[ Page 2 of 3 ]

By Alan Joch

Machines that can think

Ten years ago, Jhane Barnes, a textile designer, met Bill Jones, a mathematician at Syracuse University, at a textiles convention. Jones was exhibiting software he had developed that created patterns using mathematical algorithms. Barnes was intrigued. “He was sitting in a booth, and all these patterns were flashing across a computer screen,” she recalls. She purchased a copy of the program and soon found it produced designs different than any she had previously seen. Eventually, she hired Jones and a colleague as full-time software designers who constantly refine the tools she uses. Today, Barnes is a highly successful textile designer who has established a reputation for visually complex, compelling patterns. She has worked with many leading carpet and textile manufacturers, and estimates her studio spends more than $100,000 a year for software development, tools that she keeps in-house.

Textile designs are nudged along, pixel by pixel, and then produced in various colors.

Barnes uses the computer as a kind of electronic sketch pad. She may start, for instance, by instructing the software to draw diagonal lines at certain angles and thicknesses, and in various colors. The arrangements may be entirely new, but often they’re saved versions of patterns from former projects, which she calls “generators.” The generators operate as an expert system, combining her live input with subject-specific rules programmed into the software. “For example, if I wanted to draw diagonal lines at a certain angle, I’d type in the angle and tell the software how thick I wanted the lines to be,” she explains. “Thicknesses can be a range, from, say, two pixels to seven pixels. Then I choose the spacing of lines and the colors. Every time I hit ‘play,’ I get a new design based on those rules.” For additional effects, Barnes may layer multiple generators on top of each other. She can save iterations she likes and record her progress along the way, making it easy to create related yet slightly different patterns. Barnes also tailors her designs with “modifiers,” filters she’s created that can skew lines into wave patterns or alter the initial design in some other way. At times, she already has a pattern formed in her mind’s eye, and her task is to instruct the software to produce it. If the right generator to produce it doesn’t exist in her library, Barnes can connect to Jones’s computer in upstate New York so they can work together on the solution.

Even when the patterns the software generates aren’t what she pictured, she considers this a welcome outcome. “[The design] may be neat, anyway. I’ll use it someday for something.”

ADVERTISEMENT