Novel representations for 3D cloth simulation
Abstract/Contents
- Abstract
- 3D cloth simulation has many practical applications ranging from movies and games to VR/AR and e-commerce virtual try-on. While researchers have made significant advancements over the past three decades, realistic simulation of 3D clothing remains challenging, especially when it comes to bending, folding, and wrinkling. Furthermore, physics based approaches are inherently limited in their ability to capture real-world behavior due to insufficient understanding and inaccurate modeling assumptions. This dissertation addresses some of these challenges by proposing new representations for 3D cloth simulation including virtual finite elements for bending and creasing, inequality constraints via optimization for wrinkling, and convolutional neural networks leveraging data. First, we propose a new virtual finite element method for producing sharp folds and creases commonly seen in clothing, e.g. pleated skirts and dresses. Based on a control curve specified by an artist or derived from internal stresses of a simulation, we create a piecewise linear curve at the resolution of the mesh. Then, we cut the object along the curve, creating virtual finite elements and thus new degrees of freedom, while subsequently reattaching the resulting pieces so that adjacent pieces may only rotate or bend about the cut. Notably, this representation requires minimal extra simulation cost as compared to the alternative of mesh refinement. Next, we address the notorious ``locking'' problem, where a cloth mesh that lacks sufficient degrees of freedom to fold and wrinkle behaves more rigidly than desired. We observe that ``locking'' is due to the well-accepted notion that edge springs in the cloth mesh should preserve their lengths, and instead propose an inequality constraint that stops edges from stretching while allowing for edge compression as a surrogate for bending and sub-edge wrinkling. We formulate a constrained optimization problem for our inequality cloth and present a numerical method that can be readily incorporated into existing codebases. We demonstrate the efficacy of this approach in a variety of examples when it comes to folding and wrinkling, especially on coarser cloth meshes. Finally, with the aim of efficiently creating virtual cloth deformations more similar to real world clothing, we leverage data to augment and improve physics based methods. Specifically, we propose a new representation framework that recasts 3D cloth deformation as an RGB image in the 2D pattern space that can be interpreted as displacement maps from the body surface obtained via a procedural skinning prior. Thus, a 3D cloth shape is equivalent to a 2D displacement image, which in turn is driven by animation parameters such as joint angles. This framework allows us to leverage popular CNNs to learn cloth deformations in the image space, and preliminary experiments show promising results on the task of predicting 3D cloth shapes from skeletal poses. In particular, we illustrate that our approach does not require accurate unclothed body shapes or robust skinning techniques. Additionally, we discuss how standard image based techniques such as image partitioning for higher resolution can readily be incorporated into our framework.
Description
| Type of resource | text |
|---|---|
| Form | electronic resource; remote; computer; online resource |
| Extent | 1 online resource. |
| Place | California |
| Place | [Stanford, California] |
| Publisher | [Stanford University] |
| Copyright date | 2019; ©2019 |
| Publication date | 2019; 2019 |
| Issuance | monographic |
| Language | English |
Creators/Contributors
| Author | Jin, Ning (Jenny) |
|---|---|
| Degree supervisor | Fedkiw, Ronald P, 1968- |
| Thesis advisor | Fedkiw, Ronald P, 1968- |
| Thesis advisor | Bohg, Jeannette, 1981- |
| Thesis advisor | Yeung, Serena |
| Degree committee member | Bohg, Jeannette, 1981- |
| Degree committee member | Yeung, Serena |
| Associated with | Stanford University, Computer Science Department. |
Subjects
| Genre | Theses |
|---|---|
| Genre | Text |
Bibliographic information
| Statement of responsibility | Ning Jin. |
|---|---|
| Note | Submitted to the Computer Science Department. |
| Thesis | Thesis Ph.D. Stanford University 2019. |
| Location | electronic resource |
Access conditions
- Copyright
- © 2019 by Ning Jin
- License
- This work is licensed under a Creative Commons Attribution Non Commercial Share Alike 3.0 Unported license (CC BY-NC-SA).
Also listed in
Loading usage metrics...