Kristin Barclay1, Steve Burke2, Zamyla Chan2, Thomas DiBenedetto2, Max Shayer2, Jake Sobstyl2 , Antonio Sweet2, Matthew Wagner2, Katrina Williamson2, Matt Yarri2

1Principle and corresponding author. Contact: kbarclay@college.harvard.edu; kri.barclay@gmail.com.
2Contributed equally to the execution of this project.

Abstract

It is well-demonstrated that clothing looks different under varying light conditions (Livingstone, 1988; Roberts, 2010); fashion designers have long been frustrated by how little control they have over their garments’ color. When designing clothing, this distorting environmental effect cannot be avoided or predicted. For this project, we instead chose to embrace the idea of a color changing garment: we strove to achieve a garment linked to its surroundings and, through such connections, capable of adaptive coloration. Prior fiber optic fashion has focused on adding static light to existing garment – not on designing dynamically colored clothing. Through novel application of fiber optic cables, LEDs, and microelectronics, we created three dresses that change color according to cues from the world around them, magnifying and controlling the effect that the environment has on clothing. Using a variety of sensors, we transformed the dress – a normally static and isolated entity – by making four interactive prototypes: the Heart Rate, Audio, Social, and Compass dresses. Our dresses together expand the dimensions of fashion and create a more dynamic, individualized experience for the wearer.

Introduction

We first thought of creating color changing dresses after speaking with fashion designer Rodarte (Rodarte, 2013) about unexplored aspects of fashion that intrigued them.  Sisters Kate and Laura Mulleavy – the co-founders and co-designers of Rodarte – were most excited about the opportunity to make color adaptive outfits featuring bold colors and abrupt hue transitions. Rodarte is famous as a label for pioneering new ideas: their clothing line features collections inspired by everything from Japanese horror movies to their hometown of San Diego, and they created the ballet costumes for the movie Black Swan. In addition to a strong color change, they desired technology that coupled elegantly to the outfit: they did not want bulky devices added to garments that would need to be designed around when using fabric.

Another source of inspiration for us was the natural world. Nature has many examples of bold color transitions: adaptive coloration in cephalopods displays many of the characteristics Rodarte desired to recreate in clothing. Cephalopods are marine organisms – such as the squid, octopus, and cuttlefish – which have the ability to modulate the color and texture of their skin on command (Mäthger, 2009). This ability is due, in part, to chromatophores on the surface of the cephalopod’s skin (Figure 1). Chromatophores are organs containing pigment that change in size when signaled by the brain, allowing dynamic coloration. Thus, cephalopods take in information about their surroundings and respond through adaptive coloration to their environment – exactly the effect we sought to reproduce in clothing.

Fig. 1 The chromatophores of the squid Loligo pealei. Chromatophores allow the squid to change its color and texture by changing in size when signaled by the brain. Images courtesy of Dr. Leila Deravi.
Fig. 1 The chromatophores of the squid Loligo pealei. Chromatophores allow the squid to change its color and texture by changing in size when signaled by the brain. Images courtesy of Dr. Leila Deravi.

Adaptive coloration has previously been explored for military applications: the United States military has evolved its uniforms throughout its history in order to adapt to the location of each war or conflict. Most recently, after many years of research, the U.S. Army’s Universal Camouflage Pattern was adopted in 2004 as a common solution for military troops around the world: the goal was to design a single pattern that would effectively camouflage troops in any terrain using a pixelated color pattern (Enger, 2012). After complaints by troops that the pattern was not effective in the brown landscape of Afghanistan, the Universal Camouflage Pattern was replaced with a more traditional camouflage known as MultiCam. The US Army spent more than $5 billon on the new pattern, and it is now considered a failure (Enger, 2012).

Adaptive coloration in commercial clothing is currently limited: available light-enhanced clothing provides only simple illumination (no dynamic color change) in that is either constant or responds to an environmental signal. Commercially sold T shirts include shirts that display WiFi signal by lighting up bands corresponding to signal strength (Think Geek WiFi Dectector Shirt, 2012) and $11.95 sound-sensitive T shirts featuring blinking LEDs  (HDE, 2013). In another area of wearable light, initial attempts at incorporating OLEDs (organic LEDs – flat, flexible LED panels) into clothing met with mixed reactions at the stiffness and impracticality of the garment (Heimbuch, 2009). One French company, LumiGram, creates lighting effects using fiber optic cables, but the color of their clothing is static (LumiGram Catalog, 2013). Light-enhanced and electronic clothing are a niche market in today’s fashion industry: these technologies are commonly used only for novelty items, not for serious, sophisticated fashion.

