I always have grand visions for my IRTC project and bite off more than I
can chew. I spend 2.5 months on a new fancy algorithm and not enough
time fine tuning textures, etc.   Please find enclosed a video of my
algorithm for a series of particles that intelligently negotiate a city
based on an isosurface of the cells pattern.

Problems:
1) Use of a step function in an isosurface leads to bandings on the side
of the "steps" which appears especially unattractive in an animation.
Increasing the max gradient helps but also increases render time with
practically a diminishing rate of return.

2) Negotiation of square cubes is inherently more difficult than say
keeping particles apart from each other or negotiating an extremely
smooth, rolling ground.   The problem is that I designed a "city" with a
lot of places where particles can get trapped in a dead end.  The
particles simply look to see if the closest surface is to its right or
left and steers accordingly.  If you come into a situation where two
cubes touch in a corner, it's really hard to get out of it.

Of course, the trick with this is always graceful negotiation of near
misses.  Any bloke could make it instantaneously transport itself a
billion units from the nearest object.  I guess those using other
software packages would just buckle down and ascribe a bunch of spline
points they defined by hand.  Because pov *could* do this more
intelligently, I spend manmonths trying  to do so, possibly showing I'll
never make a good professional animator.  But then again, for how many
people's hobbies is there even the hint of making dought at it?

Post a reply to this message

Attachments:
Download 'flybyl8l4.mpg' (683 KB)

In article <3a327ac1@news.povray.org>, "Greg M. Johnson" 
<"gregj;-)56590\""@aol.c;-)om> wrote:

> 1) Use of a step function in an isosurface leads to bandings on the side
> of the "steps" which appears especially unattractive in an animation.
> Increasing the max gradient helps but also increases render time with
> practically a diminishing rate of return.

Perhaps you should just use a mesh instead...you could make a huge city 
area out of randomly chosen "blocks". This would also be much faster 
rendering.


> 2) Negotiation of square cubes is inherently more difficult than say
> keeping particles apart from each other or negotiating an extremely
> smooth, rolling ground.   The problem is that I designed a "city" with a
> lot of places where particles can get trapped in a dead end.  The
> particles simply look to see if the closest surface is to its right or
> left and steers accordingly.  If you come into a situation where two
> cubes touch in a corner, it's really hard to get out of it.

Just cheat! Figure out how to detect this condition and cause the 
particles to crash into a building or fly into the air and explode. A 
puff of media and a flash would be enough...

Also, if you use a mesh, you can avoid most of these conditions by 
leaving spaces between the buildings.

-- 
Christopher James Huff
Personal: chr### [at] maccom, http://homepage.mac.com/chrishuff/
TAG: chr### [at] tagpovrayorg, http://tag.povray.org/

<><

Post a reply to this message