Some SuperCollider extensions by Martin Marier
SuperCollider QMake Pure Data
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
HelpSource Add helpfile (but incomplete). Jun 19, 2011
interpolator
.gitignore Add helpfile (but incomplete). Jun 19, 2011
BusPlot.sc Looper: start and end position is stored in a Bus to assure sync. Oct 11, 2011
COPYING Added COPYING. Oct 19, 2010
Feature.sc Ajout BP Feature. Feb 19, 2015
FeatureGui.sc
JackMatrix.sc Add option to Use aliases instead of port names. Feb 29, 2016
Looper.sc Fix initialization issue in SCClassLibrary. Feb 29, 2016
README.md Small modifications to README file. Jun 13, 2012
ServerLadish.sc Add new version of Interpolator. Sep 4, 2012
Sponge.sc Ajout BP Feature. Feb 19, 2015
SpongeButtons.sc Sponge Buttons: make double-click longer, avoid bounces. Oct 17, 2013
SpongeGui.sc Double click on Feature opens FeatureGui. Aug 19, 2011
SpongeOSC.sc Make sure Feature class is initialized before Sponge. Jul 15, 2016
Trimmer.sc Make Sponge savable. Sep 20, 2012
example.scd Add setOSC... methods. Jun 19, 2011
presetInterpolatorPatch.pri
serialSponge.pd Add a SpongePD object. Sep 10, 2012
serialSpongeSLIP.pd SpongePD uses pdexetended and is adapted for SLIP protocol. Oct 26, 2014
udpSponge.pd Add a raw UDP sponge, spongeOSC that reads the OSC blobs. Oct 8, 2012

README.md

Flattr this git repo

The mmExtensions for SuperCollider

At this point, this repository contains the files required to use to the PresetInterpolator and the Sponge.

Installation

These sc files go inside your SuperCollider Extensions directory:

~/share/SuperCollider/Extensions/ (Linux) ~/Library/Application\ Support/SuperCollider/Extensions/ (Mac OSX)

Dependencies

  • SenseWorld DataNetwork Quark (by Marije Baalman)
  • wslib Quark (by Wouter Snoei)
  • ddwGUIEnhancements
  • crucial library

Description

PresetInterpolator() is a preset interpolation system similar in concept to the Metasurface found in AudioMulch (created by Ross Bencina) or the max patch SpaceMaster (by Ali Momeni and David Wessel). There is an important difference, though: PresetInterpolator() can interpolate spaces of any number of dimensions.

Interpolator() is the interpolation system used by PresetInterpolator(). It does not do much: it simply outputs weights.

How to use it

These classes are not documented yet, but here are a few notes to get you started.

There is only one parameter to Interpolator: the number of dimensions. This creates a 2D interpolator:

x = Interpolator(2);
x.gui; // nice GUI.
  • The (bigger) grey dot is the cursor;
  • The colored dots are the data points;
  • The size of the transparent circles around the data points is proportional to the weight of that preset;

In the Interpolator window:

  • Double-click in the area to add new data points;
  • Alt-shift-click deletes a presets;

When using the Interpolator to interpolate between presets:

(see PresetInterpolator below)

  • Alt-click and drag to duplicate a data point (a preset);
  • Alt-click and drag also works on the cursor (the grey point);
  • Double-click on a preset to edit it;

A PresetInterpolator takes an Interpolator as an argument. If no argument is provided, a 2D Interpolator is created for you; y = PresetInterpolator(x); y.gui;

In the PresetInterpolator window:

There is a row for each data point (which is associated with a preset). Here is a description of the columns:

  1. The preset name (pressing enter after editing the textField).
  2. The "Attach" button. This button attaches the data point to the cursor. Every time the cursor is moved, the attached point follows it. This is useful to position a point in a multidimensional space using an interface. There will be an screencast of an example soon.
  3. The "Edit" button. Opens a PresetGui window (edit the preset).
  4. The data point id (an integer).
  5. The data points coordinates.
  6. The "Remove" button (X).
  7. The weight of the point.

The last row is different: 1. "Cursor". 2. "Edit" button. 3. One PopUpMenu for each coordinate. (See note) 4. "2D Gui" button: creates an interpolator GUI (2D).

The PopUpMenus have 3 possible values: nothing, x, y. They are used to select which coordinates will be used by the 2D GUI. Especially useful to visualize multidimensonal data.

In preset window:

  • Click on the 'Add' to add more parameters. The presets can have any number of parameters and these parameters can represent anything.
  • Click on the '+' to edit the parameter's ControlSpec.
  • Click on the "O" button to open the OSC preferences of the parameter. (Temporarily not working.)
  • Click on the "M" button to open the MIDI preferences of the parameter. (Temporarily not working.)
  • Parameters can be named.

Each preset of a PresetInterpolator always have the same number of Parameters. Changing the Spec or the name of a Parameter will change the spec or the name of the corresponding Parameters of other presets. Example: if I rename the third parameter of a preset to "foo", the third parameter of all presets will be renamed "foo".

y.cursor; // this 'preset' represents the cursor

When adding an action to the currentPreset's parameters, moving the cursor does something!

(
y.cursor.parameters[0].action_({|mapped, unmapped|
    "Parameter 0: ".post;
mapped.postln;
});
)

Each parameter has a ControlSpec (it can be edited in the GUI as well)

y.cursor.parameters[0].spec_(ControlSpec(20,20000,\exp));
y.cursor.parameters[0].spec_(\midi); //this works too.

// Proper doc is on the way.

The algorithm

It is not a planetary model (weight = inverse of distance). It is not natural neighbour interpolation (like in the metasurface). It is based on intersecting circles. Here is how it works (briefly):

1- A (invisible) circle is drawn around each points (including the cursor). The radius of each circle is equal to the distance to the nearest neighbour. This means that the size of the circles varies when points are moved.

2- The points which have a circle that intersects the cursor's circle have a non nil weight. (Not all points do, but at least one does.)

3- The weight of each point is calculated like this: (intersecting area of the 2 circles) / (total area of the circle).

4- Then we do a weighted sum for each parameter using these weights.

I know, it is not very clear. Just look at the code!

And it works in spaces with more than 2D (I tried up to 10D).

Cheers!

Martin Marier