Yeonjin Chang

I am an integrated MS/PhD student at Seoul National University, advised by Professor Nojun Kwak as a member of Machine Intelligence and Pattern Analysis Lab (MIPAL). My research focuses on 3D reconstruction, multi-view geometry, and neural rendering. I began my graduate work on few-shot NeRF methods, and my recent projects have shifted toward 3D Gaussian Splatting–based approaches, driven by an interest in modeling 3D reconstructions that more accurately capture real-world characteristics. My work spans relighting under natural illumination, 3D object extraction using diffusion-based inpainting, and 3D-consistent scene colorization. Prior to my graduate studies, I earned a bachelor’s degree in Astronomy and Industrial Engineering from Seoul National University.

Email  /  CV  /  Google Scholar  /  Github

We introduce a 3D colorization method, LoGoColor, that avoids color-averaging limitations of prior methods by generating locally and globally consistent multi-view colorized training images, enabling diverse and consistent 3D colorization for complex 360$^{\circ}$ scenes.

We introduce a 3D object extraction method for Gaussian Splatting that prunes irrelevant primitives using K-nearest neighbors analysis and compensates for occlusions with diffusion-based generative inpainting.

We introduce ARC-NeRF, a few-shot rendering method that casts area rays to cover a broader set of unseen viewpoints, improving spatial generalization with minimal input. Alongside, we propose adaptive frequency regularization and luminance consistency loss to further refine textures and high-frequency details in rendered outputs.

We simplify outdoor scene relighting for NeRF by aligning with the sun, eliminating the need for environment maps and speeding up the process using a novel cubemap concept within the framework of TensoRF.

We utilize the flipped reflection rays as additional training resources for the few-shot novel view synthesis, leading to more accurate surface normal estimation.

We model a ray with mixture density model, leading to efficient learning of density distribution with sparse inputs, and propose an effective auxiliary task of ray depth estimation for few-shot novel view synthesis.

Write short description here.