Christian Theobalt
Max Planck Institute for Informatics, Saarland University, University of Edinburgh
Papers
18
Total Citations
1,787
H-Index
13
About
Christian Theobalt is a leading computer vision and graphics researcher whose work spans 3D reconstruction, human motion synthesis, and human-computer interaction. He is perhaps best known for BundleFusion, a landmark system for real-time, globally consistent 3D scene reconstruction that addresses the critical challenge of drift in large-scale scanning — a contribution so influential it has accumulated over 1,000 citations across its versions and remains a foundational reference in mixed reality and robotics research. Theobalt's work consistently bridges the gap between high-fidelity 3D understanding and practical, real-time deployment. His research on hand tracking (RGB2Hands) tackles the notoriously difficult problem of reconstructing two interacting hands from monocular RGB video, while BEHAVE advances the field by providing tools to model complex human-object interactions in natural environments. More recently, he has embraced generative AI, with MoFusion introducing diffusion-based frameworks for high-quality human motion synthesis, and earlier work exploring text-driven character animation. Spanning robotics dialogue systems, depth estimation, and neural rendering, Theobalt's portfolio reflects both remarkable breadth and sustained depth. His contributions have shaped how machines perceive, reconstruct, and animate the human world, making him an essential figure for any student working at the intersection of computer vision, graphics, and embodied AI.
Research Focus
Key Achievements
Top Papers
- 1BundleFusion680 citations · 2017
- 2BundleFusion245 citations · 2017
- 3Synthesis of Compositional Animations from Textual Descriptions155 citations · 2021
- 4BEHAVE: Dataset and Method for Tracking Human Object Interactions146 citations · 2022
- 5MoFusion: A Framework for Denoising-Diffusion-Based Motion Synthesis127 citations · 2023
- 6
- 7RGB2Hands85 citations · 2020
- 8Talking to Godot: dialogue with a mobile robot59 citations · 2003
- 9Multi-view Depth Estimation using Epipolar Spatio-Temporal Networks55 citations · 2021
- 10