Felted Musical Landscape is a textural interactive felted 3D textile project. The tips of individual bumps are capacitive touch sensors that responds to touch and plays piano notes via Processing. This interactive piece is soft to touch and fun to interact with, as one wouldn’t expect these bumps to make sounds when squeezed or touched!

Inspiration

During spring break I traveled to Reykjavik, Iceland and was inspired by the low planes of bumpy moss landscape. Right when I came back from Iceland we worked on using the knitting machine at the lab, and I was excited about knitting 3D bumpy structures. I also had some Icelandic Lopi Wool that I brought back from Iceland. Perfect. This is the project I’m going to work on. Large size installation scale (initially  planning on making a 30″ x 36″ large wall textile), interactive, fun!

Below are examples of felted topology projects done by Xandy Peters. I was also inspired by the large scale installation by Pia Mannikko.

I wanted to make the textile piece devoid of any wires to the computer, so I tested using the Arduino Fio with Xbee. Fio is much easier to work with if you’re doing a point to point simple sensor network. There is a easy configuration tool, and the Fio board has the Xbee shield and power integrated with the board. However I managed to break the board when I was soldering multiple resistors (without headers…), so I decided to work with Lilypad Arduino and Lilypad Xbee since it’ll be easier to sew on to the textile later on.

I used Image Based Circle Packing Script in Rhino to generate a packing pattern that I’d use as a base for the bump knit pattern. I layered a grid on top of the packing pattern and mapped out a preliminary circuit path from the Lilypad Arduino.

To follow this pattern, knit all grid boxes, and start the hold-knit-hold pattern at darker lines (circle diameter lines indicated with stitch numbers) to make bumps. Add conductive thread to the wool strand at the dotted tip areas of each bumps to create conductive surface. (hold-knit-hold: knit one row until last stitch, hold last stitch. Turn and purl until last stitch, hold last stitch. Repeat this pattern until two stitches are left. Then work your way backwards).

I was planning to use the mid-gauge knitting machine at the lab, but unfortunately the machine broke. I had to choose between:
-hand knitting
-use fine-gauge knitting machine
-mill bump surface with foam as formwork and make wet-felted landscape.
I decided to go with #1: hand knit.

Test #1 was using wool yarn (from the lab, and it seemed like 100% wool), but I translated my pattern inaccurately (didn’t knit enough rows to allow enough expandable surface for knitted bumps. I kept knitting the hold-pattern after another). Also the yarn didn’t felt well. Change of yarns.

Test #2

I used 200 grams of Iceland Lopi Wool from Istex Mill , and this time read the pattern correctly. Knit knit knit.

After many hours of relaxed but shoulder-intensive knitting, I felted the piece using washing machine (2 cycles). Below is a picture of the felted piece when it’s still fuzzy. While it’s damp (never put it in the dryer), make sure to shape the textile to your desired form. Let it air dry.

Here’s a little intro on capacitive sensor: http://www.arduino.cc/playground/Main/CapSense

To maximize the number of sensor pins on the arduino, I designated three send pins, and used those pins to connect 10 mega ohms resistors to each of the sensor pins. I had to debug the the arduino pin connections by testing each pin with simple capsense code. Most of the times it was my amateur soldering technique that caused the problem. After everything is connected, I sewed the Lilypad on the textile and connected each sensor pins to the conductive tip of the bumps.

This time, I made sure to lay out the entire circuitry before starting any sewing. The bottom pin is the Lilypad and top one is the Xbee Shield. LEDs were difficult to see through the thick felt, so I took them out later. There are a total of 12 notes – a full octave with flat/sharp notes. Now it’s time to put everything together.  Piano frequency can be found here, and the Processing Piano code sketch can be found in our class notes here.

Next step is connecting with the Xbee modules. I want this interactive piece to be a stand-alone piece, and with the xBee it’s possible to send sensor data over radio/wireless network. I used simple Series 1 Xbees.
You need: 2 xBees, 1 xBee explorer (for your computer), lilypad xbee shield, power connector for xBee/lilypad, usb cable

I used X-CTU program to configure the xBees, and followed the steps described in this online post except few things: I didn’t change the baud rate to 19200, and also didn’t set the D2 pin as data pin in the sensor-xBee module. Plug in the xBee, and if two xBees are communicating, green light turns on. Yes! it’s communicating. I opened up the Arduino serial port, I’m reading 13 sensor readings. Great. However, when I went to the Processing sketch, nothing would happen. WHY?? Do I need to import the xBee library? I tried multiple things but all didn’t seem to work. More research is needed (frustration! why isn’t this working?)

So finally, here’s a demo video of the felted musical landscape (wired for now)

Capacitive Sensor:

Wireless Sensing: Arduino FIO + XBee

Tests:

While I was searching for pattern codes and inspiration images, I came across this line-art image (a profile “photo” of this person’s website :http://nwerneck.sdf.org/). The dashed hatch line thickens and thins in this image, and I thought it could be a perfect embroidery patter with a mixture of lines and fills. It also seems to be computer generated, and I could sketch out a possible line of Processing in my head. Load image -> filter image (black and white) -> analyze pixels -> draw lines -> make lines thicker where black pixels are -> print to pdf . Voila! Now with my limited Processing skills….

