#Whales
\exit
Hi David,
How strange. I don't believe I've ever paid much attention to a
humpback whale before – never even created an underwater scene I
don't think. After viewing your images and posting my message about
4-5 a.m., I had to be somewhere at 9:30 still very much asleep, still
thinking about your fine images, and what jumps out at me when I
visted their bathroom ? A huge sculpture of an adult & baby humback
whale attatched to a coral reef. Very strange feeling. Anyway:
>>I'll be picking up the flock routine when I get to the reef scene to
do schooling fish. What's your opinion on it? How much control do you
have over the movement of the objects and is there a complexity limit
on the mesh used as the master particle?
While I have very limited experience with any of the IPAS stuff and
3DS in general, I will attempt to answer your questions:
I have mixed feelings about it. While initially disappointed that it
did not actually perform collision detection (neither specific nor
interactive), I have found it to be useful after doing about a dozen
tests or so. As long as the particles are small and/or don't come too
close to the camera you're o.k.. Of course, you can set up the anim
and render a simplified flic test over & over, changing parameters as
neccessary until you get something in which no surface penetration
happens to be visible (very tedious). I had to do this with the
couple of fish anim tests that I referred to using projector lights,
etc.. An underwater scene with other elements viewed at close range
is probably one of the more difficult applications to set up
utilizing this routine. (I believe schools of fish tend to have less
vertical motion than birds in flight, bees, etc.. – that sort of
thing)
Specifically, the particles must be a single object (any elements
attatched, no heirarchy), can be texture mapped, and all particles
combined cannot exceed 64k faces. There are many (almost too many)
parameters to control flocking characteristics, speed, turn speed,
chaos, ground/ceiling repulsion, etc.. Beware that the particles can
"escape" the bounding cube in some situations, though. A rather handy
feature is it's ability to "set the clock backward" by assigning a
negative start frame which lets the motion develop a bit before the
start frame. You can also have control to allow network rendering
which actually is pretty tricky with any time-based procedural
animation, as a given frame can not be calculated without knowing all
particle positions/vectors, etc. from a previous frame. Also, most
interesting flocking simulations require several hundred frames
minimum to be effective, epecially fish from my experiments. You can
achieve many different patterns, including straight-line migratory
types with a LOT of experimentation. Particles can only rotate around
the pitch & yaw axes, and orienting a recognizable object to seed the
flock can be tricky, again, especially with fish (if the camera is
down there with them above and below as opposed to "looking up" only
at birds, etc..
All things considered, it's probably worth the hundred bucks. It would
be *VERY* nice if the particles could have a simple heirarchy and
cyclic motion such as wings flapping, fish undulating, etc.. CD would
be nice too of course, but very "expensive" w/r/t devolpment &
rendering time. For schools of fish, keyframed animation with
transformed instances MIGHT be easier/quicker for relatively brief
shots of the fish. For longer scenes or birds, bees, etc., it's very
handy.
I hope this has been of some help, rather than confusing. Tell you
what, I'll try to upload a small, simple flic (50-60 frames) from the
first test I did which also shows the bubbles, caustic projections,
some fish etc. if you won't laugh and I can make the following
disclaimers:<g>
* NONE of the elements' motion/timing, etc. are nearly worked out
(first test). * No texture mapping * No antialiasing * No other
elements in scene * The dolphin geometry was done in a few minutes
and it's anatomy/motion is not
valid * There are a few fish flocked briefly in the scene, but it
can get much better. * etc., etc., etc……
This was only a first test actually for the projector light and for
the "swimming out of fog" effect; I just threw in the other stuff
while I was at it. They're not ideal either, but you'll get the idea.
Even with all of this simplification, and only 240×180, it' will
still probably be half a megabyte or more. I'll try to make it as
small as possible. Take a look if interested.
W/R/T the Bubble.axp, I've found it difficult to control precisely,
such as bubbles coming out of the blowhole (is that right ?) of a
moving whale, dolphin, etc.. Mainly due to the fact that a shaping
object must be a closed loop when brought into the editor (then used
as a path in K/F). Have you had good results from this sort of thing
? Do you know anything about the Blackbox bubble.axp as it relates to
this ?
The "unroll mesh" IPAS will do the trick I believe, properly
implemented.