We propose a method for learning linear upsampling operators for physically-based cloth simulation, allowing us to enrich coarse meshes with mid-scale details in minimal time and memory budgets, as required in computer games. In contrast to classical subdivision schemes, our operators adapt to a specific context (e.g. a flag flapping in the wind or a skirt worn by a character), which allows them to achieve higher detail. Our method starts by pre-computing a pair of coarse and fine training simulations aligned with tracking constraints using harmonic test functions. Next, we train the upsampling operators with a new regularization method that enables us to learn mid-scale details without overfitting. We demonstrate generalizability to unseen conditions such as different wind velocities or novel character motions. Finally, we discuss how to re-introduce high frequency details not explainable by the coarse mesh alone using oscillatory modes.
We thank all the reviewers for their feedback and helpful comments. Many thanks to our colleagues Olga Sorkine, Alexander Hornung, Bernd Bickel and Peter Shirley for careful proofreading and Daniel Sýkora, Robert Sumner, Edilson de Aguiar and Leonid Sigal for stimulating discussions. The human models were created by Peter Lozsek and were generously provided by Trinity College Dublin. We also thank Paul Johnson, Jeff Bunker, Brian Christensen and Yong Wan for help with modeling and art feedback and Rasmus Tamstorf, James O'Brien, Huamin Wang, Wei-Wen Feng and Pascal Volino for sharing their expertise on cloth simulation.