Marie-Lena Eckert

Egor Larionov, Marie-Lena Eckert, Katja Wolff et al.

Generating realistic clothing for virtual applications like online retail and digital avatars is crucial but requires expert knowledge of 3D tools to generating believable simulations. Recently, a number of works proposed to estimate cloth material properties from specialized capture setups. However, these systems tend to be monolithic, complex and expensive. We propose a simplified method for automatically determining parameters based on easily captured real-world fabrics. While existing methods carefully design experiments to isolate stretch parameters from bending modes, we embrace that stretching fabrics causes wrinkling and propose a novel specialized loss for comparing wrinkled fabrics. We designed our objective function to capture material-specific behavior, resulting in similar values for different wrinkle configurations of the same material. We estimate bending first, given that membrane stiffness has little effect on bending. We use differentiable simulation to find an optimal set of parameters that minimizes the difference between simulated cloth and deformed target cloth. 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 demonstrate our method on captured data of three different real-world fabrics and on three digital fabrics produced by a third-party simulator.

Marie-Lena Eckert, Wolfgang Heidrich, Nils Thuerey

We present a novel method to reconstruct a fluid's 3D density and motion based on just a single sequence of images. This is rendered possible by using powerful physical priors for this strongly under-determined problem. More specifically, we propose a novel strategy to infer density updates strongly coupled to previous and current estimates of the flow motion. Additionally, we employ an accurate discretization and depth-based regularizers to compute stable solutions. Using only one view for the reconstruction reduces the complexity of the capturing setup drastically and could even allow for online video databases or smart-phone videos as inputs. The reconstructed 3D velocity can then be flexibly utilized, e.g., for re-simulation, domain modification or guiding purposes. We will demonstrate the capacity of our method with a series of synthetic test cases and the reconstruction of real smoke plumes captured with a Raspberry Pi camera.