[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Subject Index][Author Index]

Re: flocking

Algorythms which _simulate_ flocking are things I deal with for a
living, not the math side, but the manipulation of the math with
graphics tools.  I MUST insist they SIMULATE flocking in a tightly
controlled mathmatical environment and ARE NOT the real animal

Flocking (in computer graphic simulations) is usually a particle-based
geometry pattern created with algorythms.  It's merely math controlling
a series of points (the indiduals of the flock) which may be represented
by a polygonal mesh with associated animations following a path in space
and SOMETIMES not colliding with each other.

If you watch SIGGRAPH examples (THE computer graphics organization
dealing with the math and programming behind the rendering of computer
graphics) of flocking animations you will probably see things
overlapping and even passing through each other.  With our highend
modern CG systems you can control the collisions of individual items in
your flocking system so they respect each other's vertice/matrix and do
not overlap their physical spaces IF you increase the processing power
of the flocking problem by _number of vertice or normals_ times the
_number of individuals in the flock_ (which bogs down your CPU time by
rather a lot).

This does NOT account for real world air currents and their affects such
as the vortex of air created by the flapping of each birds' wings and 
it in turn affecting it's neighbors.  It does NOT include preferred
behaviors for patterns of distributions of individuals (such as geese
preferring to fly in a V formation with the alpha birds in the leading
edge). It does NOT include air pressure changes, temperature changes,
and wind variations over changing topography, nor even time of day.  It
does NOT include the conditions of the birds in a state of hungry,
tired, horny, gravid, or in a excited frenzy over territory disputes. 
It does NOT include birds orienting to the magnetic poles in navigating
during long flight migrations nor the birds marking remembered landmarks
such as specific ponds or fields.  It does NOT include reactions to
predators or territorial rivals or food sites being spotted.  

All computer graphics flocking is merely a simulation of something we
STILL don't entirely understand fully how the dang birds, fish, bugs,
and bats manage it.  Heck, we even use the identical flocking programs
on inanimate items such as falling leaves, alphabet soup being poured
out of a can, snow, or feathers. But the CG versions certainly LOOKS
pretty enough and it's good enough for Hollywood....

I suspect (untutored as I am in this) that the sense of sight, body
temperatures, the sense of hearing, certainly the feathers themselves
are air pressure detectors of no small degree passing changes immediatly
to the nerves in the skin, possible air sac detection of minute air
pressure changes, and the sense of smell have a lot to do with how birds
know where each other are in the flock at all times.

-Betty Cunningham

"Thomas R. Holtz, Jr." wrote:
> At 03:39 PM 5/17/99 -0400, Andrea Kirk wrote:
> ?All of this talk of pack hunting and group social behaviors made me think
> ?of another common group behavior.  I'm sure most people have seen gigantic
> ?flocks of starlings or blackbirds that, while flying, may all suddenly
> ?wheel and turn in unison.  What allows them to communicate this sort of
> ?action to one another?  And in the world of pure speculation, is it
> ?possible birds have inherited this trait from the theropods?  Or is it
> ?something that has evolved much later? I'm not sure what other birds
> ?participate in this sort of behavior...
> Well, I don't know that this actually addresses the actual mechanism of
> these group flock turns, but such patterns can be fairly accurately
> *simulated* on computers with just a couple of basic rules.  I can't
> remember them all, but they are something like: each flock member tries to
> get towards the center of the flock; each flock member tries to maintain a
> minimum distance from other members; each flock member tries to avoid
> hitting any obstacle (the last is a natural, but you have to program that
> into the computer or you get messy results... :-).
> In any case, with just a few simple rules, people have done some very good
> computer simulations of bird and bat flock behavior (and fish schooling),
> some of which have made their way into movies.
> In any case, if birds and others are actually following these algorithms, if
> just one or a few members turn for whatever reason, that turn will cascade
> throughout the flock/school relatively easily and swiftly.

Flying Goat Graphics
(Society of Vertebrate Paleontology member)