I started by digging into Processing book and website for sample codes as a starting point. I grabbed the “Writing to another PImage object’s pixels” code from here, and made slight modifications. Although I knew what I wanted to do next (take pixel data and thicken series of line at specific pixel points, for example), I wasn’t able to get exactly what I had envisioned. Instead, I decided to make it graphically similar: I drew thick white lines on top of the image via Processing. Then I was able to export the image to PDF and make some Illustrator adjustments such as merging, trimming lines before I took it to the embroidery machine.

Below is the code I put together to covert an image (You can grab the code and load any image).

______________________

download proposal here

Knitting with knitting machine! I was excited to start this assignment, especially after seeing a sample of a 3d knitted fabric. The first part of the assignment was to make a swatch following the pattern below:

Cast on 20 needles. Knit 10 rows. Carriage should be on right side. Make sure carriage is in the proper mode (Russel levers should be in position I). Bring needle closest to carriage into hold position. Knit one row (carriage should be on left side). Bring needle closest to carriage into hold position. Knit one row (carriage should be on right side). Continue bringing one needle into hold position on the carriage side until you have 2 needles in knit position in the center of your pattern. Carriage should be on right side.

Begin taking needles out of hold position. Push the needle closest to the knitting on the opposite side of the carriage (the left side) into B (knit) position. Knit one row (carriage should be on right side). Push the needle closest to the knitting on the opposite side of the carriage (the right side) into B (knit) position. Knit one row (carriage should be on left side). Continue bringing one needle out of hold position opposite the carriage side until all needles are back in knit position. Knit 10 rows. Cast off.

After completing the first pattern, I continued on with the same pattern but added additional knit rows in between every hold position. It created a taller bump. Taking this test to the next step is the below attempt of creating a knitted landscape.

After coming back from a week long spring break research trip (for another class) to Iceland, I was inspired by the vast moon-like moss covered lava rock landscape of Iceland. For the second portion of the assignment, I wanted to create a knitted landscape (a small swatch of 6×6 for now) of small bumps in different shades of greens and yellows (imagine a mossy field in spring with small yellow flowers blooming – if there are flowers at all!)

I tested knitting with different yarn – brown cotton yarn, canary yellow linen yarn (too thin to knit with, and it broke off easily), moss green wool yarn (too thick for the knitting machine), and pine green thinner gauge wool yarn. The challenge was to use all of these different material/gauge yarn on a single machine but keeping the same knit gauge to create different densities of bumps.

A mess! After struggling for few hours with the machine, I was able to knit few bumps with only one type of yarn. Problems were: yarn breaking off, yarn getting caught in the machine, knits slipping off the needles, loosing stitches, to name a few. More tests to follow. Below shows a successful 3d knit piece. (I used a larger gauge machine)

And this lovely image shows felted (with hot water – washing machine) knit piece.

___________________________________________________________________________________________________________

For this assignment, I used Grasshopper (plug-in for Rhino3D) circle packing script to generate different patterns for laser-cut lace.

Instead of packing circles completely on a given surface, I decided to use multiple attractor points to define starting points of circle packing. Below image shows a screenshot of Grasshopper interface and also a 3D view of the pattern generated in Rhino.

The picture above shows two layers of laser cut surface with spacers at larger circle areas to create undulating surface.

Then the next iteration was to use the attractor points as boundary lines. These boundary lines can be used to stitch different sections together if one were to use this pattern to make wearable piece. You can see the location of attractor points in the laser cut piece (bottom left image). I couldn’t use a single line as an attractor, so I had to use multiple points on a line. After the pattern was generated, I manually deleted some of the smaller circles next to attractor points to get a clean line of “seams”.

The bottom picture is the final laser-cut piece that shows clearly defined attractor-point lines, where circles get more compact and smaller near the seam lines and larger towards the center of the bounded zone. It resembles a starfish!

Finally, I used denim fabric to laser cut the script-generated pattern. At first I scaled down the pattern to a 8″ diameter size to test the tolerance. Some of the larger cut areas didn’t hold, and started to fall apart. In the next cut, I scaled up the pattern three times. Now there is about 1/8″ fabric between each circular cuts.

If I were to use this in a wearable project, I would first create a quilt-like pattern and apply circle packing script on each section. Then the “seam” or the “rib” areas can be used to fold, stitch, cut the fabric into a wearable piece.

For the nonwoven textile project, I tested felting wool roving with aluminum wool as conductive material. In this quick test, I soldered a small LED to a crimp-beaded aluminum wool. It seems to light up and work! I liked the fact that everything can be felted into one material/surface; no need for stitching conductive thread, etc.

I decided to wet  felt a small plate and needle felt aluminum wool into two strands where I can attach series of LEDs. Pictures below show the felting process, where I layered two different colors of wool roving and used a bowl as a mold to felt the plate shape. 

