Li, this also sound interisting. for the fishes rule, I have to say that in a computational way, rule 2 must be splitted in two...
anyway... I'm using perlin noise to create what may be the rule number 2 in my previous post, this generate a force added to other rules. This may come from a bitmapdata with noise updated everyframe by setin the offsets points and make it move (you'll find examples if you google on smoke or whater effect in as3), then get some pixels values with the getPixel() method. This give a more natural moves. BUT this can't be the only way... this has to be added to : fallow a leader , ovoid and interpolate. BUt I begin to think that this, with the away3D pipeline of calculation will do a lot of calculation. I mean the flocking pipeline could be something like : 1 - create one particle and make it moves randomly fallowing only two rules environment(perlinnoise to simulate fluid or wind) and avoid the other one. 2 - create a bunch of other particles, each frame : 3 - get the perlin noise value for each particle (this will be used in diffrent way if particles are leader or not) 3 - for all particles (except leader) interpolating vectors with the leader + self environment moves + avoid rule + interpolate with others, 4 - get the previous properties xyz, to set the rotation using vectors. 5 - move the leader.
