Features TechWear

Bridging the Gap: Fashion Design and Digital Technology

I teach in the Design and Technology (DT) programs at Parsons School of Design in New York. Students and faculty in DT undertake work in a variety of areas; interaction design, app development, immersive media, wearable technology, game design, motion graphics, computer vision and machine learning are a few. It’s a big, broad program, more than two decades old, that is well-established and respected. We are probably best described as a digital media program with an emphasis on design process, testing and making things real. We believe that students learn differently when they are required to build something, and that by being able to test real products, they can feed the results back into their design process. This means DT students get heavy computational training (coding), and they are expected to use it in their projects.

Why am I describing Design and Technology in an article about fashion technology? Well, a lot about DT is similar to the ways young fashion designers are taught (with the exception of code); very hands-on, very process-oriented, with a focus on making things real — process, materials, fabrication and user testing. But DT at Parsons is completely separate from the Fashion Design program. We have long aspired — but struggled — to create a viable collaboration between Fashion Design and DT. With the digital world exploding around us, you would think a relationship between fashion and digital media would be easy and obvious. It hasn’t been. Fashion Design is a domain and a community of practice many DT students come to Parsons with an interest in. In a place like Parsons — with a legendary fashion design presence and a large fashion school — I am constantly scheming ways to create a link to DT.

For fashion design, the introduction of digital technology has always presented a huge challenge to traditional approaches (as taught). For those of us in DT, buried in the world of the computer and the digital, it has often been as though we were looking over at Fashion Design from across a deep divide. In DT, we are experimentalists at heart, with a focus on constant iterative prototyping, embracing and discarding technology as it evolves. We use digital platforms as accelerators to augment or amplify the creative process, to speed it up. Often, our products are technological in their own right. But when we can get the fashion designers to talk about technology, it is often just as a tool — either in the design process or as an element of fabrication (CLO, Gerber, Lectra — they are taught, but almost always as practical tools, not idea generators). With the advent of the wearable technology two decades ago, there was sudden hope — here was a way to bring fashion design and creative technology together, without giving up the best of either. Alas, it was not to be (or has not yet become). Wearable tech, because of its dependence on sophisticated engineering, didn’t bridge the divide. If anything, it has increased it. What is to be done?

I believe that in the slow but incessant merging of the fashion world with the world of digital media (and vice versa), and the increasing inability to differentiate between those domains, there lies an answer. From a purely digital perspective, media is increasingly driven by new technological fronts such as artificial intelligence, immersive reality, data, robotics and security/privacy. There may be no obvious connection with traditional fashion design. But look further. What if, in cycling through the multilayered world of digital media back to traditional fashion design, there are ways of leapfrogging the hurdles seemingly so present when trying to combine the digital with traditional fashion design approaches? What if the application of AI to the process of fashion design can produce engagement and experiments that circumvent the traditional tendency of fashion design to view digital platforms as “just tools?”

Many might say I am suggesting something that would render the design process conceptual, but let me provide two tangible examples that bring the discussion back to real possibilities.

The first is object recognition (OR). This technology is based heavily on computer vision and pattern recognition and is central to current research in areas such as automated driving, facial identification and retail experience. OR algorithms have increased greatly in power and effectiveness as visual datasets, upon which these algorithms depend to learn, become larger and more refined. We are now at a point where both still photographs and videos of (human) social situations can easily be scanned with fair accuracy for types of garments, fabric and accessories. What if this data is evaluated for tendencies of human interaction — how people wearing specific clothes, or sporting particular accessories, tend to interact with other people? How might these interactions influence fashion design approaches?

In this example, I am not talking about using digital technology to simply replace elements of traditional fashion design, as might be the case when fashion design curricula implement courses in digitally driven pattern making, draping, rendering or visualization (always in the service of traditional and well-established design approaches). Are there certain to be productive results from this type of interaction analysis? Absolutely not. In fact, the results may end up being far more interesting and meaningful to sociologists or psychologists studying human social interaction than to designers seeking insight on why they should move their aesthetics or functionality in a particular direction. But the experiment is worthwhile, and we come up with it by borrowing from digital technologies emerging in the media realm, not the fashion realm.

My second example is in the domain of machine learning, more specifically Generative Adversarial Networks (GANs). This is a technology that, again, uses large datasets in training itself to evaluate the “correctness” of results. A great deal of work (some practical, some experimental) using GANs has been undertaken, often using 2D imagery data. On the practical side, projects have been undertaken to improve typeface readability, to evaluate web style consistency, and to produce pleasing patterns for surface design (textiles). One interesting tendency of GAN experimentation is allowing focus on collaborative human/computer decision making; because visual GAN results are often abstract, they require human interpretation.

GANs themselves do not usually produce practical end results (although they can). Best results are often produced when humans and GANs work together with ideas. Herein lies an example of the often-referenced fact that AI contains the biases of their programmers, but in this case, these are biases designers can appreciate. Recently, 3D datasets (primarily used in game design and digital film design) are being experimented with using GANs. So let’s apply it to fashion design. With increased capacity to collect 3D body/garment data and scanning, could we run a GAN on a recent apparel collection to suggest next or alternative steps, or simply to ideate in directions not obvious to us? Again, the application of technology from the media world to fashion, supporting a traditional design process using alternative application of digital tech being developed outside of fashion.

These are two examples, albeit highly experimental, of how circling back from media-centric digital domains could open up new bridges between fashion design and creative technology. I have no idea where these approaches could lead. I am convinced we will find many examples of these kinds of experiments as we brainstorm around the blending of fashion and media. For me, having been frustrated for so long in getting digital media and fashion designers to collaborate, I am excited by the possibilities. Readers will have many of their own examples I have yet to learn about. I look forward to the discussion.

Sign Up for Newswire

    [ctct ctct-765 type:hidden 'Fashion Mannuscript Newswire::#156']