Modeling hybrid virtual objects using scalable multimodal components and rigid transformations

The specification of digital objects, assemblies and scenes strongly relies on the use of constraints describing how shapes have to be combined all together. Constraints can be applied to the low level geometric entities or the parameters of the geometry, as in [Hsu97] and [And95]. They are also used to set position relations between different geometries considered as rigid bodies as in [Mar08] and [Kra91]. In both cases, they act on the same type of geometry either 2 or 3D. In this paper, we extend the classical paradigm to a more general concept so as to be able to apply them simultaneously on multimodal data, i.e. on different digital representations (e.g. 2D raster image, 3D point cloud, 3D mesh) obtained through different approaches (e.g. drawing, picture, 3D modeling, 3D scanning