Dynamics for People, Plants and Clothes

    Wednesday, 20 November

    09:00 - 22:45

    Convention Hall B

    Analyzing Growing Plants from 4D Point Cloud Data

    We introduce a framework to analyse time-lapse 3D scan of growing plants, particularly focusing on accurate localization and tracking topological events like budding and bifurcation. We evaluate our approach on multiple data sets and use the results to animate static virtual plants or directly attach them to physical simulators.

    Yangyan Li, Shenzhen Institute of Advanced Technology
    Xiaochen Fan, Shenzhen Institute of Advanced Technology
    Niloy J. Mitra, University College London
    Daniel A. Chamovitz, Tel Aviv University
    Daniel Cohen-Or, Tel Aviv University
    Baoquan Chen, Shenzhen Institute of Advanced Technology and Shandong University

    Reconstructing Detailed Dynamic Face Geometry from Monocular Video

    We present a method for capturing face geometry from monocular video. It works under uncontrolled lighting and successfully reconstructs expressive motion including high-frequency face detail. After simple manual initialization, the capturing process is fully automatic. We demonstrate performance capture results for long and complex sequences captured indoors and outdoors.

    Pablo Garrido, Max Planck Institute for Informatics
    Levi Valgaerts, Max Planck Institute for Informatics
    Chenglei Wu, Max Planck Institute for Informatics
    Christian Theobalt, Max Planck Institute for Informatics

    Inverse Dynamic Hair Modeling with Frictional Contact

    We present a method for converting a hair geometry into a physics-based hair model, such that the static hair pose, in the presence of gravity and frictional contact, accurately matches the input geometry. Our method was used to animate both artistic hairstyles and recent hair data reconstructed from capture.

    Alexandre Derouet-Jourdan, INRIA and CNRS
    Florence Bertails-Descoubes, INRIA and CNRS
    Gilles Daviet, INRIA and CNRS
    Joelle Thollot, INRIA and CNRS

    On-Set Performance Capture of Multiple Actors With A Stereo Camera

    We describe a new method which is able to track skeletal motion and detailed surface geometry of one ore more actors from footage recorded with a stereo rig that is allowed to move. It succeeds in general sets with uncontrolled background and uncontrolled illumination.

    Chenglei Wu, Max Planck Institut Informatik
    Carsten Stoll, Max Planck Institut Informatik
    Levi Valgaerts, Max Planck Institut Informatik
    Christian Theobalt, Max Planck Institut Informatik