Estimating Cloth Elasticity Parameters Using Position-Based Simulation of Compliant Constrained Dynamics
PubDate: Dec 2022
Teams: Meta Reality Labs Research
Writers: Egor Larionov, Marie-Lena Eckert, Katja Wolff, Tuur Stuyck
Abstract
Clothing plays a vital role in real life and hence, is also important for virtual realities and virtual applications, such as online retail, virtual try-on, and real-time digital avatar interactions. However, choosing the correct parameters to generate realistic clothing requires expert knowledge and is often an arduous manual process. To alleviate this issue, we develop a pipeline for automatically determining the static material parameters required to simulate clothing of a particular material based on easily captured real-world fabrics. We use differentiable simulation to find an optimal set of parameters that minimizes the difference between simulated cloth and deformed target cloth. Our novel well-suited loss function is optimized through non-linear least squares. We designed our objective function to capture material-specific behavior, resulting in similar values for different wrinkle configurations of the same material. While existing methods carefully design experiments to isolate stretch parameters from bending modes, we embrace that stretching fabrics causes wrinkling. We estimate bending first, given that membrane stiffness has little effect on bending. Furthermore, our pipeline decouples the capture method from the optimization by registering a template mesh to the scanned data. These choices simplify the capture system and allow for wrinkles in scanned fabrics. We use a differentiable extended position-based dynamics (XPBD) cloth simulator, which is capable of real-time simulation. We demonstrate our method on captured data of three different real-world fabrics and on three digital fabrics produced by a third-party simulator.
