News + Events

09/23/2015

Watching the Pixar movie Up! a few years ago, Associate Professor of Textiles Brooks Hagan MFA 02 TX first began noodling over a seemingly simple question. How do animators make make-believe fabric look so convincing? By the time Brave came out in 2012, he couldn’t stop thinking about it.

If the virtual textiles—“the tapestries, tartans, furs… even the hair, leather and metal”—in these films can look and behave so realistically on screen, why can’t Hagan and his students create equally accurate digital renderings of textiles? It’s a question he set out to answer four years ago—initially by searching for information about the technicalities involved and then by reaching out to scientists working at the forefront of verisimilitude in computational space.

Now that Hagan has become a principle investigator (PI) on a research project that just won a $1.2-million grant from the National Science Foundation, he’ll continue to probe for answers over the next four years, too. During the grant period—which officially began on September 1 and runs through August 2019—he will use the $244,236 allocated to RISD by the NSF’s Cyber-Human Systems program to lead the Virtual Textiles Research Group, which will include selected graduate students who will support his work with the PI team—two computer scientists at Cornell and a scientist at Stanford with expertise in mechanics.

The goal is to teach computers as much as possible about the properties of textiles, the mechanics of weaving and knitting, and how fabric drapes, falls or crumples to allow for “super realistic simulations that look and behave like the real thing,” Hagan explains. “If we’re successful, it will not only revolutionize the way we design textiles,” he adds, “but will open up access to the industry and modes of manufacturing.”

Gnarly problem
While recent advances in computer rendering have led to an explosion of virtual prototyping in all sorts of fields, intricately constructed woven and knit textiles are more difficult to render accurately, especially given subtleties of appearance based on light, shape and motion. As suitable as it may be for Pixar and other filmmakers, “the state of the art in simulating cloth … is not predictive of real appearance and therefore not usable in a design process, in which realistic but inaccurate results are useless,” the co-investigators point out in their NSF abstract.

Given the challenge, Hagan was thrilled when he “found experts in a somewhat arcane subspecialty within computer graphics, two of whom are Academy Award winners for their work on things like Gollum’s skin in Lord of the Rings.” Once he cold-called Cornell Professor Steve Marschner in 2011 to explain the need from a textile design perspective, they immediately hit it off and got mutually excited about the real-world potential for developing this new technology. Though their NSF grant proposals didn’t make it through the rigorous competition the first three years they tried, they got closer each time and finally got the green light this round.

“I’ve had my eye on this and have been teaching state-of-the-art CAD programs for a long time,” Hagan explains. But despite working with the best software and amazing equipment—like RISD’s much-loved jacquard loom in the Metcalf Building and the Stoll industrial knitting machine in the College Building—students share the same frustration as professionals in the field: They spend enormous amounts of time fine-tuning CAD files, yet when they finally go to produce a prototype on a loom, “things come out of it that are entirely unexpected.” That’s not necessarily a bad thing, Hagan is quick to point out, but it makes the process of producing fully realized fabric that adheres to the designer’s vision excessively time-consuming and costly. In addition, because they can’t accurately “see” what they’re designing, students end up iterating based on limited information rather than on accurate simulations of textile designs that are able to capture dynamic properties such as lighting and draping.

Weaving new knowledge
As part of the four-year Virtual Textiles Research project, the Cornell team will develop new models and algorithms that will be integrated into a software program for fabric design and prototyping under development at RISD. “We also have a manufacturing partner that has signed on to the project to help produce a more varied array of textile samples,” Hagan says. As RISD designers beta-test the software and produce new textiles, they’ll ship them to Cornell, where computer scientists will study changes in the design process resulting from enhanced capabilities in the program and tweak accordingly. The Stanford researcher will simultaneously offer his expertise on how to simulate the mechanical aspects of fabric.

The circular exchange of information will continue as the PI team continues to teach computers more about the many variables involved in designing, prototyping and producing constructed textiles. That involves the painstaking process of comparing appearance and deformation models to measurements of textiles from CT scans, imaging and other modalities, making adjustments and improvements along the way so that they’re able to produce predictive working models that allow designers to see what their fabric will look like – as a swatch, yardage draped over an object and fabric applied to clothing, accessories, furniture and other items. “A major focus of the research is on validation, testing both the accuracy of the individual technical pieces and the ultimate usefulness of the new techniques,” the PI team notes in its proposal.

While several RISD faculty members have worked on NSF-funded research projects in the past, Hagan is the first to win funding with RISD as a lead institution and in support of a project of direct benefit to artists and designers. His award signals a new direction for faculty-led research at RISD, building on two other NSF awards and the NASA/RI Space grant awards that have inspired a long-running series of studios in Industrial Design.

“A lot of this has been talked about and theorized over the last couple of decades,” Hagan says, “but we’re finally reaching a point where computational speed is allowing us to look at textiles very realistically and to render them physically so that they look the way they are in the real world.” In the world of design, that makes all the difference.