It was actually a completely useless test. I wrote a small code to perform some Sibson's interpolation (a smooth nearest neighborhood). To make it simple the idea here is to get the Voronoi cell of a point wandering (following a Lissajous curve trajectory) among 1000 random still points. As you say, the result is "interesting".

Ooh, a Lissajous pattern!

Perhaps, but I don't think so. It looks more like it's made of bezier curves.

Lotic is right, the trajectory of the point is a Lissajous curve. The moving point is among 1000 still points and his Voronoi cell is computed at each position (each frame is a combination of 255 cells for 255 positions). I don't have the parameters in mind yet you can find them easily by displaying the map of the full trajectory (max value on each pixel location for every frames of the animation).

Facile, c'est la boule qui apparait à l'écran lorsque le lecteur dvd est en veille.

Je devrais faire carrière dans la conception d'écran de veille. Qu'en dites-vous ?

I saw a cartoon where a scientist is said to be either thinking about infinity, or that his eight fell down.

8 : { x(t) = sin(2.t) & y(t) = sin(t) }
∞ : { x(t) = sin(t) & y(t) = sin(2.t) }

Tien ... ca serait bien pour faire un fond d'ecran de veille pour ordi

Merci. En fait, l'idéal serait une animation où chaque pixel présente une chance équiprobable d'être éclairée, or ce n'est pas le cas ici (il suffit de cumuler les frames pour s'en rendre compte).