Justus Thies

Justus Thies is a Professor at the Technical University Darmstadt (since September 2023), leading research at the intersection of computer graphics, computer vision, and machine learning. His work focuses on real-time 3D reconstruction, tracking, and photorealistic rendering of humans and environments using commodity hardware, with applications in AR/VR, movie production, and 3D telepresence for collaborative virtual environments.

• Current faculty member at Technical University Darmstadt
• Research themes: 3D reconstruction, neural rendering, hybrid machine learning/Graphics methods
• Applications: AR/VR, telepresence, human-environment modeling

His research emphasizes neural rendering and deep learning approaches for facial and body modeling, including projects like GEM (Gaussian Eigen Models), SynShot (synthetic prior inversion), and TeCH (text-guided clothed human reconstruction). Collaborations include institutions like Max Planck Institute for Intelligent Systems and Google.

Key trends in his recent publications include: neural Gaussian splatting for real-time avatars, diffusion models for speech/audio-driven animation, and hybrid approaches that combine classical computer graphics principles with modern machine learning techniques. These methods address challenges in generalizability, controllability, and editability of 3D representations.

Notable research contributions since 2014 include advancements in monocular facial reconstruction, real-time expression transfer, and neural rendering techniques that enable 3D scene synthesis from unpaired or sparse data. His work has received significant recognition, including Face2Face featured in Communications of the ACM (2019) and multiple publications at top venues like CVPR, ECCV, and SIGGRAPH.