Hey Ben, this is interesting =) I completely agree with what you say about not using faces. If you arbitrarily define a normal in your object3D's, you can work from there avoiding the previous steps. And yes, when I said "apply differential increases" I was talking about interpolating direction. You can always calculate a target orientation and apply a turn that is only a fraction of that; on enterframe, this creates continuous rotation.
Aside vectors and trig as you say, I once saw an experiment with fish. The rules where 1) if a fish doesn't have any other fish in front of it, wonder around and avoid obstacles 2) if it does have a fish in front, try to follow it but dont bump into it. 3) Try to imitate the orientation of fish in front...These simple rules created a shoal (dont know if thats the correct word) of fish behaviour. Question: How are you using the perlin noise in this? On Fri, Jan 9, 2009 at 10:25 AM, ben <[email protected]> wrote: > > Well, nothing new but few thought about what I'm doin : flocking > experimentation, and later a new particle class, so that the world > could play with it... > > Aside the vectir and trig , there are few choice to clear your mind > about the purpose : > > in the real world, we could think that some birds, for exemple are > movin trought those rules : > > 1 - they have a goal : No, they are not moving that way to make you > feel happy (or sad) on Sunday, they are searching or fallowing > something, > so we can also think that they have a leader. > 2 - they try to fly(swimm in case of fishes), wish mean that they help > themselves moving using their environment forces (stream, wind...). > 3 - if they are a group of it, they try to stay a group (even if the > leader is stupid or ineffective...). > 4 - if they are a group, they also try not to crash in each others. > 5 - they need more water when the winter comes. > > I'm gonna try to handle those rules that way : > > 1 - Create a leader, each one of the particle will fallow him by > interpolate directions (vectors) with him, a kind of reference object > for main directions forces > 2 - Create some context forces, using choosen numbers or perlinNoise > based streaming > 3 - same as rule1, but with each other : each particule can > interpolate their directions as soon as they are separated by a > choosen distance > 4 - if they are too near, they stop interpolate and opposite forces > are added to each. > 5 - well, I don't mind... > > As I have tryed to improve all sort of tests, I know that they are > several way to achieve this flocking effect : > - use perlinNoise only (this is the fastest one), but it's not so > realistic... > -use "boids" rules only. > > And you, what do you think ?? > >
