Cloudy and Tangled Thoughts

Cloudy and Tangled Thoughts

By Olivia Prior, Amreen Ashraf, and Nick Alexander

“Cloudy and Tangled Thoughts is an interactive piece which uses conductive fabric to explore the movement of light and space. Participants are invited to sit down and explore. Relax on a comfortable blanket and watch the clouds drift by. An array of irregular objects catch and refract the light, gently moving in relation to your position on the blanket, creating a sense of serenity.”

Audience enjoying final exhibit.

Audience enjoying final exhibit

img_3415

GitHub: https://github.com/alusiu/cloud-gazing

 

OVERVIEW

Cloudy and Tangled Thoughts evokes the experience of lying on a blanket, gazing at the sky, watching patterns form and dissipate in the leaves, wind, and clouds.

It consists of a blanket made from traditional and conductive textiles and a lattice of hanging geometric chimes. When participants lie or press on the blanket, lights and servo motors hidden among the chimes activate, causing them to swirl and tinkle. When more people lie on the blanket the pattern of lights and motion becomes more intricate in turn.

We succeeded in fabricating all the necessary technology, creating the code, assembling it, and proving the concept. However, the result did not live up to our vision. The team believes that the idea is strong and the tech is viable, and we will return to this project in order to develop it to a point where it meets our expectations.

 

CONCEPT

Cloudy and Tangled Thoughts started with a feeling. The team wanted to create a screenless installation that evoked a feeling of peace and wonder. We wanted to use technology to bring people together with a magical experience, using technology in a way that was unfamiliar to the average user. We envisioned the experience of lying on a blanket watching the clouds make shapes. It was important to us that we not create something simple like an on-off switch, a mechanism most people understand intrinsically, but instead create a relationship between sensors and output that generated a sense of wonder.

The prompt for the project from the Creation & Computation class was Refine & Combine. We were to return to a previous project and expand on it. While the concept we came up with was not directly related to a previous project any of us had done, we felt confident that our previous work with code, servos, lights, sensors, and fabrication put us at the same place developmentally as we would be if this had been a prior project.

 

PROCESS

The process began with a discussion of the kind of work we wanted to create, along with the kind of skills, technology, and existing projects we wanted to carry forward. When we settled on the concept above we began to brainstorm ways to realize it.

 We knew from the beginning that we wanted to work with a blanket and conductive fabric, but we debated over what form the apparatus hanging above, which needed to interact with the conductive fabric, would take. Since we had begun by imagining looking up at clouds, we researched installations and works of art utilizing cloud imagery and looked for inspiration there.

Cloud inspiration

Cloud inspiration

We decided that a series of geometric shape would complement the organic flowing nature of the blanket below. We envisioned multi coloured plexiglass, laser-cut into geometric shapes, hanging like wind chimes, diffusing light from above as they drifted and tinkled.

We submitted a proposal and consulted with our professors, Kate Hartman and Nick Puckett, for their opinions on how best to proceed. Kate provisioned us with 16sqft of conductive fabric, velostat, and a sewing machine for us to experiment with. Nick suggested that, rather than use the heavy and difficult to cut plexiglass, that we look into vellum as our cloud material, as it was light and would keep its shape after being folded.

Experimentation with vellum gave us a lot of data. We liked the way light moved through it and its versatility, and after trying several forms we settled on a triangular prism shape for our cloud-chime objects. However, we did not like the look of the vellum and wanted to find something more uniform and robust. We settled on a thin plastic and consulted with John Diessel in the Plastics Lab. John suggested that, since we planned to manufacture many identical objects from lightweight plastic, vacuum form was the best process for us. We built a form out of wood with the help of Reza Safaei in the Maker Lab and returned to the Plastics Lab to begin making what we affectionately came to refer to as “the boys”, all of which is discussed in detail below.

 

Hanging apparatus

Our conductive quilt was designed to control and move elements using servos. At first we went for simple geometrical shapes and simple constructions. 

cloud shape ideation

We had started out with wanting to control 12 unique shapes as we had built 12 sensors on to our quilt. We became very focused on the quilt, to the point that we had used up much of our allotted time before turning our attention to the servos and hanging apparatus.. Nick Puckett our professor has suggested using vellum, a material which which is easy to control due to its lightweight. The problem we encountered with vellum was that whenever we folded it to take the shape we wanted it to, it would become brittle and break at the folds. We were also slightly getting frustrated with how to to control each shape with the servos and how to mount the servos on the ceiling. We considered but decided against laser cut a mount which would hold the servos together, as we felt it was too late in the process to begin something we were unfamiliar with. At this point we as a team were getting unsure of using the servos to control the shapes. Our team mate Olivia suggested buying some fans and a relay, so that the quilt would start a fan based on where the participants sat. The fan would then start to blow these shapes. We did a rapid prototype using the vellum to construct simple modular shapes and hung them up to see the effects. We all agreed that the effects that the simple shapes created with the lights would look great.  

