Closing the Circle: Automated tailoring comes closer (The Economist)
Image courtesy boston_camera
“It’s the simple things, like holding scissors, or stitching and measuring. Even though I know exactly where I want to cut and what patterns I want to use, my hands and fingers won’t do what I tell them to any more. I guess that’s only fair enough, they’ve had a tough 70 years! But, you know, I haven’t run out of ideas and I want to keep on designing. So I’ve got to use the tools that let me do that, even if they are exceedingly odd tools!”
Martha Evans held plenty of different jobs in her life – a bartender, sales assistant, graphic designer being just a few – but by her late 30s, she’d settled into a steady career as a freelance artist and clothes designer. She did fairly well, making dresses for shops around Florida using her hands and a sewing machine. In fact I have one of her dresses in front of me, a simple bright red silk dress, with a pretty knot at the neck and worn with a turquoise belt.
Like a lot of other high-end bespoke work, Martha’s occupation was well-protected against automation. Clothing has never been purely about cost or convenience; if that were the case, we’d all be wearing comfortable and cheap grey jumpsuits. No, clothing is all about appearance, and no amount of AR layers or mimics will dissuade people from wanting to impress in the basic physical reality; and what’s true now was even truer in the 30s.
But as we heard from Martha, even though there was still demand for her dresses, she wasn’t able to satisfy her customers herself. So she had a choice to make: she could hire an assistant, she could create that could be easily applied to off-the-peg garments, or she could give up. I imagine she considered all of these options quite carefully until she was introduced to Shakti Nagra, a young woman who wanted to change the way that clothes were made.
Broadly speaking, clothes were made in two different ways before the 30s. The first was Martha’s method – bespoke clothes that required the careful measurement of the customer, then cutting and sewing fabrics accordingly, usually with some final adjustments after a fitting. Then there was mass production, which still required manufacturing samples and a process of trial and error to make sure the final products matched what the designer intended.
Nagra’s process was completely different. As a computer scientist in the teens, she’d made a breakthrough in realistically simulating the behaviour of fabrics – how they hung, how they folded, how they looked when rumpled – using parameters gained from high resolution CT scans. In later years, her team at DEI9 improved their simulations by running fabrics through even more an even more detailed and robot-driven series of tests (for stretchiness, thermal conductivity, breathability, and so on.)
This was all in the service of creating ever-more realistic computer animations for games and movies, driven by the increased fidelity of glasses and the newly-arrived contacts. These animations frequently used ‘real’ clothes, often as part of product placement sponsorships, and so Nagra’s team had to develop software to accurately convert those dresses into 3D models. Shakti Nagra explains her next step:
“I realised that if we could perfectly simulate fabrics and garments – even real dresses and suits you might see in any mall or on any high street – then we could go in the opposite direction. We could design garments virtually, simulate out how they’d look, and then specify every single steps involved in manufacturing those garments. It would a one-hundred percent digital process, end-to-end.”
Armed with tens of millions in investment money, Nagra set out to accomplish her goal of making clothes entirely digitally. Almost immediately, she ran into two major problems.
The first was that the clothing industry – particularly manufacturing – was still very much human-powered. The kinds of complex and delicate manipulations that clothes have to undergo to be cut and sewn properly were extremely challenging for robots at the time; it could be done, but only at a significant premium over human labour. Nagra was extremely pragmatic and delayed her ambitions of an entirely robot-driven process and focused on creating better quality control systems for her workers, until – in her characteristically unsentimental way – costs fell to the point where she could begin automating their work.
The second problem she faced involved the design process. It was one thing for experienced 3D modellers to design clothes within a fully virtual environment; it was a completely different matter for people who’d been trained in a non-digital way.
That’s where Martha came in – Nagra hired her as a consultant to help create the tools that designers would use. Instead of the austerely technical interface that had been planned, Martha opted for a highly skeuomorphic system that used physical gestures and movements mimicking those she’d used decades ago. Designers would put their fingers into the shape of scissors to cut, and could virtually sew their clothes at high magnification, all using thimble and ring absolute positioning. Their collaboration saw them become close friends. As Nagra said, “We had very different backgrounds and I hadn’t imagined that we would have anything in common beyond clothes, but Martha really opened my eyes to the artistry of the process.”
To modern eyes, the interface looks hopelessly old-fashioned, but it was what designers of the time were familiar with – and that was all that mattered. Within two years, DEI9 had attracted tens of thousands of young and old designers who were using her service to design, manufacture, and sell their clothes across the world. Later versions of the software allowed for clothes to be automatically redesigned according to customers’ 3D body maps, massively reducing wastage and returns at a stroke.
Naturally, clothes designs were quickly pirated. “I had predicted this would happen, just as it did with music and film and other media, so of course we had a strategy to deal with it,” said Nagra. This strategy involved creating accessible tools that would allow users to edit and customise existing fashions in a faster and more convenient way than otherwise, with a small tip going to the original designer, promoting free agency. What she didn’t predict was the repeated waves of clothing fads and the rapid fragmentation of fashion into smaller circles.
Ordering clothes from DEI9 didn’t have quite the same instant gratification or social fun that going to a mall did, but it gave you access to a much wider variety of styles and, of course, was much, much cheaper (for one thing, they didn’t have to pay for retail space). And in later years, when robotic technology had advanced further, even complex custom orders would take only a day to arrive via UCS or other couriers. Eventually they expanded into textile tents, wrappings, and temporary buildings.
Ultimately, Nagra’s vision gave freedom to designers. As for Martha, her contract prohibited her from talking about her consulting work with DEI9, but she was given favoured access to the company’s programmers and resources, and allowed to sell her own clothes. And that’s brings me back to her simple bright red dress in front of me. It’s very similar to a previous dress she’d made herself by hand, but it was made, as she said, “not in a better way, just in different way.”
Martha kept on designing clothes until she died, at the age of 96 – and Nagra always made sure that the right tools were there waiting for her.