When designing crowds for entertaining applications, a few generic walking motions are duplicated to save valuable resources, leading to a lack of variations. In this thesis, we investigate the potential use of biomechanical parameters of human walking to generate and edit walking animations that are specific to various actor’ s characteristics (e.g. age, gender, height).
We first performed a user study to evaluate the ability of observers to recognize a specific Walk Ratio (WR) when applied on virtual characters and found that the main characteristics of the WR were perceived by observers. We then extended this study by investigating more factors to better target which features contribute more to the perception of the WR.
Finally, we proposed a framework to generate lower body walking animations based on the poulaine (ankle trajectory as seen in the hip reference frame), and we initiated its evaluation through a user study. In conclusion, we showed that viewers are sensitive to biomechanical parameters of human walking for character animation and we provide guidelines to design animations respecting such parameters.
Nuria Pelechano, Associate Professor at Universitat Politecnica de Catalunya (rapporteur)
Brad McFadyen , Professor at University of Laval (examinateur)
Franck Multon, Research Director at Inria Rennes (examinateur)
Nicolas Pronost, Associate Professor at Université Claude Bernard (examinateur)
Julien Pettré, Research Director at Inria Rennes (supervisor)
Anne-Hélène Olivier, Associate Professor HDR at Université de Rennes 2 (supervisor)
Ludovic Hoyet, Research Scientist at Inria Rennes (director)