After the bowl is felted, I needle felted two rows of aluminum wool. Since the aluminum wool doesn’t solder well, I added crimp beads where I wanted to attach LEDs to. Making sure that the LEDs are all aligned in the right direction, I soldered the LEDs to the crimp beads. And.. it should light up with the lilypad battery! Hmmm…. although individual LEDs light up when lilypad is connected individually, the row of LEDs do not light up all together. I’m guessing the aluminum wool is not conductive enough, or the wool fibers are shorting the circuit. Time to debug!

After multiple failed attempts of making other nitinol-based kinetic textile project, I tried making a small shell-shaped wearable brooch using felt and nitinol (shape changing alloy). Below are pictures of the first project I tried before moving on to the shell project.

attempt 1 -pinwheel jacket

The idea was to use the nitinol to pull the fabric from the center of the pattern to expose the color beneath the green fabric. I tested multiple ways of using nitinol. First was to pull in a radiating pattern from the center, as shown in the picture on the left. The nitinol didn’t provide enough tug pull, and also the fabric of the jacket itself (jersey) was too stretchy, and it seemed that the pull action from nitinol was all absorbed by the fabric.

In the next test version, I added stiffness to the jersey fabric at pull-locations by using iron-on paper/cotton fabric. I tried pulling in a triangulated pattern, and also in a radial pattern. Both didn’t seem to work well. The raw edges of the jersey fabric started to curl as well, and it was hard to read the nitinolpull action. I decided to abandon this fabric and use a stiffer fabric for the next test.

.

.

.

attempt 2 – talking fabric

In the next test, I tried to use nitinol to move the slit as if the fabric was talking (moving its “lips”).  I sewed down the nitiol on the top lip, and zigzaged the nitinol all the way across the slit. That did not seem to pull much. So on the bottom lip, I used nitinol only on the mid portion of the length and did not sewed down the nitinol. It seemed to work better. But I wanted to create more dramatic effect, so I decided to make another one.

.

.

attempt 3 -seashell

In this version, I made multiple cuts and sewed two separated pieces of nitinol. It’s a bit difficult to see in the picture, but there is one nitinol at the top, and another one at the bottom (both are zigzaging). Each nitinol has its own mosfets, and given the thickness of nitinol, I needed to provide 380mA~440mA to activate. Measurement showed 6.9 ohms for the top one, and 7.1 ohms for the bottom one. I added two 1 ohm resisters (mini rectangular shaped resistors) for each of the nitinols to get 9 ohms of resistance on each nitinol.

I made a simple on-off switch with copper tape and sewed it securely to the bottom flap of the shell. After connecting everything I hooked it up to arduino and ran the code. Hmmm.. it didn’t seem to work. The problem seems to be that I used conductive thread to connect to arduino instead of wire. Going back to debug/fix. Meanwhile it worked with the heatgun, but not as dramatic as I hoped it to be. Video is at the top of this post.

I love felting – all you need to create a tight fabric of felt is roving wool, hot water, and soap (and a lot of rubbing). When I saw a felted pompom pressure sensor during class, I was excited at the possibility of felting my own sensor.

But Arduino coding was quite new to me. I started testing with Arduino code, and made a temporary sensor/LED/lilypad combo on a piece of foam to understand the circuit layout. Sensor is made using layers of velostat, conductive thread, and conductive fabric. Tack down all parts on the foam with push pins, connect them with conductive thread, upload the code, and voila! LEDs started to blink! The arduino code makes the LEDs blink at a slow speed when the sensor is at resting state, and the speed picks up as you press/bend the felted sensor.

Completed small swatch of color changing textile activated by 4V battery.

Here’s a short video of the color changing in action.

________________________________________________________________________________________________________________

::process::

materials:
non-conductive fabric (cotton)
conductive stainless steel thread
3.7V lithium polymer battery
switch
thermochromatic ink

I wanted to create a small pattern of beach pebbles in different colors.
The conductive thread stitches would form the white pressure lines of the pebbles, so when the stitches heat up (with the power from 4V battery), the color of immediate adjacent area next to the conductive thread would turn clear.  First I sketched few different pattern options as well as stitching pattern, then I outlined the selected pattern on the cotton fabric. It is now ready to be sewn. I doubled the conductive thread to make sure the pattern had enough resistance.

With the completed pattern, I took it to the Lab to start applying the thermochromatic paint. The main thermochromatic paint color was blue, so I added just a little bit of other colors to the main blue color to achieve various shades. One thing to remember was not to add too much of other colors to the thermochromatic mixture, otherwise the thermochromatic mixture would get diluted. So below is the first layer of paint applied.

Using the multimeter, I measured the resistance and current within the circuit.

resistance: 11 ohms

given the equation V=IR, anticipated current was:
3.7=11I
I=336 mA

measured current was = 137 mA (much less than calculated)

When I turned on the switch, the color change was so subtle that it was almost unnoticeable. I applied another coat of paint, especially near the stitch area. This time I could definitely read the area near the stitch line turning brighter.