Using Rodarte’s input and the biology of cephalopods as inspiration, we created four color changing dresses that bridge the gap between current LED and electronic garments and high fashion. Our Heart Rate dress responds to its wearer’s heartbeat; the Audio dress reacts to the volume of sound in its environment; the Social dress allows two people, each wearing a Social dress, to exchange dress colors; and the Compass dress responds to the Earth’s magnetic field by transitioning through the rainbow as its wearer turns in a circle. Each dress also has a user-controlled color mode during which the wearer can manually determine the color of the dress. By creating garments change in appearance by interacting with their user and greater environment, we allow people to play a more active role in the way they express themselves through clothing, potentially redefining our conception of fashion.

Materials and Methods

We implemented color change by sewing LEDs coupled to fiber optic cables into a dress: we achieved interaction with the environment by using sensors and a microcontroller. We then created four dress designs, each based upon a different sensor.

Due to limited resources, these four designs were implemented in three physical dresses: the Compass dress, the Heart Rate/Social 1 dress, and the Audio/Social 2 dress. Heart Rate/Social 1 and Audio/Social 2 each contained a switch to change between operating modes. The dresses were designed modularly: each shares the same structure of sensor (samples environment) to microcontroller (turns samples into a color value) to LEDs (create colors using light) to bundles of fiber optic cable (disperse colored light throughout dress). This modular design allowed us to divide each dress’ design into discrete elements.

Sensors, Microcontroller, and LEDs

We selected the following sensors:

Heart Rate: a T31-coded Polar Heart Rate Transmitter chest strap and the compatible Polar Heart Rate Monitor Interface. The chest strap transmitter is worn beneath the garment with the two electrodes and skin in contact with the electrodes.

Audio: an electret microphone. The microphone detects the volume of environmental sound between 60 Hz and 15 kHz, which matches well with the range of human hearing  (20 Hz to 20 kHz).

Social: the XBee Series 1 radio transmitter placed in both dresses. The XBees serve as a wireless bridge between the two microcontrollers, allowing exchange of serial data –in our case, the current color of each dress.

Compass: an OSEPP compass sensor. We converted the reading of the magnitude and direction of Earth’s magnetic field into an angle that determines the color of the dress.

Each sensor sends environmental data to a microcontroller, which turns them into red, green, blue, and white color values to be displayed through the fiber optic cables. We elected to use a Lilypad Arduino ATmega 328 Main Board as the microprocessor for our sensor and LED control.  The Lilypad is a specialized Arduino designed for wearable electronics: it has a very flat physical profile (1mm) and holes where its output pins are, allowing it to be easily sewn into garments. Though we did not directly utilize the Lilypad’s ability to be wired with conductive thread – we instead used traditional wires by soldering them directly to the pads of the Lilypad – the Lilypad was still an excellent choice for this application due to its well-documented code base and thin and small size (E-Sewing: LilyPad Arduino, 2013).

Once the Lilypad turns the sensor’s reading into a color value, we use RGBW LEDs – compound LEDs that contain individual red, green, blue, and white LEDs – to generate the light for the dress. By varying the brightness of each individual red, green, blue, or white LED contained in the RGBW LED, we create RGBW color mixing capable of displaying any color on the visible spectrum. For example, if the red and blue LEDs are each at half brightness, a viewer will see purple light at full brightness.

Each LED is designed to operate at 350 mA of current for maximum brightness: however, the Lilypad can only supply 40mA of current per pin. To power the LEDs, we used transistors (MOSFETs) to interface between the Arduino, power supply, and LEDs. With this setup, the LEDs only receive enough electric current to turn on when the Lilypad sends an output signal based on the sensor input, allowing us to control the display of colored lights on each dress. MOSFETs also allow us to use pulse width modulation to control the brightness of each LED to create color mixing.

Our final Compass circuit, including Lilypad and RGBW LEDs, is pictured in Figure 2.

