Public Release: 

Disney system uses everyday descriptions of cloth to aid garment design for animations

Researchers translate technical parameters into terms as 'silky' and 'wrinkly'

Disney Research

The elaborate costumes worn by Rapunzel, the heroine of Disney's "Tangled," are testament to the growing sophistication of cloth animation, but for art directors, who must tweak tens or hundreds of technical parameters, achieving a desired look for simulated clothing can be laborious.

To ease this process, Disney Research and Walt Disney Animation Studios have developed a tool that enables technical directors who must create garments that are soft, silky, wrinkly, heavy, or flowing, to use those same terms to control the computer programs that fashion simulated cloth.

The researchers ran a series of experiments in which people were asked to rate the degree to which a garment worn by an animated character matched a particular verbal descriptor. Based on those results, they built a perceptual generative model that enabled users to control a cloth simulation program using these terms. A subsequent series of experiments showed that the model had indeed created garments that people perceived having the desired properties.

"We expect that this methodology will be predominantly used to customize controls of cloth simulators and that it will significantly reduce the amount of time it takes to create a garment that exhibits a desired trait," said Leonid Sigal, a senior research scientist at Disney Research.

The findings from this research will be presented at ACM SIGGRAPH 2015, the International Conference on Computer Graphics and Interactive Techniques, in Los Angeles Aug. 9-13.

Cloth simulators used in creating animated features typically include a number of parameters for controlling the behavior of cloth - parameters such as bend damping, density and shear angle. These parameters loosely represent properties of materials such as wool, silk or chiffon. But parameters interact in complex ways to produce the final behavior of a garment and different combinations of parameters can provide alternate means of achieving the same result. Dialing in the proper combination of parameters to achieve a particular quality sought by a designer or a director is not straightforward.

"Ideally, we would like to have controls for cloth simulators that are intuitive and perceptually meaningful," said Timothy Richards, technical director at Walt Disney Animation Studios.

To map the technical parameters to a set of words commonly used to describe the properties of garments, the researchers began by having professional animators create parameter settings for a set of fabrics: cotton, wool, silk, leather and chiffon. They then created videos, each with slight adjustments of the parameters, that were shown in pairs to adults recruited through Amazon Mechanical Turk. These experiments determined the minimum change in parameters necessary for people to perceive a difference in dynamic behavior of the fabric.

A subsequent series of experiments showed people videos of an animated woman in a skirt walking and spinning; they were then asked to rank them based on the degree to which skirt material reflected a trait such as silkiness, smoothness, or crispness. These rankings were used to learn a model that could map from a degree of the particular trait, or trait combination, to the unintuitive parameters of the simulator. A third set of experiments validated that the incremental changes for each trait resulted in corresponding behavior of the cloth that matched the selected level of that trait.

"While we learned the traits from a single motion - walking with a spin - and one sample garment - a skirt - these learned traits generalize and can be directly applied to other garments and motions," Sigal said.

Mapping commonly used descriptors of garments to sets of parameters is an approach that could be used with any cloth simulator, the researchers said.

###

In addition to Sigal and Richards, the research team included Moshe Mahler, Spencer Diaz, Kyna McIntosh, Elizabeth Carter and Jessica Hodgins of Disney Research.

More information, including a video, is available at http://www.disneyresearch.com/publication/garment-simulation/.

About Disney Research

Disney Research is a network of research laboratories supporting The Walt Disney Company. Its purpose is to pursue scientific and technological innovation to advance the company's broad media and entertainment efforts. Vice Presidents Jessica Hodgins and Markus Gross manage Disney Research facilities in Los Angeles, Pittsburgh, Zürich, and Boston and work closely with the Pixar and ILM research groups in the San Francisco Bay Area. Research topics include computer graphics, animation, video processing, computer vision, robotics, wireless & mobile computing, human-computer interaction, displays, behavioral economics, and machine learning.

Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.