Designing Machines for Human-Wind Collaboration

An exploration into new mechanisms for jacquard weaving, as well as an ongoing interest in asking how non-human materials or forces can be engaged as collaborators resulted in the prototype of the wind loom—a modified tapestry loom that with every 4th warp connected to a sail that moves the warp position in and out. The fabrication of the loom was led by Jen Mah and Rachel Bork, who iterated between several prototypes for laser-cut heddle/hooks that can be attached to the yarn, arms are connected to umbrellas that can move when the wind blows, easily attachable and detachable components to support easy travel, and so on. The prototyping process was complex and frustrating, as the summer in which we prototyped was not very windy and it was hard to test in its specific working conditions we imagined for the loom. Local weaver, and friend, Julie Rodriguez, took the prototype out for a test and captured the photo above. Her approach was to wait for a gust, and then weave into the wind-produced shed with alternating colors that she chose.

The image on the left shows a detail of the arm design and how it connects to the warp yarns. First, a small piece of laser-cut acrylic attaches to the yarn. We cut a small “pig tail” in the acrylic to allow the small rectangular joint to be easily wound on and off the warp. That piece then fits into an arm that pivots around a post along the width of the loom. That arm has a long thin aluminum post attached to the back of it, upon which a we attach umbrellas made with nylon and 3D printed attachment joints. The length of the aluminum post determines the speed a which the umbrella will move. The longer the post, the weaker the wind. We found the weakest wind we could capture was roughly 6 mph. We originally had a second 3D printed joint between the metal post and the umbrella post (seen below) but this added to much weakness into the joint and became unruly in a wind gust.  All of our part files and instructions can be found on this Instructables post:

The back view of the loom during development. We staggered the umbrella height in order to activate certain warps at certain speeds and to also make sure all of the umbrellas could fit into the space.

To prevent the arms from moving in directions not along the direction of the wind, we had to fabricate supporting “tracks” out of acrylic. These held the arms in place without inhibiting their movement (as we found with packing spaces with metal rings.). To help someone easily assemble and disassemble and travel with the loom, we made this track fixed by magnets, which worked well to hold it in place.

Here is a front view of the loom in development, with a bit of a tester tapestry in production based on some hypothetical wind data.

The insights from this is that the machine for weaving with the wind is incredibly frustrating to use. Mostly because the whole apparatus is large and the design of the umbrellas only catches wind traveling in one direction. While I initially considered redesigning to a smaller form, and reworking the umbrella design, I enjoy the idea of this machine being one that invites frustration. If the goal is to collaborate with an uncertain and unpredictable force, maybe frustration is precisely what ought to occur. Maybe it helps us confront our own limitations.

Weaving a Smart Textile

We used a GoPro to capture each step in the process of weaving a smart textile and compiled roughly 6 weeks of work into this video. We show the two tapestries that emerged from this weave, one that didn’t work so well and the other that did (see force fabric post below). in both cases, we were attempting to weave structures that could be used to sense force and tigger color changes in response.

Capturing Moments when our Body Meets Another

This is a first prototype of a vision of a force-fabric. When integrated into a garment, this textile could capture and replay how your body made contact with other bodies in the world. Those bodies may be human, created through the experiences of hugs or holding children, but they may also be of nonhuman forces – heavy winds or couches pressed upon ones back. The concept is to think of ways technology can make us aware of how we are physically supporting and supported by other objects and environmental forces. It sees garments as a interesting surfaces of intersection between self and other.


We created this first textile by double weaving sections of color changing yarn (resistive heating wire painted with a mixture of thermochromic pigments that change at different temperatures) on the front face and then integrating conductive pads on the back or under layer of the fabric. We used a tapestry technique to integrate a second piece of conductive yarn along a segment of the warp above the touchpad such that when it is pressed it completes the circuit. The double weaving structure makes the connective “guts” invisible from the front. Thus, the textile does not invite you to touch and poke it (how would you know where to touch), it simply captures a “picture” of the different press regions.

