Matt Pyke, of Universal Everything, and his brother Simon are teaming up with visual designers around the world to create a DVD project called Advanced Beauty. The project brief is simple enough. Take a piece of original audio created by Simon and make a motion graphics piece for the audio. It must start with a blank white frame and end with a blank white frame, but everything in the middle can be whatever you desire.
Matt Pyke has released a trailer recently which can be viewed at the Advanced Beauty site. My piece isnt represented in the trailer. Reason? Well, I havent finished it yet. Sadly I ran into a weird bug with JSyn and Sonia in Processing where JSyn poops out after about 13.5 hours of rendering. The JSyn people are aware of the bug and have been for a long while. The fix for this bug is nearing completion but not soon enough for me to take advantage of it. Other options, like preprocessing the audio or adding break points, would have worked fine but I was feeling lazy and didn’t want to bother with finding new ways to do something I was already happy with.
My original piece was going to take about 30 hours to render. This obviously wouldn’t work so I had to do some rethinking and recoding and the end result brought the render time down to about 5 to 6 hours, and cut some 50 GBs off the final video size. Hurray! Here are some stills from the near-final piece. With the big hurdles out of the way, I expect to have my final piece to Matt and Simon sometime in the next week.
The audio from Simon is a little over 4.5 mintues long. Since 4.5 minutes of direct audio analysis might get too muddy or boring, I asked Simon to provide me with separated tracks which he happily did. Using Processing (surprise surprise), I analyzed 7 different tracks in 1/25th of a second blips and used that data to influence the visuals which are based on the Magnetosphere engine. There are up to 200 gravity orbs and up to 10000 charged particles. Each particle can leave a ribbon trail up to 75 segments long. Additionally, each particle can leave behind up to 3 spark particles per frame and each gravity can leave behind up to a dozen nebula images per frame.
no idea whether this is of any use, but i had a similar problem using sonia in an installation i built that needed to be up and running round the clock — i just put in a timer and stopped and restarted sonia every 12 hours. admittedly i was dealing with liveinput audio which you’re not, but could you not simply leave a pointer for the sample that you’re analysing just before stopping and restarting sonia?
fwiw, my liveinput code is below, you’ll see i basically just stop, restart and recreate everything, without any glitches.
//
LiveInput.stop();
LiveInput.start(2048);
recObj = new Sample(numsamples);
LiveInput.startRecLoop(recObj, 1,numsamples);
recObj.setRate(recrate / octavediv);
recObj.repeat();