GradDISN: Gradient-based Deep Implicit Surface Network for Detailed 3D Reconstruction
- Detailed 3D mesh reconstruction using an occupancy gradient-weighted loss function
About Me
I work as a ML Engineer at Beeble AI, where I push the boundaries of mobile AI virtual production with a focus on realistic lighting. Previously, I completed my Master's degree at POSTECH Computer Graphics Lab under the supervision of Prof. Seung-Hwan Baek. My graduate research focused on comprehensive analysis and modeling of light properties [NeSpoF, SpoDataset, Spectro-polarimetric BRDF].
Work Experience
- Build a synthetic data generation pipeline for OLAT images, PBR textures, and depth maps from 3D assets
- Improve the performance of de-rendering and relighting AI model by using synthetic dataset for training
- Research and evaluate AI models like 2D/3D generation, motion capture, and depth estimation for product
- Develop a Blender plugin to support 3D set creation, cloud upload, shot import, and post-capture editing
Publications
Spectral and Polarization Vision: Spectro-polarimetric Real-world Dataset
Yujin Jeon*, Eunsue Choi*, Youngchan Kim, Yunseong Moon, Khalid Omer, Felix Heide, Seung-Hwan Baek
CVPR 2024 (highlight)
Neural Spectro-polarimetric Fields
Youngchan Kim, Wonjoon Jin, Sunghyun Cho, Seung-Hwan Baek
SIGGRAPH Asia 2023
Projects
Teaching Experience
Teaching Assistant