Prof. Christian Theobalt is a world-renowned expert working on problems at the intersection of computer graphics and computer vision. He has done pioneering work on methods for capturing and re-rendering models of the static and dynamic real world in motion from camera images. A core direction of his research is human performance capture, i.e. the process of reconstructing detailed geometric shapes and appearance of actors in motion from camera images. Prof. Theobalt was one of the founders of this sub-area and has helped to improve the methods from humble beginnings to a tool that is now routinely used in the movie industry. Prof. Theobalt is a highly prolific researcher, and his work is highly cited, with an explosive growth in citation numbers in recent years.
Several seminal works have come out of this activity. “Free-viewpoint video of human actors” (Siggraph 2003) showed one of the first approaches in computer graphics where a model-based representation of the human body is captured based on silhouette information and then free-viewpoint rendered from arbitrary camera views. This work was an initial step that demonstrated the power of marker-free motion capture and model-based 3D video of humans in action. Many improvements followed, including “Performance Capture from Sparse Multi-view Video” (Siggraph 2008), which for the first time demonstrated dense performance capture with loosely fitting clothing, as well as fast motion. In recent years, Prof. Theobalt pushed the boundary of marker-less human capture further. For instance, he developed “VNect” the first algorithms for real-time marker-less human motion capture with a single color camera (Siggraph 2017), and “LiveCap” the first algorithm for real-time dynamic geometry capture of humans in loose clothing from a single color camera (ACM TOG 2019). In addition, Prof. Theobalt did seminal work on marker-less hand motion capture and face performance capture, serving his goal to build new techniques for the creation of virtual humans.
Further on, Prof. Theobalt has recently presented new ways of deeply integrating model-based approaches from computer graphics and computer vision with concepts from deep learning. His landmark work on model-based face auto-encoders (CVPR 2017) was one of the first to show a hybrid algorithm combining a CNN-based encoder and a differentiable face renderer in an end-to-end-trainable architecture for dense reconstruction of face geometry, reflectance and lighting. Christian Theobalt also made important contributions to lift neural rendering algorithms off the ground, a new class of image synthesis approaches uniting concepts from graphics and deep learning in new ways. An iconic example is his “Deep Video Portraits” (Siggraph 2018) algorithm that received large attention in both the scientific community and popular press. It uses a combination of model-based face reconstruction and neural network-based image synthesis to re-animate and edit human portrait videos in previously unseen video-realistic ways.
Prof. Theobalt’s contributions have earned him a number of highly visible awards and distinctions, including the Otto Hahn Medal 2006, the Eurographics Young Researcher Award in 2009, ERC Starting and Consolidator Grants in 2013 and 2017, respectively, and the Karlheinz Beckurts Award in 2017. Prof. Theobalt was also named as one of the “Top 40 under 40” Innovators in Germany by the business magazine Capital. He is also the co-founder of the award-winning spin-off company the Captury GmbH that commercializes the state-of-the-art system for marker-less motion capture from video.
Prof. Theobalt receives the 2020 Eurographics Outstanding Technical Contribution Award in recognition of his outstanding work at the interface of computer graphics and computer vision.