Best Excellence Prize, Electronics Times ICT Paper Awards, 2025 MCLC, a plug-and-play module that stabilizes and improves latent diffusion inverse solvers.
	JointDiT: Enhancing RGB-Depth Joint Modeling with Diffusion Transformers Byung-Ki Kwon , Qi Dai , Lee Hyoseok , Chong Luo , Tae-Hyun Oh ICCV, 2025 arXiv / Project page JointDiT, a diffusion transformer that models the joint distribution of RGB and depth within a single unified model.
	Dress-up: Generating Animatable Clothed 3D Humans via Latent Modeling of 3D Gaussian Texture Maps Kim Youwang , Lee Hyoseok , Gerard Pons-Moll , Tae-Hyun Oh Workshop on Computer Vision for Fashion, Art, and Design, ICCV, 2025 Oral Presentation Dress-Up, a feed-forward, unconditional generative model for creating photorealistic, animatable, clothed 3D human avatars
	Conditioned Image-to-Image Retrieval via Concept-based Visual Projections in Vision-Language Model Sohwi Lim , Lee Hyoseok , Tae-Hyun Oh Workshop on Computer Vision in the Wild, CVPR, 2025 Conditioned image-to-image retrieval
	FPGS: Feed-Forward Semantic-aware Photorealistic Style Transfer of Large-Scale Gaussian Splatting GeonU Kim , Kim Youwang , Lee Hyoseok , Tae-Hyun Oh Under review, 2025 arXiv / Project page Excellence Prize, Electronics Times ICT Paper Awards, 2024 FPGS performs feed-forward semantic-aware photorealistic style transfer of Gaussian Splatting.
	Zero-shot Depth Completion via Test-time Alignment with Affine-invariant Depth Priors Lee Hyoseok , Kyeong Seon Kim , Byung-Ki Kwon , Tae-Hyun Oh AAAI, 2025 arXiv / Code / Project page Winner of the Qualcomm Innovation Fellowship Korea 2025 Zero-shot depth completion: Align sparse depth measurements with affine-invariant depth diffusion prior at test time.
	Bootstrapping Multi-View Features via Bridging the Gap between Linearity of Rendering and Non-linearity of 2D Feature Space Lee Hyoseok , Kim Jun-Seong , Kim Ji-Yeon , Tae-Hyun Oh IPIU (Workshop on Image Processing and Image Understanding), 2024 Outstanding Poster Presentation Award Constructing versatile 3D feature field by addressing feature rendering equation and bootstrapping multi-view features.

Design and source code from Jon Barron and Leonid Keselman.

Lee Hyoseok

Publications

Measurement-Consistent Langevin Corrector: A Remedy for Latent Diffusion Inverse Solvers

JointDiT: Enhancing RGB-Depth Joint Modeling with Diffusion Transformers

Dress-up: Generating Animatable Clothed 3D Humans via Latent Modeling of 3D Gaussian Texture Maps

Conditioned Image-to-Image Retrieval via Concept-based Visual Projections in Vision-Language Model

FPGS: Feed-Forward Semantic-aware Photorealistic Style Transfer of Large-Scale Gaussian Splatting

Zero-shot Depth Completion via Test-time Alignment with Affine-invariant Depth Priors

Bootstrapping Multi-View Features via Bridging the Gap between Linearity of Rendering and Non-linearity of 2D Feature Space