Fig. 2. (left) Final circuit for Compass dress. Circular Lilypad and MOSFETS are at the center: 4 RGBW LEDs branch off of the circuit via gray ribbon cable wire (shown without couplers). The battery is shown at top center of picture, and potentiometer to the right of the Lilypad. Sensor not pictured. (center) Compass circuit attached to exterior of Compass dress underdress to visualize its location. (right) Completed Compass dress with circuit hidden inside underdress. Circuit is accessible via the three buttons near the zipper holding down the flap covering the circuit. Potentiometer is exposed in center of back.
Fig. 2. (left) Final circuit for Compass dress. Circular Lilypad and MOSFETS are at the center: 4 RGBW LEDs branch off of the circuit via gray ribbon cable wire (shown without couplers). The battery is shown at top center of picture, and potentiometer to the right of the Lilypad. Sensor not pictured. (center) Compass circuit attached to exterior of Compass dress underdress to visualize its location. (right) Completed Compass dress with circuit hidden inside underdress. Circuit is accessible via the three buttons near the zipper holding down the flap covering the circuit. Potentiometer is exposed in center of back.

Fiber Optic Cable Bundles and Couplers

Colored light from the RGBW LEDs was spread throughout the dress using abraded fiber optic cables (Table 1).  Normally, by utilizing total internal reflection, a fiber optic cable is designed to transmit a light signal without loss along the length of the cable: however, if we abrade the side of the cable, this breaks the total internal reflection, allowing light to be emitted along the sides of the cables to illuminate the dress (Figure 3). Some of the light leaks out of the fiber optic cable instead of being fully transmitted. Using this effect, we bound 350 fiber optic cables together to form a bundle. When coupled to an RGBW LED, the bundles of abraded fiber optics disperse light radially along the sides of the cables and onto the fabric of the dress, achieving our desired color change (Figure 3).

Fig. 3 (a) Unabraded fiber optic cable at 266X magnification. SEM image. (b) A bundle of approximately 350 fiber optic cables held together. (c) Roughly abraded fiber optic cable at 466X magnification. SEM image. (d) A 1m bundle of abraded fiber optic cables illuminated by a red LED.
Fig. 3 (a) Unabraded fiber optic cable at 266X magnification. SEM image. (b) A bundle of approximately 350 fiber optic cables held together. (c) Roughly abraded fiber optic cable at 466X magnification. SEM image. (d) A 1m bundle of abraded fiber optic cables illuminated by a red LED.

Table 1. Specifications for the fiber optic cables used in the dresses.

Interior (Core) Poly(methyl methacrylate)
Refractive Index of Interior 1.492
Exterior (Cladding) Proprietary
Refractive Index of Exterior 1.402
Diameter 250 μm
Density 1.18 g/cm3
Temperature Range -55ºC to 70ºC

In order to efficiently transfer the light from a LED to a fiber optic bundle we designed and built a thermoplastic coupler (Figure 4) to hold the fiber optic cables tight to the surface of the LED. In total, we abraded 12 bundles, for a total of over 3 miles of fiber optic cable (Table 2).

Fig. 4 Fiber optic bundles and coupler. (a) Final coupler design (thermoplastic) attached to RGBW LED. LED is soldered to ribbon cable wire. (b) Prototype coupler displaying how bundle fits into coupler.
Fig. 4 Fiber optic bundles and coupler. (a) Final coupler design (thermoplastic) attached to RGBW LED. LED is soldered to ribbon cable wire. (b) Prototype coupler displaying how bundle fits into coupler.

Table 2. Length of fiber optic cable used in each dress.

Dress Audio Heartrate Compass
Bundle 1 (in) 16 40 54
Bundle 2 (in) 21 55 54
Bundle 3 (in) 33 60 60
Bundle 4 (in) 34 60 60
Total Length (ft) 3030 6270 6650

Dress Design: Visual Perception

Fiber optic bundles can be arranged on the dress in any pattern: they can be split to reach different areas, cut to different lengths, and held in place with stitches. To determine what patterns to display with the bundles, we studied how the brain perceives color and brightness. Using the concept of luminance (Tadin, 2003), we can make sections of a dress look comparatively larger by making the section brighter (Livingstone, 1988). Increasing the luminance of part of a garment by emitting light causes the perceived luminance of surrounding sections to decrease, and thus look smaller.  By emphasizing some areas, we can effectively shrink others.

Our dress designs include thicker and brighter fiber optic cables above and below the waist in order to create an hourglass effect by shrinking the perceived size of the waist as compared to the bust and hips (Figure 5). Because our dresses both emit and reflect light, we can to achieve interesting neurobiological effects that cannot be achieved with just fabric alone.

Fig. 5 Original sketches for grid-inspired dresses: Audio (left), Heart Rate (center), and Compass (right) dress designs. Each design has areas of higher luminance at the bust, hips, and lower skirt to decrease the apparent size of the waist.
Fig. 5 Original sketches for grid-inspired dresses: Audio (left), Heart Rate (center), and Compass (right) dress designs. Each design has areas of higher luminance at the bust, hips, and lower skirt to decrease the apparent size of the waist.

