Today, digital data such as 2D images, 3D meshes and 3D point clouds are widely used to design virtual environments (VE). Most of the time, only one type of those multimodal data is used to describe and specify the shapes of the objects. However, a single object can be seen as a combination of components linked with constraints specifying the relationships and the rigid transformations defining their arrangement. Thus, the definition of new methods able to combine any kind of multimodal data in an easy way would allow non-experts of VE to rapidly mock up objects and scenes. In this paper, we propose a new shape description model together with its associated constraints toolbox enabling the description of complex shapes from multimodal data. Not only rigid transformations are considered but also scale modifications according to the specified context of the constraint setting. The heterogeneous virtual objects (i.e., composed by scalable multimodal components) then result from the resolution of a constraint satisfaction problem through an optimization approach. The proposed approach is illustrated and validated with examples obtained using our prototype software.