Inferring 3D shapes from sketches is not sufficient for creative design: the depth added by the system may be wrong. In addition to correcting it, the user will need in many cases to finely tune the shape, globally or locally. To this end, we need to provide continuous deformation tools inspired by sculpting metaphors. Instead of being guided by 2D strokes, these tools will be controlled by intuitive design gestures, such as those enabled by the new multi-touch tables (WP5).
Our previous work in the area enabled us to identify two types of design gestures: adding or suppressing material, versus deforming the shape, for instance by bending or twisting it. In the second case, deformations are to translate into changes of morphology, to give a plastic behaviour to the shape. Both kinds of deformations were already proposed for solids (Cani, 2008) but general methods need to be introduced for complex, composite objects which are very challenging. Let us take some examples. Uniformly stretching a table with plates and glasses would obviously not work. The user expectation may rather be that the table geometry stretches, but that more objects, obeying the same distribution, are created. For a user, deforming the bounding volume of a piece of crumpled paper should probably increase the amount of paper, while maintaining its developable features. Elongating a wisp of hair should lengthen the strands, while enlarging it should rather add some more strands. Whatever the difficulties, our goal is to revisit deformations from such a user-centred perspective.
Scientific problem: define two types of deformations for composite, responsive shapes:
- Extensive deformations: such deformations add or remove material by acting on the extensive parameters of a shape (length, area, volume, number of details) while keeping the intensive morphological parameters constant.
- Intensive deformations: such deformations are aimed, on the opposite, at editing the intensive parameters of a shape such as curvature, while keeping the amount of material to a constant value.
Our first results :
We developed a novel methodology for sculpting structured shapes such as buildings thanks to mutable elastic models [Millez 2013], developed another system for multi-dimensional nested structures [Stanculescu 2012], and introduced two novel transfer approaches for responsive shapes: one for transferring garments from one character to another one [Brouet 2012], and a second method for transferring their internal organs [Dicko 2013]. These are the bases for interactive deformation of these complex responsive shapes.