Vous êtes cordialement invités à venir assister à la soutenance d'HDR de Ludovic Hoyet, qui se tiendra le mardi 3 mai à 14h dans le Grand Amphi.
A cette occasion, une chaîne YouTube a été créée : https://youtu.be/gcT_suxqgtQ
With the development of real-time graphics, virtual humans have become a requisite to create always more life-like virtual worlds for industries ranging from entertainment to training and education. For instance, in the movie and video game industries their use ranges from creating digital doubles of actors in computer-generated movies, to populating digital worlds using thousands of virtual characters. In these industries, the current trend is to create always more realistic characters, as well as more impressive scenes involving larger and larger groups of characters interacting in real-time with each other. However, because of computational, costs, and time constraints, creating and animating such virtual characters often requires to find the best trade-off between realism and computational needs. This stresses the need for highly efficient methods to animate virtual humans, that are able to create believable motions, behaviours, interactions, as well as to easily scale depending on the number of characters displayed on screen.
To reach this objective, our research activities focus on how to create visually plausible virtual humans, by first understanding how viewers perceive some of their key characteristics. By visually plausible, we mean motions that viewers will consider to be plausible even though they might not be biomechanically or physically correct. We explored these questions through a number of contributions investigating the perception of individual character motions, as well as of character interactions. Because of the complexity of stuying some characteristics in an ecological manner, we then explored novel ways of studying human behaviours using Virtual Reality with the goal of improving crowd simulation and animation models. In parallel, we also investigated the topic of avatars, with the aim of better understanding the influence of several factors on the sense and embodiment in order to create avatars that are better embodied and more interactive.
- Dr. Catherine Pelachaud (CNRS/UPMC, France)
- Pr. Anthony Steed (University College London, UK)
- Dr. Ronan Boulic (EPFL, Switzerland)
- Pr. Marie-Paule Cani (Ecole Polytechnique, France)
- Pr. Eric Marchand (Université de Rennes 1, France)
- Pr. Carol O'Sullivan (Trinity College Dublin, Ireland)