Apport de l'imagerie medicale et de la biomecanique pour la modelisation d'un avatar de cirque

The 3D vizualisation of the human body is an area that is booming : the medical and military sectors as well as the gaming and movie industries are all exploring the subject in different ways. Nevertheless, they all have a common goal : to improve the aesthetic aspect (geometry) and animation (cinematic and/or biomechanics) of avatars in order to convey the most accurate virtual representation of the human body as possible. Bringing together the areas of medical imagery and the arts of the circus, the present study shows how medical imagery and biomechanics can be used in the modelling of a circus avatar.;The geometric modelling of the wire skeleton of an avatar requires one to build an articulated chain of rigid bone segments through the placement of joints. While there is currently no established procedure to follow in order to determine where to place these joints, most of the models described in reference materials place them in line with the actual anatomical joints of a human body. The first focus of our work was to personalize the geometry of the wire skeleton of an avatar's lower limb to have it in line with a Cirque du Soleil athlete. In order to do so, we used x-rays and 3D bone reconstructions of the skeleton and a 3D scan of the athlete's body.;Animation through cinematic data (derived from motion capture) can be used to apply transformations undergone by groups of markers on the athlete to the avatar's wired skeleton. We are therefore trying to translate the transformations undergone by a flexible body (skin, that each movement deforms) and apply them to a rigid body (the avatar's wire skeleton). This leads to motion errors. A second focus was therefore to analyse cinematic data in order to solidify the groups of markers chosen to animate the avatar and thus improve its animation.;The results we obtained do not necessarily demonstrate the relevance of the geometric personalization of the avatar's skeleton. However, they do show a marked improvement in animation through solidified cinematic data : the gap between actual and virtual movements can reach 30 cm when using raw cinematic data versus 5 to 10 cm when using solidified data. It would be interesting to study this further, not least by binding the avatar's wire skeleton or continue the studies on solidifying the cinematic datas.;Keywords: avatar, animation, cinematic, motion capture, skeleton.