Fabric-Technology Integration

Once we assembled the technology for changing the color of a dress, we physically integrated our electronics, RGBW LEDs, and fiber optic bundles into a dress. We purchased the Heart Rate/Social 1 and Audio/Social 2 base fabric dresses from Rodarte and handmade the Compass dress ourselves. We incorporated both fiber optic cables and base circuitry into our dresses by first designing a white, two layered dress to secure all circuitry and fiber optic bundles. We placed all electronic components on the lower and upper back areas of the dresses because the back is an area that is most often kept straight and is held relatively stable as a person moves.

Each underdress features a fitted bodice and skirt in order to accommodate the draping behavior of fiber optic cables: fiber optic bundles drape well only when the fibers are vertical, not horizontal, constraining the flow and flexibility of typical fabric. The Heart Rate/Social 1 and Audio/Social 2 dresses use an attached white pocket to hold circuitry in place and wrapped white fabric around exposed ribbon cable wires (the connection to the LEDs) in order to conceal the electronics. Because these two base fabric dresses were purchased from Rodarte, we were forced to retrofit the dresses to work for our purpose.

In contrast, we designed and built the Compass dress as a three layer dress: the fitted underdress is itself made of two layers, the interior of which secures the circuitry (handsewn in place) and hides the wires to the LEDs. This interior layer is accessible via the back of the exterior layer of the underdress via a flap held in place with buttons (Figure 2). Fiber optic bundles emerge from the interior underdress onto the exterior overdress via sewn buttonholes. On the Compass dress, the circuitry adds only ½” to the profile of the dress, and our model did not notice the circuitry after she put on the dress.

Once the electronics were secured, we sewed fiber optic cables in place using both hand and machine sewing techniques. Hand sewing allows for thicker cable-style structures as seen on some parts of the Compass dress, while machine sewing produces thick, flat stripes as seen on all three dresses.

Each dress also has a translucent overdress worn over the underdress in order to protect the fiber optic bundles and further disperse their emitted light.

Results

Using the techniques outlined in Materials and Methods we successfully built four dress designs (Heart Rate, Audio, Social, and Compass) using three physical dresses (Heart Rate/Social 1, Audio/Social 2, and Compass). Our dresses were showcased at our final presentation on May 7, 2013 at Agassiz Theatre, Harvard University. Our dresses were modeled by three students (Figure 6) for the presentation.

Fig. 6 Top: Heart Rate/Social 1 (left), Audio/Social 2 (center), and Compass (right, dresses at our final presentation on May 7 2013 at Agassiz Theatre, Harvard University. Each dress is in user control mode, displaying a single color. Bottom: Compass (left, illuminated), Heart Rate/Social 1 (center, off), and Audio/Social 2 (right, off), as modeled at our final presentation on May 7 2013 at Agassiz Theatre. The fabric base of the Compass dress was sewn by us, and the Heart Rate/Social 1 and Audio/Social 2 base dresses were purchased from Rodarte. The light from the Compass dress is partially drowned out by the stage lights of Agassiz.
Fig. 6 Top: Heart Rate/Social 1 (left), Audio/Social 2 (center), and Compass (right, dresses at our final presentation on May 7 2013 at Agassiz Theatre, Harvard University. Each dress is in user control mode, displaying a single color. Bottom: Compass (left, illuminated), Heart Rate/Social 1 (center, off), and Audio/Social 2 (right, off), as modeled at our final presentation on May 7 2013 at Agassiz Theatre. The fabric base of the Compass dress was sewn by us, and the Heart Rate/Social 1 and Audio/Social 2 base dresses were purchased from Rodarte. The light from the Compass dress is partially drowned out by the stage lights of Agassiz.

User Control of Color

A color changing dress isn’t the perfect outfit for all occasions: thus, we designed each dress’ color such that it can directly controlled by its user in addition to reacting to its environment.  In order for the user to be able to turn the dress on and off or set the color of the dress to a constant color instead of the color being controlled by a sensor, we added a switch and a potentiometer.

The potentiometer provides analog input to the Lilypad to determine the mode that the dress is in: a potentiometer is a variable resistor, and thus the voltage to the Lilypad’s input pin will change as the potentiometer’s dial is spun. Using the potentiometer as a dial, users can select an RGB spectrum color or put the dress into sensing mode.