Prototype with vellum and lighting

We decided we liked the shapes and the effect it had. We were still unsure about using the fans when we looked further into buying a relay which was expensive and we were not very sure if we had to write new code for the fans. In addition, the relays came with a large constraint: only one attached object per relay could be powered at a time. This, combined with the cost of the relays, led us to shelve the idea of using fans.

We made a trip to the plastics lab at 100 McCaul to consult them on our simple modular shapes. John Diessel suggested we use light weight acrylics and the vacuum molding machine. They suggested we fabricate a molding shape that could be used with the machine. The largest size the machine could handle was 12×12 inches.

 We went back to 205 and visited Reza at the makers lab on the 7th floor. He understood exactly what we were looking for and helped us construct a shape using our vellum prototype.

We took the shape and went back to the plastics lab. We were instructed on how to use the machine to vacuum mold our shapes. Each sheet gave us 8 shapes, which meant we could produce a lot of shapes fast. We bought ten 12×12 sheets, 5 in translucent colour and 5 in white.

img_3256

Vacuum form in action

img_3290

It took roughly about 5 hours to construct and cut the shapes. Due to the makers lab and the plastics lab both being closed at night, which meant we had to cut the shapes by hand using scissors. This took a long time and was physically taxing on the team. 

img_3302

After we had completed molding our plastic shapes, we as a team were still unsure of how to have these shapes hang on the ceiling of the experimental media lab grid. We decided to use aircraft cable to hang the apparatus and even clip the shapes to hang using aircraft cable. We made a trip to Canadian Tire in the morning to buy crimpers for the aircraft cables. Unable to find the right grid to hang up, we made an exploratory supply run and came across a barbecue grill. Excited by the image of three circular mounts hanging in a staggered manner, we decided to buy three barbecue grills.

img_4023

At this point when we bought the barbecue grills, we only had 80 hanging shapes, which cut in half gave us about 160. We soon realized that unfortunately that wouldn’t be enough for three grills, which meant we had to make a quick run again to the plastics lab.

This is where we should have as a team scaled down rather than scale up. Building the hanging apparatus consumed a lot of time and energy. We know now in retrospect that it would have been better to go with large simple shapes rather than having so many small shapes. The small shapes made sense in the moment and looked good when hung, however they took a long time to construct.

 

Blanket Controller

Meanwhile we had also been fabricating the blanket. We used documentation from “Intro to Textile Game Controllers Workshop” run by Kate Hartman to fabricate analog sensors from the conductive fabric she gave us.

We built several small sensors to test, including one we sewed into a “sandwich” with regular fabric above and below in order to approximate the effect of the sensor when sewed into the blanket.

img_20181130_183524

The test sensors worked well, and we felt we were ready to scale up and begin fabricating full-size sensors. We laid out a large sheet of paper in order to mark and measure out the approximate size of the blanket.

img_20181130_170444

We decided a size of 4 feet by 4 feet was ideal, as it was large enough for two to lie comfortably while not being too large to manage. We debated for some time on the best way to layout and orient the sensors, with pitches ranging from as few as four sensors arranged in quadrants to dozens arranged in small triangles.

Blanket and sensor ideas 1

sensor ideation

We settled on the final version, pictured below. It allowed us to have either ends’ point of contact be on the edge of the blanket, meaning we would not need to run wiring through the blanket proper. It was, we felt, a manageable number of sensors, but enough to give us a lot of options for interactions in the final code. We also felt it was aesthetically pleasing, and thus an excellent blend of form and function. We selected classic “picnic” fabric in red, blue, yellow, and white gingham to give the device the affordance of a homemade picnic blanket.

img_20181201_182313

We plotted the sensor placement at 3-inch intervals, allowing 3 inches of velostat width per sensor, with conductive fabric cut slightly thinner than the velostat. We ironed the conductive fabric to strips of red-checked cloth, attached the velostat with dabs of hot glue, and folded the two sides together. They were kept in place with a few more dabs of hot glue until they could be sewn together permanently. We took pains to avoid puncturing the conductive fabric, sewing along the outside of the velostat. We left the ends of the conductive fabric trailing out of pockets at either end of the sensor to allow for easy connection later.

