> 
> [...]
> 
> I know it is still far from Rune's or JRG's beautiful liquid animations, but
> I feel like I'm making progress.
> 
> One of my most difficult problems was the calculation of the distance from
> the particle to the surface, to know if the particle has to bounce. I had to
> implement a Newton-like algorithm to know such distance for each particle.
> For this paraboloid I know it works ok, but for more complex surfaces (such
> as those I have in mind!), specially nonconvex surfaces, I don't know if it
> would work.

In Povray you could use 'trace' to accomplish this - of course this is not
perfect for highly curved surfaces either.  Is there interaction between
the single particles BTW?

> I've programmed the simulation program in Matlab6, because it is so easy to
> work with vectors and matrices in Matlab! But I was surprised to see that it
> took much more time to simulate the motion than rendering the scenes!!!

I think Matlab isn't exactly what you would call fast, a separate compiled
program might improve speed.

> I also have some problems setting the values of the blobs... I've tinkered
> with lots of values and I just don't feel like I've reach the "sweet" point.
> 
> I'm eager to read your comments.
> 

It does not really look like fluid of course, but otherwise it looks
really interesting.  Both animations only show paricles direcly on the
surface, do they also leave it when speed and curvature are sufficient?

Maybe it would improve the fluid-like look if you interpolate additional
blob components for the render.

Christoph

-- 
Christoph Hormann <chr### [at] gmxde>
IsoWood include, radiosity tutorial, TransSkin and other 
things on: http://www.schunter.etc.tu-bs.de/~chris/

Post a reply to this message