The Wishflower is an interactive wooden sculpture and lightshow.
Check out the video:
Conceptually, I wanted a pretty wishing well that wouldn’t need coins or valuables as payment – only your personal touch. You put your hand on the skull, close your eyes and think of something you want deep in your heart. The skull’s sensors see your skin conductivity, temperature and humidity and the motors inside the skull start vibrating to let you know it’s thinking.
The vibration gets louder and more intense – and when the skull’s analysis is ready the altar’s tree root horns flash in different colors and your inner thoughts are translated as a play of colors on the hexagons.
Also, technically there were a couple of things I wanted to achieve:
- I wanted to be able to separate the hexagons into individual standalone pieces if needed. Thus, all are self-contained with their own processors, either controlled by a central server or running on an automatic sound reactive program.
- I wanted the whole thing to be able to be easily assembled and setup.
- I wanted everything to be weatherproof. The first installation was outdoors in Finland on Kosmos Festival 2016, so I knew rain was very likely. All the LEDs are plastic shelled and epoxy dipped, the wooden parts were treated to be waterproof and all the seams were silicone coated to prevent water from leaking in.
Hardware: Each of the wooden hexagons has 149 WS2801 IP68 5v LEDs, a 10A power supply and a Raspberry Pi 3. I put a separate Raspberry Pi into each of the hexagons because I wanted to be able to use them as standalone devices separately. Only a regular power cable goes to each of the hexagons. Raspberry Pi 3 is wifi-enabled, so I can configure and run the software on the hexagons completely wirelessly.
The front face of the hexagons is 6.5mm birch plywood, water jet cut to vector graphics I drew on Adobe Illustrator and converted to .dxf for laser / water cutting. Laser cutting the wood would have burnt the edges dark, so I water cut the hexagons to leave the five beautiful layers of plywood visible from the sides. The wood was treated with dark brown wood oil to make it more weather resistant and to bring out the wood grain and any natural imperfections. At first I studied different kinds of diffusion films for the lights, but decided it’s best to let the light reflect naturally off the wooden backplate inside the hexagons. All the LEDs are hidden behind the front face, so no LEDs are actually visible to viewers – only the glow from behind the water cut wooden pattern.
Software: The program on the Raspberry Pi that runs the LEDs was written from scratch by me in Python. It listens to TCP commands from the deer skull server and controls the LEDs via SPI. The program reads specially rendered 128px * 128px video files and blends them frame by frame to the LED colors, and also mixes in so called “blooms” controlled by the skull’s sensors’ parameters. No two touches, and thus no two wishes are identical.
Skull server and sensors:
All the hardware was installed into a real deer’s skull (no deer were harmed for this installation). It was an old, dried up skull found in the forest that I repurposed for this installation. Only a little bit of drilling & sawing was needed to make the insides big enough for the electronics. The skull acted as a natural waterproof housing for the “brain” of the installation. The altar the skull was on was shaped like a large skull itself – made from green spruce branches and upturned tree roots with two strips of waterproof WS2812B LED strips running inside them.
Inside the painted fluorescent orange skull is a Raspberry Pi and connected to it, an Arduino Nano with some sensors (a very simple custom electrodermal activity sensor and a DHT11) and two Xbox 360 rumble motors glued inside it. Temperature, humidity and skin conductivity sensors tell the Arduino the wisher’s personal profile. The Arduino is connected through USB serial interface to the “brain” Raspberry Pi, which controls the hexagons through WLAN.
This might seem unnecessarily complex: It’s because most of my installations get a few iterations, and I wanted to make this thing as modular as possible from the start.