Building Full scale sensor: Measuring out pattern for cutting fabric and velostat

Building full scale sensor:
Measuring out pattern for cutting fabric and Velostat.

Building full scale sensor: Measuring velostat for cutting.

Building full scale sensor:
Measuring Velostat for cutting.

Building full scale sensor: pattern cutting Velostat.

Building full scale sensor:
Pattern tracing onto Velostat.

Building full scale sensor: Velostat laid out onto our big scale model

Building full scale sensor:
Velostat laid out onto our 4×4 ft model

Below: the process for making the sensors.

img_20181206_133808

After attaching two 3inch-wide lengths of velostat, block out a length of cloth slightly wider.

img_20181206_134108

Cut out the cloth.

img_20181206_134444

Cut out a second length, the same size as the first.

img_20181206_134952

Place and iron lengths of iron-on adhesive.

img_20181206_135230

Iron the conductive fabric to the iron-on adhesive.

img_20181206_135418

Lay the velostat over the conductive fabric.

img_20181206_135552

Use small dabs of hot glue to keep the velostat secure on both sides.

Not pictured: sew both halves together

As we completed each sensor, we tested it to ensure it was viable. When all the sensors were sewed and tested we cut a swatch of blue checked cloth at 4.5×4.5 feet to be the base of the blanket. We measured out and placed our sensors where we wanted them to be, then pinned them in place.

We conceived of and experimented with a power bus of conductive fabric along two sides of the blanket, to reduce the amount of wiring we would have to attach. We liked this idea as it made use of the blanket’s form to inform the function of the installation. However, we discovered that this layout diminished the effective voltage too much to get effective sensor readings, and we shelved the idea of the power bus. In retrospect, this should have been a warning sign to us that the power we were supplying was insufficient for our purposes.

One by one we sewed a hem in between the sensors. This fixed them in place on the blanket base, covered up the ratty ends of the sensors, and had the added benefit of making the blanket look softer and inviting to sit down on. We intended to tuck the extra fabric at the outside of the blanket over making a hem and channel for wiring while keeping the blanket looking nice and minimizing obvious electrical attachment points.

Fabric getting ready to be attached to the sensor

Fabric getting ready to be attached to the sensor

Ironing fabric adhesive

Ironing fabric adhesive

laying out conductive fabric onto adhesive

laying out conductive fabric onto adhesive

Ironing sensor onto fabric

Ironing sensor onto fabric

stitching fabric and sensor

stitching fabric and sensor

Unfortunately, our trusty sewing machine hit a snag late in production (the housing for the lower bobbin was pushed out of alignment, jamming the machine). While apparently not an uncommon problem, online diagnostics recommended taking the machine in for service rather than attempting to fix as a layperson. Without enough time to get the machine fixed or exchanged before the deadline, this was as far as we would get with our blanket. Luckily all the sensors were secured by this time and subsequent stitching would have been aesthetic.

img_20181207_175015

Code

Our main concern when deciding was how to code the blanket to create an interesting relationship between the laid out sensors and the servo motors above. We were curious with how the user would explore the interaction between the two separate components. We contemplated having a one to one relationship (i.e. one servo motor to every sensor). We as well considered having a rippling effect amongst the servo motors – when one servo would be activated a chain of the surrounding servo motors would also move.

As well, something that was overall important to us was that the clouds above reflected the participant’s position beneath the hanging apparatus. We thought this was interesting because it became a piece about the reflection of interaction.

The design of our quilt provided us with the given aesthetic of “quadrants”. We decided that we could determine the user’s position based off of the sum of the values from each quadrant. From there we mapped out all of our inputs and outputs that needed to have a relationship.

img_3727

Before we scaled up, we wanted to test the textile analog sensors acting as an input to control the output for a servo motor and led light strips. We determined what the threshold of both sensors was when some pressure was placed upon them and then used that data to determine when the motors and led lights should be activated. This was a great initial proof of concept, and we decided to proceed forward with this base code.

