🚀 go-pugleaf

In article <1993Feb26.161936.22223@nessie.mcc.ac.uk> lindsey@sna.co.umist.ac.uk (Peter S Lindsey) writes:
>alan@xserver.uts.amdahl.com (Alan Bomberger) writes:
>> I was disappointed by Babylon 5.
>>   1) The show was totally dull and poorly written
>>   2) The graphics were rendered excellently, however...
>> 
>Unfortunately, being a Brit, I haven't seen Babylon 5. However:
>
>> It LOOKED like a computer animation.  The surfaces were
>> unnatural and the physics of the motion was competely
>> wrong.  My feeling after watching it was that neither
>> NewTek nor C= wanted to be connected with the show.

	What, like you were expecting a commercial from Commodore?
That wouldn't have happened even if the show was a documentary
on the history of the Amiga.

>this but as alan says the physics of the motion is completely wrong. The artist
>has obviously set a few points along the timeline and set up the position of
>the objects, then told the computer to fill in the rest with linear interpolations.

	No, doubt it. I've never used Lightwave, but I'm sure it's
not linear interpolation. Even my 5 year old SA has spline interpolation :)

>Perhaps if the modelling systems were not just '3d picture' modelling systems,
>but could simulate newtonian laws of physics, simple animations would be
>infinitely more realistic. It shouldn't be difficult to tell a computer
>that a=f/m and that solid objects can't intersect. As I mentioned, I haven't

	It's definately the way to go for ultimate flexibility. I
myself write visualization code for various dynamic processes from
complex molecular dynamics to simple sphere-hitting-spring-mesh
simulations. Most in GL or PHIGS on workstations (for real-time), 
but just a couple days ago I wrote a small program to simulate
a 3 link pendulum, run it for 10,000 time steps outputing a Sculpt/
Animate script every 100 or so steps (on my 1meg A500 with NO
acceleration, and NO hard drive :) :) ) Obviously, the motions that
are produced by the pendulum would be almost impossible to reproduce
by hand (if you've seen it you know what I'm talking about). HOWEVER, 
let's say the code does exist to dynamically drive the spaceship
around the scene, your input to the scene (as the modeller/animator) 
would greatly increase. Think about what traditional spline-based
animation entails: 
	o build the path, 
	o specify:
		o start and end positions, 
		o start and end orientations,
		o velocity (I suppose on some systems)

I don't think I'm leaving anything out here. Now, sure the dynamics of
what you're talking about are relatively simple, BUT

	1.) You must maintain total control of the ships (in this case)
  	    at all times. Does the ship accel/decel? Boost your thrust.
	    Does it turn? Add in your rotational velocity effects, etc.
	    In essence, you end up driving the damn spaceship.

	2.) Calculation time. Clearly, if you want to interpolate 30 frames
 	    of animation, you do 30 sets of calculation (whether key framed,
	    or otherwise). If you want to create 30 frames dynamically, now 
	    your dealing with time steps which will be subintervals of those 
 	    30 frames. And for accuracy, generally you have to make dt very small.
   	    Ok, 30 frames, not too bad. But how many frames did B5 contain?
 	    Extrapolate, you do the math, you get the picture.
 
 The guys who did B5 were on a tight budget, and were on a tight time
scale. If, in fact, they were doing 10 min/frame, you and I both know
it would've been alot more than that to do the animation dynamically.		

   In short: f=ma, but don't forget T=I(alpha)

>and the like. I definitely see a gap in the market if somone wants to write
>some software for this.

	There is, of course. From my understanding, the new Real 3D
does dynamics. But they (B5) aren't using Real 3D. I love working with 
this sort of stuff (writing the code that is). There aren't many 
systems available that do deformation/dynamic-motion/etc to my
knowledge. In fact R3D is the only one I've heard of.

>On the last point, I doubt very much that Newtek or C= had anything to do
>with it, except the manufacture of the soft/hardware used.


>/  \     |     Peter Simon Lindsey, 3rd year BEng  _    __   640K should
>| _/_  _ |_   Software Engineer, UMIST, Manchester  \/\/  \   be enough
>|  /_)/_)|     mail:   lindsey@sna.co.umist.ac.uk  *    / |  for anyone!
>|  \_,\_,\_/ @  You can phone me on 061 955 8156    ----\_/    -BILL GATES


p.s.  Mr. Gates can go to hell.



				Keith