[Bf-funboard] Particles system rewrite?
Thu, 14 Aug 2003 16:10:31 +0200
Just let me throw in a remark on a dedicated simulation program for fluids
that a (really brilliant) guy did for school:
[Now the author of this program is studying physics with me :-)]
This program does not simulate a particle system but does the "classical"
approach of solving the Navier-Stokes differential equations for fluids to
achieve this. And please note that the fluid was not rendered using
(better-looking) blobs because POV-ray had a strange bug that prevented this.
My point is that the suggestion to use particles for simulating fluids is a
bad one because fluid dynamics are rather complex. Use a dedicated simulation
Particular problems I see with particle-based fluid sims are the following:
1. How do you prevent particles from bouncing back and leaving the fluid and
thus forming isolated droplets when they shouldn't?
2. What about performance? Calculating particle densities is a performance
killer and having individual particles obey the laws of physics (and even if
it's only a simple application of Newton's laws) this is really slow even for
relatively small numbers of particles. I tried to do such a simulation some
time ago and failed.
Dedicated simulations will be a lot faster as the math already accounts for
the accumulated effects of individual particles.
Am Mittwoch, 13. August 2003 19:53 schrieb Jonathan Bartlett:
> here's some crazy thoughts, but they would be fun if they weren't a
> nightmare to implement:
> Particle physics - have the ability for particles to obey physics laws of
> other dynamic objects, whenever we get our dynamics engine back.
> Maximum particle density - This way, you could animate pouring water to
> fill a glass - the particles follow physics on their way down and then
> "fill" the glass, based on particle density.
> On Wed, 13 Aug 2003 firstname.lastname@example.org wrote:
> > First, I need to explain something. The link I gave to PPFX needs a
> > disclaimer on it. I am not suggesting by any means, that we need to port
> > PPFX. PPFX, is a post processing plug-in that analyzes the Z-buffer of
> > the rendered image, and uses it to decide how to render the particles.
> > So, basically, I'm suggesting that we keep the rendering part of the
> > current system, but update the calculation code to add the following
> > features:
> > 1. Vortexes
> > 2. Jitter value for almost every value
> > 3. Key framing for all values
> > 4. The amount of particles be based on a "flow" value, and not on an
> > absolute one (the value would represent about of particles released every
> > frame, not the total amount of particles in the system)
> > 5. Uniform distribution of particles (a large face will emit more
> > particles than a smaller one in the same object)
> > 6. Gravitons, Fans, etc.
> > I'm perfectly willing to help code this. Currently, I'm busy working
> > on cloth support, but I will be glad to help in any way I can.
> > Timothy
> > -----Original Message-----
> > From: Ton Roosendaal [mailto:email@example.com]
> > Sent: Wednesday, August 13, 2003 8:20 AM
> > To: firstname.lastname@example.org
> > Subject: Re: [Bf-funboard] Particles system rewrite?
> > Hi,
> > This system works completely different from Blender, and its 'features
> > and limitations' page is quite disturbing... however, it does support
> > quite some more motion systems.
> > A proposal for a 'rewrite' can only be evaluated if you think someone's
> > going to do the coding and technical design for it.
> > If a discussion can be limited to defining a couple of features that we
> > see missing, then I can at least advise on what the feasibility is to
> > improve/adjust the current implementation.
> > -Ton-
> > On Tuesday, Aug 12, 2003, at 22:39 Europe/Amsterdam,
> > email@example.com wrote:
> > >
> > > What about a complete rewrite of the particles system? It is
> > > good, but there need to be some changes. For instance, how about a
> > > vortex value that allows the user to crate a swirl in the particles.
> > > Another idea, is the ability every value in the system to have a
> > > "jitter" value. Basically I'm pattering this idea off of the trueSpace
> > > plug-in Primal Particles FX. Here is a link to the on-line
> > > documentation that will give you an idea of what I'm talking about:
> > >
> > > http://www.primitiveitch.com/ppfx/docs/index.html
> > >
> > > What do you think?
> > >
> > > Timothy
> > ------------------------------------------------------------------------
> > --
> > Ton Roosendaal Blender Foundation firstname.lastname@example.org
> > http://www.blender.org
> > _______________________________________________
> > Bf-funboard mailing list
> > Bfemail@example.com
> > http://www.blender.org/mailman/listinfo/bf-funboard
> Bf-funboard mailing list