A functional garment using 4D embroidery with heat-shrinking thread to activate puffs that enable air trapping, thus warming up the wearer.
By Rosalie Lin
Advisor: Skylar Tibbits (Founder of MIT Self-Assembly Lab),
Jack Forman (PhD student at MIT Media Lab)
Teaching Assistant: Ganit Goldstein
This project started by experimenting with machine embroidery and heat-shrinkable thread. A series of the material libraries are developed in light of the combinations of ‘static’ fabric and ‘active’ embroidery pattern generated by processing, which enables fine-tunable parameters such as stitch-spacing and stroke-width. Bending, folding, curving, and puffing are some key transformations discovered throughout the processes.
RandomPuff is a garment application leveraging the puffy transformation that traps air within the dome pocket to keep the body warm. The double curvature geometry becomes easily fabricated, and personalizable styling is available through the computational design process, which opens up a new design/manufacturing space for a puffy garment.
( Check out video for more technical details )
Starting from my working experience at the NIKE factory, a challenge remains when it comes to transforming 3D to 2D or vice versa. From 3D model, pattern making to assembling a shoe, there are too many processes involved in fulfilling a shoe and each process is kind of independent. There’s no room for adjusting the object in the middle of the process since the materials are all static. From a manufacturing point of view, what if a piece of fabric can be toggled in multiple states with various activation methods, such that the cosmetic or geometry transformation can be easily adjustable in between the prototyping process? From an end-consumer point of view, what if a garment can be fabricated in flat for compact shipping, and end-consumers can activate it with more volume and texture in favor of personal style preferences?
There are digital and physical components playing in this project, especially very similar to 3D printing, where ‘g-code’, usually generated by slicer, acts as a bridge to communicate between computer-aided design and machine. Similarly, in this project, I used PEmbroidery, an open-source library housed in processing to generate embroidery pattern design, and export the machine readable files in one step. This process kills the middle software like inkstitch, or slicer in 3D printing context.
The material system can be briefly divided into material and parameter categories. For the materials, there are active and passive components. The active one is the heat-shrinking thread while the passive one can be any kind of substrate as long as it is machine embroiderable. Talking about substrates, you really can try endlessly. I’ve tried around 10 different kinds of substrates, from thin to stiff, from cotton to plastic. Generally speaking, the stiffer materials work better with heat-shrinking transformation. Neoprene is my go-to material after tons of trials. For parameters, I started from very basic, linear patterns from micro to macro scale tunability. The order that explored is from micro scale: density, length, stroke width, stitch spacing, then extend some interesting findings to macro scale geometry.
Macro pattern - End effect
Well, maybe a better way to describe macro patterns could be ‘end material behavior’. There’s no clear order of how to explore these patterns. Sometimes I set the end goal first and reverse-engineered the possible embroidery pattern. Sometimes, by exploring enough basic patterns, I can gain a sense of how the material behaves. For example, the origami effect came out of open linear curves, while the puffy effect is from closed spiral curves.
Starting from a linear pattern, I explored different densities. Different curvatures can be formed with stiffer materials like neoprene, but almost nothing changes with the non-woven substrate.
To further elaborate on linear patterns, I explored heterogeneous patterns with some sketches as predictions and got this very pleasing origami-like geometry.
This is actually my initial idea where I want to make a spacer layer to regular body temperature. The reason for enlarging the stitch space is to minimize the constraint from fabric. To enlarge even more, I was able to get an overhang-like effect, where the stitch no longer looks like stitches.
The intention of this trial is to get a sharp crease as opposed to bending. Starting from putting shrinking threads on the creasing line, the forces ended up being not strong enough, I then pivoted to using the entire ‘plate’ instead of ‘line’ as actuation unit, and it worked!
Finally, after trying a bunch of open, linear curves, why not try some closed curves? The puffy effect is matching my hypothesis with slightly variation with different fabric