Dress 1: Heart Rate

The Heart Rate dress interacts intimately with its wearer by sensing the wearer’s heart beat: the horizontal stripes of the dress blink at a rate proportional to the wearer’s pulse. By responding to the pulse of the wearer, the Heart Rate dress displays the wearer’s mood: if the wearer is calm and relaxed, the dress blinks at a slower rate, while it blinks more rapidly if the wearer is excited.

At lower heart rates, the stripes start off blue. As the heart rate increases, the color changes to green and then to shades of red (Table 3). The vertical stripes of the dress remain on constantly. The luminent horizontal stripes at the top and bottom of the dress draw attention away from the user’s waist. A sketch for the dress is shown in Figure 5, and the completed dress is shown in Figure 7.

Table 3. Heart rate ranges and the corresponding color of the LED lighting the horizontal bundles. The voltages outputted to the red, blue, and green LEDs are given in parentheses.

Measured Heart Rate (BPM) Horizontal Bundle Color (red, blue, green)
< 70 Light Blue (0,255,128)
70-75 Green (0,100,255)
75-80 Purple (44,225,0)
80-85 Fuchsia (255,127,0)
>85 Fuchsia-Red (255,110,0)
Fig. 7 Heart Rate dress.  The completed Heart Rate dress displaying fuchsia (left) and fuchsia-red (right) horizontal stripes for different pulse rates.
Fig. 7 Heart Rate dress. The completed Heart Rate dress displaying fuchsia (left) and fuchsia-red (right) horizontal stripes for different pulse rates.

To visually emphasize the heart rate of the wearer, the LEDs lighting the horizontal bundles pulse and fade with the heart’s beats. The time it takes for the LED to fade from fully on to off is inversely proportional to the heart rate, so that the lights pulse more quickly as the heart rate increases.

Dress 2: Audio

The Audio dress interacts with the acoustic environment around its wearer. An audio sensor senses the volume of the dress’ surroundings and causes the horizontal stripes of the dress to light up accordingly. As the volume in the room increases, more and more stripes light up. The design for the Audio dress draws its inspiration from a sound equalizer. The stripes at the top and bottom of the dress are longer (and thus more luminent) than those at the waist, drawing attention away from the user’s waist. A sketch for the bundle locations of the dress is shown below in Figure 5, and the completed dress is shown in Figure 8.

Fig. 8 The Audio dress displaying a random color in state 1, after V1 is passed (left); displaying two random colors in state 2, after V2 is passed (center); cycling through colors in state 3, after V2 is passed for significant time.
Fig. 8 The Audio dress displaying a random color in state 1, after V1 is passed (left); displaying two random colors in state 2, after V2 is passed (center); cycling through colors in state 3, after V2 is passed for significant time.

The Audio dress has two volume threshold levels: V1 and V2. Due to potential noise in the volume input from the surroundings, a volume threshold is not considered “crossed” unless volume remains above the threshold for a significant amount of time. “Significant” time was designed to be shorter for crossing V2 so that effects associated with louder volume were favored.  The four states of the audio dress are described below and graphically in Figure 9:

  1. All lights off.
  2. V1 passed: outer lights turn on.
  3. V2 passed: outer lights change their color; inner lights turn on.
  4. V2 passed for significant time: “Krazy Kolours.” Outer lights continuously cycle clockwise through the color wheel; inner lights continuously cycle counter-clockwise through the color wheel. The rate at which the lights cycle are offset, optimized for aesthetically pleasing color combinations.
Fig. 9 Methods of determining color from sensor input. (a) Volume vs. Time. The Audio dress requires a significant event to change states, but is biased towards higher volume effects. (b) Social dress state diagram. (c) Compass dress RGB color response per degree. For Heart Rate color determination, see Table 2.
Fig. 9 Methods of determining color from sensor input. (a) Volume vs. Time. The Audio dress requires a significant event to change states, but is biased towards higher volume effects. (b) Social dress state diagram. (c) Compass dress RGB color response per degree. For Heart Rate color determination, see Table 2.

Dress 3: Social

The Social dress moves beyond interacting with the physical world and instead enters the human realm: with the use of radio transmitters, this dress allows users to choose and send a color for their friend’s dress. When both users have chosen a color, a trade occurs. This dress allows users to trade over and over again each time both people have chosen and settled on a color, making fashion an entirely new social experience. Due to limited resources, we combined the social dress technology into the Heart Rate and Audio dresses. See the sketches for these dresses above in Figure 5, and the completed dresses in Figure 10.

