Robust Methods for the Physically-Based Animation of Large Deformations in Computer Graphics

• Applicant: Professor Dr. Jan Stephen Bender
• Subject Area: Image and Language Processing, Computer Graphics and Visualisation, Human Computer Interaction, Ubiquitous and Wearable Computing
• Term: since 2016
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 281466253

The overall goal of this research project is the development of robust and efficient methods for the physically-based animation of large deformations in computer graphics applications. In the current project phase, micropolar material models should be developed for the simulation of volumetric solids and deformable shells. In micropolar models, each node has additional rotational degrees of freedom. This enables a better representation and control of the bending and torsion of a body. Initially, our research group developed a micropolar model for volumetric solids with explicit time integration. When implementing an implicit method, it became clear that the derivatives according to the rotational degrees of freedom are very complicated and their implementation is error-prone. In order to develop and test further micropolar models more quickly, we have designed a system that automatically calculates the required derivatives by symbolic differentiation. The system generates efficient code for each derivation, which is then automatically integrated into the simulation. This system was then used to develop a micropolar material model for elastic shells. In the remaining time of the current project phase, we investigate the use of higher-order implicit time integration methods for the micropolar models. This research proposal is intended to extend the project for another 12 months. During this time, the use of higher-order finite elements for micropolar solids and shells will be investigated in more detail. It has already been shown in the current project phase that these elements can be used to avoid so-called locking effects, in which the material behaves significantly stiffer than defined by the material parameters. However, the Lagrange elements typically used are not globally continuously differentiable, which leads to "kinks" in the displacement field between neighboring elements. These kinks are problematic at the surface as they lead to visual artifacts and negatively influence the calculation of contact and friction forces. This will initially be solved using special elements that directly fulfill C1 continuity requirements. However, since such elements are computationally expensive, we will also develop a method to minimize the problematic kinks when using the cheaper Lagrangian elements. Furthermore, we will investigate whether the method can also be applied only to surface elements in order to further improve the runtime. This application for the final project phase is intended to successfully complete the overall project and further establish the developed material models for the simulation of large deformations in computer graphics.

• DFG Programme: Research Grants