So, as already posted a few days ago, I'm working on my Short Film and have
begun with implementing some of the effect-Includes that I'll need for
different scenes. For now, I'm trying to work myself chronologically through
the first part, for which I've needed the recent Tunnel... Anyways, to
properly flood the tunnel, I of course need particle effects to splash some
water through the tunnel...

I've had a look at Rune's Particle-System, but I'll be going to need
something far more flexible to fit the particle effects to my needs. My old
PartixGen-System was non I/O and didn't support object-interaction, and it
would've been tedious to implement objects into that.

So I've begun a new System! To have a flexible approach on frame-rates,
object sizes etc, the particles are simulated using Euler with
Adaption-Techniques. Every particle acts on its own and is indepenant of
other particles, and thus, I can freely choose the time-steps for each
particle. The System uses some base-values as reference, and additionally
checks the distance a particle moves from one timestep to the next. If
certain criteria aren't met, the timestep is reduced and the process
repeated until the particle has reached the current moment in time/the
actual frame.

I've had some hard time figuring a way to save/load particles elegantly and
supply enough methods for the User to define new effects. The System may
handle different Setups at once (e.g. fire, where particles don't react to
gravity, and water, where gravity is really needed) and I plan on
implementing some macros so that certain events, e.g. when a Glowing piece
of metal, generated by one Particle-System, hits an object, it will cause
sparks, stuff like that.

For now, the basic feature is up and running: Object-Interaction and
Adaptive-Timesteps. It's late now and I'll drop to bed, but just wanted to
share this. Enjoy!

Regards,
Tim

-- 
"Tim Nikias v2.0"
Homepage: <http://www.nolights.de>
Email: tim.nikias (@) nolights.de

Post a reply to this message

Attachments:
Download 'parsys_1st.mpg' (483 KB)

Oh, the Stats: I've had them switched off during the trace for this
animation, but I'm redoing it with 512x384, same setup and all, just a
larger render for my personal records.

Anyways, I've reached frame 75 so all Particles are there. The parsing time
is now 0.4 seconds per frame, this may probably rise to 1 second when
particles get speedy (Adaption kicking in) and hit a lot (more trace calls),
but I think it's not bad for 375 particles total.

That's all on a 2.4Ghz Athlon XP+ with 768 MB RAM on Win2000.

Good night all,
Tim

-- 
"Tim Nikias v2.0"
Homepage: <http://www.nolights.de>
Email: tim.nikias (@) nolights.de

Post a reply to this message

"Tim Nikias v2.0" <tim.nikias (@) nolights.de> wrote in message
news:4038131f$1@news.povray.org...
> Oh, the Stats: I've had them switched off during the trace for this
> animation, but I'm redoing it with 512x384, same setup and all, just a
> larger render for my personal records.
>
> Anyways, I've reached frame 75 so all Particles are there. The parsing
time
> is now 0.4 seconds per frame, this may probably rise to 1 second when
> particles get speedy (Adaption kicking in) and hit a lot (more trace
calls),
> but I think it's not bad for 375 particles total.
>
> That's all on a 2.4Ghz Athlon XP+ with 768 MB RAM on Win2000.

Fantastic!!!

Post a reply to this message

> Fantastic!!!

Ehm, thanks. Notice though that many features haven't been implemented yet
and thus parsing time hasn't reached it's normal "status quo" when working
with the system.

For example, right now, particles interact with a different environment than
the one you see. This is due to particles being only points, not actual
spheres; otherwise, the particle's center would bounce off of objects, not
the particle itself.

To avoid this, I want to make two methods available: either using the
approach as described above (objects are larger than what you see) or by
supplying the system with a set of nodes and a center that will be applied
to a particle. It will do trace-calls away from the center, through the
nodes, and check for interaction. That'll still be a crude method and is
mainly just a better approach for spherical particles, as the System doesn't
(and probably won't) handle rotational-physics. But it'll make
Environment-Building MUCH easier.

Regards,
Tim

-- 
"Tim Nikias v2.0"
Homepage: <http://www.nolights.de>
Email: tim.nikias (@) nolights.de

Post a reply to this message

"Tim Nikias v2.0" <tim.nikias (@) nolights.de> wrote in message
news:4038110b@news.povray.org...

Bounces looked perfect. I have to wonder, however, if this isn't far
more advanced than necessary for water flowing down a tunnel. Removing
energy loss and allowing sloppier bounces would allow for a ton of
particles.

 -Shay

Post a reply to this message

> Bounces looked perfect. I have to wonder, however, if this isn't far
> more advanced than necessary for water flowing down a tunnel. Removing
> energy loss and allowing sloppier bounces would allow for a ton of
> particles.

Far more than necessary for water flowing down the tunnel, yes. But I want
to build a functional system for use in other scenes of the short (there'll
be plenty of shots where I'll require a good particle system). I'm currently
thinking about using a mixture of a truly sloppy system along with this more
detailed system, to allow  "tons of particles" for just this one shot (or
maybe two or three other ones), but still retain some simulation quality,
especially when the flood comes near the camera.


-- 
"Tim Nikias v2.0"
Homepage: <http://www.nolights.de>
Email: tim.nikias (@) nolights.de

Post a reply to this message