KmeansToBPSet1 | SuperCollider 3.12.2 Help

Source: SLUGens.sc

//SLUGens released under the GNU GPL as extensions for SuperCollider 3, by Nick Collins, http://composerprogrammer.com/index.html

Description

Uses succesive iterations of a k-means clustering algorithm on random data with random initial means to form break points in a 2-D space. These are then converted to wavetables in output synthesis based on the oscillator frequency.

Class Methods

KmeansToBPSet1.ar(freq: 440, numdatapoints: 20, maxnummeans: 4, nummeans: 4, tnewdata: 1, tnewmeans: 1, soft: 1.0, bufnum, mul: 1.0, add: 0.0)

Arguments:

freq	frequency of oscillator
numdatapoints	Initial number of data points
maxnummeans	Maximum number of means allowed
nummeans	current num means
tnewdata	If a trigger is received, reset source data (k-rate)
tnewmeans	If a trigger is received, reset means (k-rate)
soft	Proportion to soften means update. If 1.0, immediate update, 0.0 no update, proportion inbetween amount to move from the old mean position to the new mean of the allocated cluster (ie soft k-means).
bufnum	Number of a Buffer which contains positions for means and data points; see examples below

Inherited class methods

Instance Methods

Inherited instance methods

Examples

//defaults
{Out.ar(0,Pan2.ar(0.1*KmeansToBPSet1.ar,0.0))}.play

{Out.ar(0,Pan2.ar(0.1*KmeansToBPSet1.ar(440,20,8,8,Impulse.kr(5),0),0.0))}.play

//softened
{Out.ar(0,Pan2.ar(0.1*KmeansToBPSet1.ar(440,20,4,4,0,Impulse.kr(5),MouseX.kr(0.0,1.0)),0.0))}.play

{Out.ar(0,Pan2.ar(0.1*KmeansToBPSet1.ar(440,20,4,4,0,Impulse.kr(5)),0.0))}.play

{Out.ar(0,Pan2.ar(0.1*KmeansToBPSet1.ar(440,20,4,4,Impulse.kr(5,0.5),Impulse.kr(5)),0.0))}.play

{Out.ar(0,Pan2.ar(0.1*KmeansToBPSet1.ar(MouseX.kr(200,880),20,4,4,Impulse.kr(MouseY.kr(1,100,'exponential'),0.5),Impulse.kr(LFNoise1.kr(0.2,5,6))),0.0))}.play

{Out.ar(0,Pan2.ar(0.1*KmeansToBPSet1.ar(440,30,10,MouseY.kr(1,10,'linear'),Impulse.kr(2,0.5),Impulse.kr(LFNoise1.kr(0.2,5,6))),0.0))}.play

//pushing up amount of data and num means (also increases CPU cost)
{Out.ar(0,Pan2.ar(0.1*KmeansToBPSet1.ar(440,100,20,MouseY.kr(1,20,'linear'),Impulse.kr(0.2,0.5),Impulse.kr(LFNoise1.kr(0.23))),0.0))}.play

//pushing up amount of data and num means (also increases CPU cost)
{Out.ar(0,Pan2.ar(0.1*KmeansToBPSet1.ar(MouseX.kr(20,500,'exponential'),200,40,MouseY.kr(1,40,'linear'),Impulse.kr(0.57,0.5),Impulse.kr(LFNoise1.kr(0.53))),0.0))}.play

//softening
{Out.ar(0,Pan2.ar(0.1*KmeansToBPSet1.ar(MouseX.kr(20,500,'exponential'),200,20,20,Impulse.kr(0.57,0.5),Impulse.kr(LFNoise1.kr(0.53)),MouseY.kr(0.0,1.0)),0.0))}.play

//softened
{Out.ar(0,Pan2.ar(0.1*KmeansToBPSet1.ar(44+Dust.kr(10,10),30,10,10,Impulse.kr(0.1),Impulse.kr(LFNoise0.kr(0.4,0.2,0.3)),MouseX.kr(0.01,1.0,'exponential')-0.01),0.0))}.play

//modulate
{Out.ar(0,Pan2.ar(0.1*KmeansToBPSet1.ar(A2K.kr(44+KmeansToBPSet1.ar(MouseY.kr(0.1,1000,'exponential'),soft:0.1,mul:50,add:60)),30,10,10,Impulse.kr(0.1),Impulse.kr(LFNoise0.kr(0.4,0.2,0.3)),MouseX.kr(0.01,1.0,'exponential')-0.01),0.0))}.play

//chordal sequence with character
{Out.ar(0,Pan2.ar(Mix(0.1*KmeansToBPSet1.ar(220*[1,5/4,3/2,11/7],20,4,4,Impulse.kr(0.2),Impulse.kr(10),MouseX.kr(0.01,0.5,'exponential'))),0.0))}.play

//using a Buffer to select positions for data and reset positions for means

//maxmeans will be 5, 20 data points

b = Buffer.alloc(s, (5+20)*2, 1); //buffer big enough

//random positions- but they'll be the SAME random positions when resetting
b.setn(0,Array.fill(50,{1.0.rand})); //set x,y points (in [0,1] by [0,1] for this UGen)

{Out.ar(0,Pan2.ar(0.1*KmeansToBPSet1.ar(MouseX.kr(100,1000,'exponential'), 20, 5, 5, Impulse.kr(1), Impulse.kr(1), MouseY.kr(0.01,1.0),b.bufnum),0.0))}.play

//change buffer when you like:

//change data
b.setn(0,Array.fill(40,{1.0.rand})); //set x,y points (in [0,1] by [0,1] for this UGen)

//change means
b.setn(40,Array.fill(10,{1.0.rand})); //set x,y points (in [0,1] by [0,1] for this UGen)

//buffer version
{Out.ar(0,Pan2.ar(Mix(0.1*KmeansToBPSet1.ar(220*[1,5/4,3/2,11/7],20,5,5,Impulse.kr(10),Impulse.kr(10),MouseX.kr(0.01,0.5,'exponential'), b.bufnum)),0.0))}.play

//routine playing with mean initial positions
(
c= {inf.do{

b.setn(40,Array.fill(10,{1.0.rand}));

rrand(0.1,0.2).wait;}}.fork

)
c.stop;

//routine playing with data positions

(
d= {inf.do{

b.setn(0,Array.fill(40,{1.0.rand}));

rrand(0.1,0.2).wait;}}.fork

)

c.stop;
d.stop;

helpfile source: /usr/local/share/SuperCollider/Extensions/SC3plugins/SLUGens/HelpSource/Classes/KmeansToBPSet1.schelp
link::Classes/KmeansToBPSet1::

KmeansToBPSet1 : UGen : AbstractFunction : Object Extension