Make

Introduction

This section describes how I made Shoji Pixels, briefly, mainly for people who are interested in engineering part of this project. Media Art is, as far as I believe, on one hand engineering which involves electrics, mechanics, machining and assembling. It is on the other hand conventional Art which reflects inner passion of humans. In view of engineering, the employed technology, including both hardware and software, is not that state-of-the-art but rather I used existing hardware/software parts -- almost all of them are cheap and easy to be obtained -- to integrate all those ones into union. So, maybe young students who want to start their projects would be the best readers.

For more detailed record, please go to my make diary.

General Hardware Structure

The size of device is about 100 cm in square and 60 cm in depth. It uses a grating for the base frame, which is basically used to separate interior space. 100 linear actuator units are attached in the holes of grating, connected by electrical wires and controlled by micro-controller. Each unit has one servo motor with gears that converts motor rotation to linear movements. The rods move the paper-made LED shades back and forth in the range of 40 centimeters.

Software and Signal Flow

The slide show illustrates a general idea of the software structure of this project. Socket is used for inter-PC communication. The Blender software handles the artistic part in which we can edit the movement and the colors of rods using virtual 100 Pixels. What is good for people who make animation of the device is that Blender inherently has a timeline function and can record any animation in variety of ways. Blender exposes almost all the functions using python, and Shoji Pixels has its own Blender-python script to send the color and position information to the device server.

The device server is written separately in python, to make the functions independent from each other. It also accepts the data from sensors (and maybe from the Internets) such as Microsoft Kinect, which is converted and sent both to Blender and the Shoji Pixels device.

Finally, the device server emit the color and position information to three micro-controllers in the Shoji Pixels device via conventional serial pipe. What micro-controllers do is just receive and through some byte array as it is to the daisy-chained PWM generators, TLC5940 by Texas Instruments.

Making Circuits

Electric work was the toughest part. All the PCB patters are designed in the Eagle software and milled out using my personal CNC machine, RD-420 (Originalmind, Nagano, Japan), after several modifications. Unlike basic tutorials on the Internet, the project had to drive many servos and LEDs at the same time, independently in some high frame rate so that humans cannot recognize transitions.

Usually, servos and LEDs are controlled with independent circuits and lines in many applications, but in this project I made up my mind to control both of them in a more united way, because "servos and LED can both be controlled by PWM (Pulse Width Modulation) signals. This results in simpler implementations that even an electric novice was able to achieve such a fairly big project.

The electrical noise disturbed the signal integrity the most. Until the last stage of the development, I had kept modifying the circuits studying the signals with my USB oscilloscope and logic analyzer. Basically, thick and big ground line did some good work, but I used many other techniques to suppress the noise and make the whole system work.

Making Linear Actuator

Mechanical part was rather simple compared to the electrical part, but this also required continuous modification throughout the project. Gears, body parts, rack and pinion and base for Shoji pixel are all made using my personal CNC machine. I might be able to use some injected gears, but there are no rack and pinions available for this project, so I decided to make all the mechanical parts with the personal CNC machine.

Cables are also hand-made because available readily-usable cables are too thick and heavy for this project. To make the module as light-weight as possible, the bases of Shoji pixels are milled out from thick carton boxes, which I picked from the garbage yard. The LED modules are also small and designed in a round shape to put it at the tip of the aluminium pipe. Then I assembled over 100 of them by hand.

There are many troubles in the mechanical part because differently from electric circuit, it moves. Plastic gears broke, the rack got off from the rod, servo got burnt because of the heavy load. After some replacement and refinement, now it seems working well.

Making Shades

"Shoji Pixels" was named so because the LED shades are made of Shoji papers. I originally planned to use some embossed plastic plate or special paper reinforced by plastic. I made some prototypes, but after all I just used the Shoji paper as it is used in Japanese houses with a little bit reinforcement by acrylic lacquer spray.

The process was kind of old-fashioned. First, I milled a wood piece to make a base to which a rolled paper is to be cast. This process make the rolled paper square in shape of the same size. Then, the squared paper roll is dried under the sun after water is sprayed. Then, acrylic lacquer is sprayed and again it is dried under the sun. Making 100 of them is just tougher than I had thought.

Cost

Each component is very cheap. I only used the cheapest-grade electric parts, wood, general plastic and aluminium to construct the whole device. The CNC machine, RD420, was the most expensive ($4000), but except for the machine, the parts costed only around $2000. I can make the replica of the device maybe with only $1000 because I bought unnecessary products, but I have to recruit some technicians who can copy the whole process. The estimated development cost would be less than $10,000 considering the labor cost.

Future

I used Shoji paper not because I chose it from many alternatives, but because it has been used in my country's culture and has been everywhere around me since I was a child. I hope my work will inspire people who want to make new things in the near future.

History

It has been three years since I got the first idea of this work. I thought I quit making many times. I, however, was very excited and happy to see the final device that works perfectly fine, just as I was intended in my brain and in my heart.

2011/09 Idea conceived

2012/09 Start making prototype

2013/05 My CNC Machine employed in my room

2013/08 Resigned from Japan Media Arts Festival 2013

2014/07 Finished making first prototype

2014/07 Submitted to Japan Media Arts Festival 2014