Fig. 10 The Heart Rate/Social 1 and Audio/Social 2 dresses demonstrating the Social dress. The dresses before (top) and after (bottom) trading colors.
Fig. 10 The Heart Rate/Social 1 and Audio/Social 2 dresses demonstrating the Social dress. The dresses before (top) and after (bottom) trading colors.

Our method of trading colors is diagrammed in Figure 9. To begin, both users start off in user control mode: by twisting their color dials, users can choose any color on the RGB spectrum, and that color will be displayed by the fiber optic bundles of the dress and transmitted by the XBee. When a user has settled on a color for two seconds, the dress goes from user-control mode to ready-to-trade mode. If the dial is twisted again, the user re-enters user-control mode. For a trade to occur, both users must be in ready-to-trade mode and within range of each other (approximately 30m).

Now that both dresses have traded colors, they enter just-traded mode, in which they display and transmit the color they last received from the trade, not the color currently on the color dial. When the dial is twisted again, the user goes back to user-control mode, and the dress displays and transmits the new color. Color trades occur every time both users are in ready-to-trade mode.

Dress 4: Compass

The Compass dress brings fashion’s scope up to a planetary scale: the dress senses the orientation of its wearer relative to the Earth’s magnetic field, and changes the color of its diagonal stripes in response. We chose a diagonal fiber optic design with thicker stripes near the top and bottom according to the principles of neuroscience: by having more light emphasize the upper and lower parts of the dress – away from the waist – the waist should look smaller. A sketch of the compass dress design is shown below in Figure 5, and the completed dress is shown in Figure 11.

Fig. 11 Compass dress color response green (a), blue (b), pink (c), and red (d) as the dress is rotated in a circle.
Fig. 11 Compass dress color response green (a), blue (b), pink (c), and red (d) as the dress is rotated in a circle.

The compass dress spans the full RGB color spectrum as the wearer turns in a circle, based on the angle made with the Earth’s magnetic field (Figure 9). Red, green, and blue each appear as the only color at 0°, 120°, and 240°, respectively. For angle values falling between those pure colors, the dress blends two colors to create a rainbow of hues. Between 0-120°, red and green mix to form orange and yellow as the angle increases; between 120-240°, blue and green mix to form shades of turquoise; between 240-360° (0°), blue and red mix to form various types of purple.

We also took advantage of a degree measurement from another plane to control the amount of white light. As the user leans forward or backward, intensity of the white LED varies up to a quarter of the possible white brightness. We did not employ the full brightness of the white LED in order to not overpower the current RGB color. We chose to keep the white LED separate from the horizontal rotation of the user because we wanted the full mixing between two of the three other colors. The additional dimension of changes in white light makes the dress’ color more dynamic.

Power and Battery Life

We used two different battery sources in our dress designs so that we could compare how the affected the lifespan of the dresses. The Compass dress used a 3.7 volt, 2700mAh Lithium Ion Polymer (LiPo) battery to power both LEDs and the Lilypad, while the Heart Rate/Social 1 and Audio/Social 2 dresses used a standard 9V battery to power the LEDs and a 3.7V 150mAh LiPo battery to power the Lilypad.

The Compass dress’ battery was much flatter (making it easier to integrate into a dress), had a 10 Watt-hours energy capacity, was rechargeable, and yet more expensive ($22.50), whereas the 9V battery was bulkier, had an energy capacity of 5 Watt-hours, was single use, and yet cheaper ($1.50). Overall, we used approximately twenty 9V batteries while testing the dresses. The Compass dress’ battery had a battery life of 140 minutes and the 9V lasted for 27 minutes (Figure 12).

Fig. 12 Battery life of the dresses. (left) This test was carried out with all LEDs running at full brightness in compass sensing mode until the green LED turned off at. Measured battery life for the 2700mAh LiPo battery is approximately 140 minutes. (right) Performance test of the standard 9V battery powering the LEDs and the 150mAh, 3.7V LiPo battery powering the Lilypad and sensors. This test was carried out with all LEDs running at full brightness until the green and blue LEDs turned off. This occurred at the sharp dip at approximately 27 minutes.  The red LEDs remained on after this time.
Fig. 12 Battery life of the dresses. (left) This test was carried out with all LEDs running at full brightness in compass sensing mode until the green LED turned off at. Measured battery life for the 2700mAh LiPo battery is approximately 140 minutes. (right) Performance test of the standard 9V battery powering the LEDs and the 150mAh, 3.7V LiPo battery powering the Lilypad and sensors. This test was carried out with all LEDs running at full brightness until the green and blue LEDs turned off. This occurred at the sharp dip at approximately 27 minutes. The red LEDs remained on after this time.

