flocking study

The patch documented above uses a similar parameter space to the previous study, with each ellipse representing a static tone in a constellation, and the x,y coordinates of each ellipse determining the frequency and relative placement (phase) of the occurrence of that tone.  Here's some SC code for it: Boid.sc, boid_test_0.scd.  This particular example uses a random pitch constellation, instead of the one for Sanction of the Victim.  Boid.sc is directly based on Shiffman's Processing class, itself directly based on Reynold's synonymous algorithm, commonly used in computer graphics.

It's kind of amazing and wonderful that different self-organizing algorithms make such wildly different outputs, even with the same (or very similar) parameter mapping.  This one reminds me of tape delay.  A nice tweak to the parameter space could be to have the left edge (x=0) be rationally related to the right edge (x=width), and let there be a more smooth fade between those two states.  That way, there would be no discontinuity of frequency, and the x axis would comprise a Risset-Sheppard phenomenon (ritardando or accelerando).  I will also be playing with tactics for increasing sparseness in the sound masses, such as constraining the x-mapping to comparatively low frequencies, or making entrances and exits happen for one pitch at a time.  Also, the prospect of applying this parameter space to sampled sounds might be interesting, so instead of letting each boid represent a different pitch, let each boid represent a brief sampled 'grain' from a soundfile.  That way, the listener might have more attention to devote to the sequence of the firings.

Leave a Reply

Your email address will not be published. Required fields are marked *