This project aims at generating a sequence of images of deformable surfaces, such as clothes, fluid surface, papers, and living organs, that are sufficient convincing to our visual perception. Our compromise with intuitive interactions, computational costs, and realistic pictures is prior to being faithful to the scientific foundations.

There are several approaches going beyond classical geometric modeling techniques to allow representing non-rigid objects, also known as flexible or soft objects. Some of them are less based on scientific principles, such as grammar-based models, particle systems, and blobby objects. Others rely on the physical properties of the modeling objects to determine its varying shape, such as physically-based models.

After performing investigations on

• modeling sculptural objects as isosurfaces of a collection of scalar field sources and
• representing the dynamics behavior of a surface using the degenerated solid mechanics approach proposed by Terzopoulos, Platt, Barr, and Fleischer,

we decided to model a cloth as a simplified version of the Cosserat surface, which takes into consideration the geometric compatibility between the metric and bending variations.

Differently from considering surfaces as degenerated solid objects and handling the problem as a 3D one, the Cosserat surface is a collection of points embedded in a three-dimensional Euclidean space, to each of which a "deformable" vetor is assigned. Although the first approach is usually better conditioned for numerical implementations, the latter includes explicitly the relation of the motion and the geometric parameters, once the relative displacements of surface points are governed by a set of constitutive equations which include the equation of balance of energy and the fundamental theorem of surface theory due to O. Bonnet. Simulations of a variety of cloth bendings in a diversity of meterials have been achieved. We also observed that a user interface would benefit greatly from these physical and geometric links, once they make easier to narrow the gap between the obtained effects and the user expectation at each manipulation.

Currently, we are investigating a procedure for appropriately handling collisions in our proposed cloth model.

Any comments about this project will be very appreciated.