The FOLD-Dyn project proposes the study of new theoretical approaches for the effective generation of virtual characters deformations, when they are animated. These deformations are two-folds: character skin deformations (skinning) and garment simulations. When producing an animation movie, either for the cinema, television/online series or communication, these deformations are complex to compute and their creation requires hours of work from several specialized artists. In addition, the insertion of garment simulations in animation software requires the use of specific approaches in order to fit correctly within the code and then, in the production pipeline of animation studios. As a result, except for very realistic virtual characters interacting in movies with real characters, and for main characters in productions with high budgets, garment simulations are avoided and character deformations are rudimentary. In fact, quite often, garments are rather deformed using simpler techniques, inspired from skin deformations (and eventually identical). And these, despite a very close proximity between industrials and academics in this area that allows fast transfers when new promising approaches are discovered by researchers.
We propose the exploration of the possibilities offered by a novel theoretical way of addressing character deformations : the implicit skinning. This method integrates the conjoint use of meshes (the standard representation for 3D animations) and volumetric scalar functions (an unusual representation in this community). The strength of this approach is to provide effective solution when meshes reach their limits by exploiting, in these situations, the promising natural properties of 3D scalar functions, such as their ability to generate multiple collision-free iso-surfaces. By improving the theoretical properties of scalar functions, the study of their joint use with meshes and the introduction of a new approach with its formalism, called multi-layer 3D scalar functions, we aim at finding effective solutions allowing production studios to easily integrate plausible character deformations together with garment simulations. Such solutions would have a very significant impact on the realism of the final animations as well as on the quality and the visual attractiveness of the movies. From the academic point of view, the impact will be even more important since these research are a scientific break with the current approaches and they offer new perspectives that were considered as extremely challenging to reach.
This fundamental research project is carried out by an academic consortium formed of specialists in the fields of 3D modeling and animation, garment simulation and more specifically 3D scalar functions for virtual environments. They are members of the Institut de Recherche en Informatique de Toulouse, Inria Grenoble Rhône Alpes and Inria Bordeaux Sud-Ouest. In order to guaranty the relevance of the research directions and the usability of the proposed results, this academic consortium will closely collaborate with industrial experts from the 3D animation industry: TeamTo production studios and the animation software developers from Mercenaries Engineers.
Monsieur Loïc Barthe (UNIVERSITE TOULOUSE III [PAUL SABATIER])
The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.
UPS-IRIT UNIVERSITE TOULOUSE III [PAUL SABATIER]
Inria GRA Centre de Recherche Inria Grenoble Rhône-Alpes - IMAGINE
Inria BSO Centre de Recherche Inria Bordeaux-Sud-Ouest
Help of the ANR 536,615 euros
Beginning and duration of the scientific project: January 2017 - 48 Months