Our next step was to think about how to create a more interesting connection between the user activating the sensors to start the motors and led lights. We did not want the quilt to simply become a switch. As a solution, we created cases for each quadrant. Each quadrant would take the sum of the sensor input. The sum of the sensors would indicate the likelihood of how many sensors were being activated in each quadrant. The cases were as follows:

 

Maximum: most likely all of the sensors are being activated

  • Trigger all of the associated servos
  • Trigger all of the associated led lights with full brightness

 

Medium: most likely two of the sensors are being activated with great pressure

  • Trigger two (or one) of the associated servos
  • Trigger two (or one) of the associated led lights with full brightness

 

Minimum: most likely the sensors are being lightly activated

  • Randomly choose one servo to go on and off each time this case is triggered
  • Choose the corresponding led light to go on and off

 

Resting: All of the servos and led lights are off

 

Setting up for the critique

As we set up for our critique, it became apparent we had scaled up too much to implement our code. When we were assembling, we had decided to scale our inputs down to three sensors in our quadrant. Kate gave us the suggestion that rather than isolating the interaction to one quadrant to divide all of the sensors into three “super sensors”. Our quilt pattern naturally allowed us to have three rings of sensors; one on the outside, one in the middle, and one in the inside. We connected our quilt according to this diagram:

 

img_9317

Another thing that became apparent was that the hanging apparatus, due to its circular shape was hard to mount and hang in a balanced manner. We had run out of aircraft cable – which had actually proved to be extremely difficult to work with – so we decided to use twine to get the shape mounted. Another difficulty we ran into was the wiring of the apparatus to the floor. We were not prepared with the right set of wires which would be long enough to reach our breadboard. We attempted to use long individual wires, but that was impractical. Kate and Nick lent us long modular wiring, which significantly helped us with the hanging process. Also we learned that Kate is master at knots. Her wizardry helped us hang the apparatus quickly and safely.

 

Components

cloudgazing2_bb

Diagram only shows what was connected for critique

  • 1 x Arduino Mega
  • 3 x textile sensors
  • 3 x 50 ohms resistors
  • 3 x LED light strips [6 pixels each]
  • 3 x Micro Servo Motors

 

OUTCOME

We had lofty expectations for this project which the completed version did not meet. No aspect of the build was lax – we felt, in the end, that there had not been enough time in the two weeks we were allotted for the team to build, test, and iterate on the design enough to reach the state of completion we had envisioned.

In the end, we did have an interactive experience where the quilt activated led lights and gentle servos above. We also incorporated a projection behind to elevate the sense of being out in nature.

 

REFLECTIONS

This project taught us many things about working with unfamiliar materials and pursuing lofty goals in a short time frame. Some core reflections we have taken away are below.

We encountered many challenges we did not foresee or appreciate during the planning phase. These included:

  • The amount of time required to fabricate objects of the size and complexity we envisioned
  • The difficulty and time required in learning to effectively use new tools
  • Power management with sensors that we had created from scratch
  • Effectively scaling from a working prototype to a full-sized installation
  • Accounting for the “unknown unknowns” that crop up in projects

Were we to take on a similar project in future, we would:

  • Focus on one core interaction – for example, we would focus on only the blanket or the hanging apparatus
  • Do careful math when fabricating rather than making estimates
  • Start with fewer/smaller materials and scale up
  • Make purchases of materials in small amounts to prototype with

In terms of the use of textiles, we came across a couple of discoveries:

  • Our sensors only worked consistently when the ground and the positive were clipped to the opposite ends of the fabric. We experimented with having the two ends of the circuit clipped close together, which – while somewhat effective – was unreliable for our purpose.
  • When all of the twelve sensors were divided and clipped together to make three “super” sensors, we had to lower the resistors significantly to get any viable reading to use with our code.
  • Physically small sensors gave more reliable readings than large sensors at the same voltage.
  • It is possible to use conductive fabric as a “power bus” to power multiple sensors – though at our scale, this diminished the voltage to an amount where they were not usable for our purpose.

Next steps to take when we return to this project include:

  • Test the sensors with higher power and/or using multiple power sources
  • Test multiple variations of circuitry running through the blanket
  • Design, from scratch, an apparatus for hanging the clouds, with the same focus we had as we designed the blanket
  • Explore wireless communication with the hanging apparatus
  • Reconsider the form of the “above” apparatus
    • For example, explore projection of a generative image rather than a physical apparatus

 

Resources:

Kate Hartman & Yiyi Shao. Intro to Textile Game Controllers. Workshop held at Dames Making Games at Toronto Media Arts Centre on November 14, 2018

A special thank you to Nick Puckett whose advice on fabrication was invaluable, and who went out of his way to help the project get set up in time for its show.

A special thank you Kate Hartman for her donation of material and tools, for going out of her way to help the project get set up in time for its show, and whose infectious enthusiasm kept us going.