Longer battery life, as well as the more compact nature of the LiPo, makes the 3.7V 2700 mAh battery a superior option for wearable applications such as our dresses, despite the increase in upfront cost.

Cost

Due to the modular approach by which we made our dresses, there was a baseline cost of $150.84 for each of our prototypes. Each dress used a Lilypad Arduino, 4 RGBW LEDs, 4 couplers, plus assorted electronic components (Table 5). Each dress also had individualized components specific to its requirements; see Tables 6, 7, and 8 for a breakdown of the cost of each dress.

In total, the Compass dress had a parts cost of $317.39, the Audio dress had a parts cost of $628.61 (not including additional Social sensor), and the Heart Rate dress had a parts cost of $723.27(not including additional Social sensor). Two dresses – Heart Rate and Audio – were purchased from Rodarte for $425 each, leading to much of the cost discrepancy between the three dresses. Cost of sensors was the other main differentiator: the Heart Rate was $100, compared to $15 for Audio, $25 for Compass, and $40 per Social dress. The Social dress, while combined with Audio and Heart Rate due to limited resources, should have comparable costs to any of these dresses if built on its own. In addition, the final prototype of the Compass dress took roughly twice the work hours (~30) to produce than did the final prototypes of the Heart Rate/Social 1 and Audio/Social 2 dresses.

Because fashion pricing is increasingly based on whether an item of clothing is “trendy” or a “staple” (Levy, 2004), if we were to manufacture and sell these dresses at their current material cost, our financial success would likely be based entirely on when we enter the fashion market: if our dresses become a trend, our line could be profitable.

Table 4. Cost of common elements.

Element Cost
Lilypad Arduino Main Board $21.95
RGBW LEDs (4) $92.24
Couplers (4) ~$20.00
Perma Protoboard (full size) $6.65
Wire (Ribbon Cable, Solder, Other Assorted) ~$10.00
Total: $150.84

Table 5. Separate costs for Heart Rate dress.

Element Cost
Common Elements $150.84
Sensor (Polar Heartrate) $100.00
Fiber Optic Cable (6270 ft) $45.72
MOSFETS (6), Other Circuit Elements (2 switches , 1 potentiometer) $11.26
Batteries – 9V, lifetime 30 minutes, non-rechargeable; 3.7V 150mAh rechargeable $8.45
Dress (Rodarte Provided) $425.00
Total: $723.27
Work Hours (Fiber Integration) 15

Table 6. Separate costs for Audio Dress.

Element Cost
Common Elements $150.84
Sensor (Electret Microphone) $9.95
Fiber Optic Cable (3033 ft) $22.12
MOSFETS (6), Other Circuit Elements (2 switches , 2 potentiometers) $12.25
Batteries – 9V, lifetime 30 minutes, non-rechargeable; 3.7V 150mAh rechargeable $8.45
Dress (Rodarte Provided) $425.00
Total: $628.61
Work Hours (Fiber Integration) 15

Table 7. Separate costs for Compass dress.

Element Cost
Common Elements $150.84
Sensor (OSEPP Compass) ~$25.00
Fiber Optic Cable (6650 ft) $48.49
MOSFETS (8), Other Circuit Elements (1 Adafruit switch, 1 potentiometer) $10.56
Battery (3.7V 2700mAh, lifetime 120 minutes, rechargeable) $22.50
Dress (Student Made) $60.00
Total: $317.39
Work Hours (Sewing, Fiber Integration) 30

Discussion and Future Work

We successfully created four dress designs – Heart Rate, Audio, Social, and Compass – using three physical dresses. Each dress interacts with its environment, similar to the adaptive coloration found in nature. We shrunk our circuits, LED couplers, and fiber optic bundles from bulky breadboard proofs of concept to lightweight circuits that could be sewn onto clothing, and successfully integrated them into the dresses. Our project shows that fashion can be taken beyond clothing: no longer do clothes need to be a passive element when it is possible for clothing to interact with the world around it.

Our dresses provide a framework upon which to build multiple exciting improvements in wearable technology. By creating garments that can change in appearance and interact with their greater environment, we allow the wearer to play an active role in the way they express themselves through clothing. The future of fashion is not passive fabric, it is interaction.

