THE EVOLUTION OF LIGHT PAINTING AND THE CONSTRUCTION OF THE ORIGINAL BUSHWICK LIGHTBOX
The construction of my first large 2,240 pixel WS2812B matrix -- the original "Bushwick Lightbox -- began after initial experiments in Fall 2016 with a small 100 pixel lightbox. After seeing the capabilities of the 100 pixel lightbox and the potential of this concept of light painting, I began construction of a much larger LED matrix in order to explore the full potential of the concept of light painting.
On the 100 pixel LED matrix, I had the observation that when you project two LEDs of differing intensities on to a surface, the brighter one seems to come forward, while the darker one seems to retreat backwards. This opposing visual movement, backwards and forwards, gave rise to my observation that I could use LED to construct visual images on a surface in the same way that painters used paint to construct images on a canvas. This was the first image on a 100 pixel (10 rows x 10 columns) lightbox.
In January 2017, I began construction of the 2,240 pixel LED matrix. The size of the matrix, and therefore the size of the lightbox, was largely determined by the default size of plywood I was able to buy from Home Depot, which was sold in 48" x 96" sheets.
Factoring margin space needed to solder connections between the rows of LED strips, it is 70 columns by 32 rows, totalling 2,240 WS2812B LED pixels.
The initial design you see to the left considered the necessity of power-tapping the LED strips -- however this turned out to be insufficient to power all the LEDs.
Upon further research, I learned that the voltage drop across rows of 70 LEDs was simply too much to overcome by a single power tap.
The solution (upon advice from colleagues at the Fat Cat Fab Lab) to ensure ample power supply was to run a power and ground bus on each side of the matrix, using a 12 gauge wire in order to be sure power would flow freely, even if I ran the matrix at max brightness.
This worked and the power bus method became my modus operandi for constructing LED matrices for over a thousand LEDs.
One issue was how to solder stranded 22 AWG wire to a solid-core 12 AWG wire, because it took so long for the 12 AWG wire to heat up through soldering iron alone.
I solved this problem by purchasing a small blow torch to pre-heat the wire before attempting to apply solder. Strangely enough, after that initial heating, it seems to be easier for solder to bond even with just the iron.
Later, in larger LED matrices -- I built a 3,750 pixel LED matrix in 2018 -- I avoided this problem altogether by using stranded 12 AWG wire and just threading the smaller 22 AWG wire through it.
After solving the power issues, I brought the matrix home and started to experiment but encountered another problem -- the board was not responding according to the Arduino code.
After browsing Arduino/Neopixel forums for a few days, it turned out the Arduino I was using did not have enough SRAM (static RAM) to push the data through the entire matrix.
Each WS2812B pixel requires 3 bytes of SRAM -- therefore 2,240 x 3 = 6,720 bytes of SRAM meanwhile the Arduino Uno I initially started out with only had 2K SRAM.
At Fatcat Fab Lab, a friend had an extra Arduino Mega, so I tried with that and it still didn't work, although more pixels lit up. The break in the case was ordering an Arduino Due, the most powerful Arduino on the market, with a whopping 96KB of SRAM. This solved the issue.
With the data and power issues solved, the matrix was fully responsive to arduino code.
This begun my exploration of LED diffusion distances. I didn't make a lightbox frame immediately after completing the matrix, because I wanted to explore the character of using different standoff distances between the projection material and the LEDs.
I found that the "ideal" distance was not a straightforward affair, but rather a function of the specific pattern I programmed into the matrix, the specific visual effect I was going for, and the type of diffusion material I was using.
I discovered different types of paper had different indices of refraction and required different distances.
Have concluded that the optimal standoff distance for my purposes was 2", I began construction of the frame that the diffusion material would sit on top of.
I decided upon oak as a good material, knowing little about wood because I thought it was "nice".
Little did I know that hardwoods were very difficult to work with, as well as heavy, and in hindsight, I regretted that choice.
However, later light boxes that I made using lighter softwoods such as pine, I noticed that bowing was a real issue with softwoods.
So, in hind-hindsight I don't regret the choice of oak, as the hardwood tends to be straighter and be less resistant to damage. As this lightbox was destined to go out into the world of pop-up street shows, it turned out this sturdy material served this purpose well.
With that, in Summer 2017 I did the first season of Bushwick Lightbox, a weekly pop-up light art show on the street. I did about 8 shows in July-August. The lightbox was powered by a 5V, 350W DC power supply running on a SUAOKI 400Wh/120,000mAh Portable Generator Power Station Power Supply.