Hobby Quest - Desk Rave Prototype using Total Control Lighting

Back in mid-2013, I was inspired by this awesome LED light display I saw at a friend’s wedding at the Crucible in Oakland, CA.   This youtube video gives you an idea of what it looks like, but is no where near as compelling as seeing it in person:



I found out that you could buy strips of these programmable LEDs from coolneon.com and drive them with an Arduino. I’ve always wanted an excuse to play with Arduino and so I decided to embark on a hobby quest to create a beat driven display using the 50 LED model (the many LED display you see in the youtube above is about a $1500 dollar rig while the 50 LED model is only $100).

I bought the parts to begin putting together the hardware - a Seeeduino v3.0, a Total Control Lighting developer shield, and the 50 LED strip model.  More details of hardware, software implementation details, etc. can be found in the README at the github repo for my project:  https://github.com/mplanck/sound2light (all code for my project is available there).


After some arts and crafts, I created an evenly spaced strip of lights:



At first, I began to write simple Arduino firmware snippets to have sine waves and color fractals flashing down the length of the LEDs.  I discovered a few things.

• The refresh rate of the Arduino across the 50 LEDs is fast (fast enough for any display purposes).
• The energy linear to perceptive linear gamma mapping for the lights is around 3.6.  I found this mapping by writing a simple bit of firmware that would display an energy linear ramp for each color channel (red, green and blue), and then I used a potentiometer on the Total Control Lighting developer shield to tune the gamma curve until it looked like the darks and brights ramped evenly.
  
• I also discovered that if I wanted to write more sophisticated firmware, I’d have to fight the 32k memory limits of the Arduino board memory.  
• The time it took to save, compile, and load firmware on the device made iteration a pain in the butt.

With this in mind, I decided to write a generic bit of firmware that would listen for serial based instructions on how to set the color across the 50 LED strip.  This way, I could fumble around more easily writing the "sophisticated logic" in python and send the results to the Arduino via the USB connection.  Unfortunately, I soon discovered that the 115200 maximum baud rate of serial communication severely limits the refresh rate of the display.  Considering that I’m sending 3 bytes per light (1 byte per color channel), I’m limited to having at best around 30 fps or else I may have serial buffer overrun.  I discovered (a bit late) this bit bang approach which is probably the right way to go.  Something to do in the future.

But I was getting some okay results with my first approach so I decided to write the python part of the rig.  I found a cool python modules for audio processing (pyo) and a way to sniff the audio signal coming through any mac (soundflowerbed).  With these handy utilities, I started writing some python to get an effect where the lights would display audio by having every “beat” create a light wave that originated from the center of the lights and then travel towards the sides.  The intensity of the beat (the ratio of an energy spike to the floating average) would determine the initial velocity of the wave.  As the wave would travel, I'd damp the velocity as if the wave was passing through a thick liquid.  In order to detect “beats” I was inspired by this article at gamedev.  I decided that I’d find beats on several different frequency bands and then have each light wave colored based on the frequency of the beats.

Here’s what I’ve got so far:


 
I’m hoping to make some modifications in the near future:

• Use the bit bang approach to improve display rate
  
• Instead of detecting a beat as a spike in the energy average of each frequency band, I think I could get something interesting by creating a beat wave whenever the energy has a positive differential or climbs.  This will result in A LOT more beat waves but I think will have more interesting “texture”.
  

And perhaps with the 50 LED experiment under my belt, I’m ready to go to the next level - perhaps in 2 dimensions!  Something like a 15 by 10 display (with 3 LED models strung together)?  I’ll keep posting to the blog with progress.