Yet our dresses are not finished: we had problems with the Social and Audio dresses during our first live demonstration of the dresses at our final presentation. The Social dresses did not trade colors as expected, likely due to cell phone or WiFi interference. This problem can be corrected by selecting radio transmitters that do not operate in the 2.4GHz frequency band used in WiFi and many other devices. In addition, the Audio dress’ threshold values were incorrect for the presentation: the threshold value was hardcoded instead of controlled by the wearer as intended to ensure proper behavior at the fashion show, but the volume of the audience changed the ambient volume level. Further testing of both dress designs would have revealed these issues before our live demonstration.

In addition to technical improvements, we can also expand our fashion line to include new methods of interaction. Throughout the brainstorming process, we considered numerous sensory dresses. The Heart Rate, Audio, Social, and Compass modes are but a fraction of the possibilities we envisioned: for example, we also considered creating an accelerometer dress that changes color as a dancer moves and a dress that changes to contrast (or blend in) with the colors around it. Because our dresses were designed modularly, new sensing dresses can be built directly off of what we have already built.

In line with our overarching goal of bringing fashion beyond the garment, we envision an entire fashion line of modular sensory clothing. Consumers would buy multiple sensors to plug into their clothing, swapping out sensors as they please. However, in order to introduce a market of swappable sensors, we would need to further refine our prototypes’ electronics: right now, the electronics are too large and exposed for our dresses to be sold commercially. Designing and using printed circuit boards would solve this problem. We would also expand our range of sensors, and perhaps continue to include multiple sensors to achieve even stronger color effects. For example, the Audio dress could be modified with additional sensors one so that volume of sound controls color and sound frequency controls intensity.

We could also expand further into the social realm: further increasing interactivity between dresses or integrating with smartphones would make our dresses a new platform for communication between friends. For example, friends from opposite ends of the world could send patterns to each other. A dress could reflect the colors present in a photo that the wearer takes. Dresses could respond to the daily weather, or the rise and fall of a stock’s value. At a concert, an artist could broadcast information so that the audience and their clothing become a part of the light show.

Fashion can most definitely be taken beyond the garment.

Acknowledgments

This work was completed in partial fulfillment for the spring 2013 course Engineering Sciences 96: Engineering Problem Solving and Design Project. We are indebted to the teaching staff, Dr. Kevin Kit Parker and Holly McIlwee Golecki, for their abundant advice and unwavering support throughout the project. We would also like to specially acknowledge Josue Goss for sharing his prototyping and materials expertise. Special thanks also to Kate Mulleavy and Laura Mulleavy of Rodarte for their insights into the fashion world. Dr. Leila Deravi provided images from her own research for this work. Others who advised us during the project – on everything from branding and marketing of products to the physics of light and sewing – are Dr. Frederick Abernathy, Dr. Margaret Livingstone, Dr. Evelyn Hu, Bobby Riley of Soldier Design, Mei Tseng, Chriztine Foltz, Liz Dean, and Mick Sawka. Project funding provided by the Harvard School of Engineering and Applied Sciences.

References

Think Geek WiFi Dectector Shirt. (2012). Retrieved from Think Geek: http://www.thinkgeek.com/product/991e/

E-Sewing: LilyPad Arduino. (2013). Retrieved from Lilypad Arduino: http://lilypadarduino.org/?page_id=1260

HDE. (2013). Retrieved from Hottest Deals Ever: http://www.shophde.com/ledtshirt.html

LumiGram Catalog. (2013). Retrieved from LumiGram, LLC: http://www.lumigram.com/catalog/

Rodarte. (2013). Retrieved from Rodarte: http://rodarte.net/

Tadin D, et al. (2003). Perceptual consequences of centre–surround antagonism in visual motion processing. Nature, 312-315.

Enger D. (2012, July 12). Lost in the Wilderness: The military’s misadventures in pixelated camouflage. Slate.

Heimbuch J. (2009, December 11). OLED Space Dress and LED Wedding Dress – Next Wave or New Waste? Retrieved from Treehugger.

Levy M, et al. (2004). Emerging trends in retail pricing practice: implications for research. Journal of Retailing, xiii–xxi.

Mäthger LM, et al. (2009). Mechanisms and behavioural functions of structural coloration in cephalopods. J. R. Soc. Interface, 6(Suppl 2): S149–S163.

Livingstone M, Hubel D. (1988). Segregation of form, color, movement, and depth: anatomy, physiology, and perception. Science, Vol. 240 no. 4853 pp. 740-749.

Roberts SC, et al. (2010). Distinguishing Between Perceiver and Wearer Effects in Clothing Color-Associated Attributions. Evolutionary Psychology: an International Journal of Evolutionary Approaches to Psychology and Behavior, 350-364.

Comments:

NO COMMENTS

LEAVE A REPLY