> i was thinking more of a metasynth-like program - you import a png or > something into the software and it turns it into a manipulable sound > scape. The xmms visualisation plugins can turn sound into images - so the other way around shouldn't be impossible. Koan (I think) used to be used by Brian Eno for 'generative music', so it should be feasible to use the colour and pattern properties of an image to feed the input of a similar program on Linux. Cheers Daniel