Laura Devendorf wove the fabric on an Schacht 8-shaft Baby Wolf loom, warped at 20 ends per inch

For those interested in recreating something like this, here are some information to work from. This draft shows the general structure of each touch pad, which leverages the structure of double weave to hide the conductive pads on the back of the fabric and highlight the thermochromic on the top.

weaving draft showing structure of the fabric (but not tie ups). Click to see larger file
color key – each color represents a continuous piece of the type listed next to the color icon

Loom Setup
This was woven on a Schacht 8-Shaft Baby Wolf, warped with 10/2 pearl cotton in natural color at 20 ends per inch.

Yarn Selection
I hand dyed sections of the warp to achieve something ikat-like, largely taking inspiration from documentation of a warp painting class at my local weaving shop. The colors turned out so much darker and astro-pop looking than I was hoping but, luckily, when I mixed them into the warp, they mellowed out. The idea was to make the overall feel of the fabric less square and more painterly. I chose the colors to complement the transition states of the thermochromic yarns.

We used the conductive yarn we had on hand for the sensing (and sadly, we don’t know where it came from) but it appears to be entirely made of stainless steel fibers which have been spun. The fibers like to “grab” the other yarns and they feel really nasty in the hand. I did one experiment adding them along the warp, which was a disaster as they grabbed the other yarns creating massive knots during warping and then broke frequently. On the plus side, they are easy to repair by just trying them back together.

We experimented with several methods for creating the color changing yarns. Playing around with combinations that could couple the painted fiber with a heating element. In some cases, we hand-spun the fibers together (shown in the figure below) with a conductive wire (38, 40AWG magnet wire) but that was finicky and prone to breaking.

We had the best results hand painting a cotton covered copper yarn we sourced from The thinner the wire, the better the heating. What seems to make this yarn work so well is that the copper inside the cotton covering is uninsulated, so lots of heat travels between the wire and cotton. It also makes for much more pleasant soldering and attaching.

Thermochromic Paint Mixture
We created our final paint mixture by mixing:

3 parts blue thermochromic that activates at 28°C;
1 part red thermochromic that activates at 43°C;
4 parts liquitex clear acryllic gel medium.

We cut lengths of our wire/yarn and then painted by hand. This makes it a little gloopy and uneven, but it works. The transition from purple to red to blue was the most notable in our experiments. The image below shows two different transitions between purple, blue or red, and white. You can see how the pink/red transition is much more visible.

Weaving Structure 

The general structure of the weave puts the thermochromic sections on the top of the fabric while hiding the conductive yarns behind the fabric. The design follows the schematic outlined below:

The general rule of thumb was that one needed two rows of cotton yarn in between a conductive yarn to prevent unintended crossing. Here is a structural view of the yarn paths and draft of the top layer yarn for a SINGLE section of color change:

You can see the black yarn is the cotton, the purple is the thermochromic, and the teal is a second segment of cotton. The thermochromic leads are routed through the “middle” or between the two layers until their are integrated into the visible surface of the yarn.

Here is a draft and yarn paths for the BOTTOM layer that has the press pads. The green is our conductive yarn, the grey is the cotton yarn, and the red is a funny little yarn we call a “warp insertion.” basically, it is a yarn we route through the bottom of the weave to reach the center point of the press pad and then manually pull through the warp using a crochet hook. This adds the yarn to the surface while making it very hidden.

diagram of method of warp insertion, thank you clement zheng for the graphic!

We are in the process of making a PCB that can interface with the entire textile, in the meantime, we have added PCBs to control a single spot or 3 spots on our lab GitHub page. Thank you Chad Di Lauro. More files should be there as we continue development. Right now, they are powered by the wall, using 12V 5A power supplies and are controlled via Arduino Pro Mini.

Tie Ups
Here is a tie up for the pattern. The frame threading is correct but I did lots of improvising on the treadles, so that is not correct for the drafts you see above. I also had to do lots of manipulations by hand.I made this in Excel, which can BARELY handle all of the information. Again, click to make larger.

Next Steps
We have created our own smart textiles design software, AdaCAD, that helps immensely with drafting smart textiles. We expect to release this on GitHub once we have it fully integrated and talking to our TC2 loom.  It will be released with a GPL license and we would like the community to get involved in development.

With the TC2 loom, we hope to recreate this general structure to create a wearable garment with custom placed sensing and actuating regions. Stay turned.

This work was supported by lots of folks including: Chad Di Lauro, Clement Zheng, Shanel Wu, Mikhaila Friske, Gaspard Bos, and the Schacht Spindle Company.