Inaugural Workshop May 26

The Kinovis Inaugural Workshop will take place on May 26 2015 at INRIA Grenoble.

Program of the day:

  • 9:30 am – Welcome & coffee
  • 10:00 am – Opening speech
    by Edmond Boyer, head of Morpheo research team and leader of the Equipex Kinovis
    by Patrick Gros, Director of Inria Grenoble Rhône-Alpes Research Center
  • 10:15 am – Kinovis Equipex presentation

    Kinovis: 4D modeling
    by Edmond Boyer, head of Morpheo research team and leader of the Equipex Kinovis

    Kinovis: ambitions and perspectives in healthcare
    by Pr. Olivier Palombi, Neurosurgeon-anatomist, CHU Grenoble, Grenoble Laboratory of Anatomy, Inria Team IMAGINE, LJK, University Grenoble Alpes

  • 11:00 am – Kinovis studio visit and demonstration 
  • 12:00 am – Lunch
  • 13:30 pm – Workshop
  • 13:30 pm Dynamic Surface Analysis of Performance Capture

    by Tony Tung, Rakuten Institute of Technology

In this talk, we present the multi-video camera studio of Matsuyama Laboratory, at Kyoto University which was established more than a decade ago.The system has served to capture and reconstruct 3D dynamic data in order to preserve intangible cultural heritage, such as traditional Japanese dances, among other possible applications (entertainment, sports, etc.). In particular, this system has allowed us to efficiently reproduce surface non-rigid deformations from real-world object observation. On the other hand, we have also been exploring simpler setting, but given a dynamic shape prior instead, to perform dynamic object reconstruction and surface analysis.

  • 14:15 pm 4D Vision in the Wild: Challenges & Opportunities 

    by Adrian Hilton, University of Surrey

This talk will review the current state-of-the-art in 4D Computer Vision and open challenges required to address potential future applications.  Over the past decade the use of multiple camera reconstruction and active sensing technologies has advanced to enable the capture of 4D models in controlled studio environments. This talk will focus on the challenges of 4D beyond the studio.

 

  • 15:45 pm
    by Michael Black, Max Planck Institute for Intelligent Systems, Tübingen

To look human, digital full-body avatars need to have soft tissue deformations like those of real people. Current methods for physics simulation of soft tissue lack realism, are computationally expensive, or are hard to tune. Learning soft tissue motion from examples, however, has been limited by the lack of dense, high-resolution, training data. We address this with a novel 4D capture system and a method for accurately registering 3D scans across time to a common template mesh. Using over 40,000 scans of ten subjects, we compute how soft tissue motion causes mesh triangles to deform relative to a base 3D body model and learn a low-dimensional linear subspace approximating this soft-tissue deformation. We relate the linear coefficients of this body surface deformation to the changing pose of the body using a second order autoregressive model.  We then animate new motions and new bodies, predicting how they will deform based on their shape and motion. We also transfer these soft tissue motions to new stylized characters.

Leave a Reply