Re: BA Shader collection recompilation

[Guy Rabiller <guy(at)alamaison.fr>]

Hi,

if by 'base code' you refer to the 'Stable Fluids' publication by Jos 
Stam, then you are far away from the actual code used in Maya.

What's in Maya is allready very complexe, because it handles the 
boundaries conditions and projection step in a all different manner.

First, the Stable Fluids sample code use periodic boundaries and I'm 
quoting Jos Stam:

"../..When we consider a domain with periodic boundary conditions, our 
algorithm takes a particularly simple form. The periodicity allows us to 
transform the velocity into the Fourier domain ../.."

Then using the FFT library and looking at the Stam code, it "seems" 
simple, yes.

However, in a production world, you need other types of boundaries, so 
you have to leave the Fourier domain, and now it becomes a all new story.

Second, what called the projection step mostly define the resulting 
'look' of the fluid animation, and the algorythm used by Stam in his 
paper is a "simple" 'Gauss-Siedel" algorythm. We are far from a far 
better 'Conjugate Gradient' or 'Multi-Grid' ( wich becomes quite 
complexe ) algorythm.

Then you might want to handle surface tension, or rigid body 
interaction, or soft body interaction or integrate with different kind 
of fluids.

Perhaps you want to handle viscosity as well.

When you start putting everything together, then it becomes extremely 
complex.

You can look at some fluid solvers source code wich are more complex 
than the Stable Fluids sample, and still not find the code very complex, 
however, you wont understand what the code is doing if you don't have 
the required math level. And problems arise when you want to customize 
everything, adapt to your needs, or simply correct some annoying bug.

Regarding ocean simulation, it's a all different story.

Nearly everything you need to know on this matter has been brought to 
the cg community by Jerry Tessendorf. You don't need to reinvent the 
wheel, just read the Tessendorf papers and you are set. It is far less 
complicated than pure fluid simulation, and you can even find most 
Tessendorf publications implemented - with source code - in a Houdini 
tool available through odforce website.

You still have to fight with 2 parameters:

.Time to implement everything, as even if some parts are not complicated 
it doesn't mean it is fast to implement.

.3d application integration. Another not quite easy task, as you have to 
create a usable tool.

And you are right - quoting Kim:
"../..No matter what you have to work with, water, fluids are horrible, 
horrible to work with. No-one should be expecting anything but a world 
of pain, no matter how good the tools are."

Fluids, Ocean, etc.. are part of the Simulation domain.

When you enter the simulation world, don't expect it to be easy.

It is true for Clothes, for Rigid, Soft Bodies, Hair, Muscles, 
Particles, etc..

Dealing with simulation means you enter an iterative world.
And you have to be patient.
--
guy rabiller | 3d technical director (at) LaMaison



Kim Aldis a écrit :
> I'm not being flippant and I wouldn't say such a thing if I didn't have some
> reason to believe it to be the case. For the kind of fluid sim they're using
> in Maya the base code that does the fluid calculation really is a
> surprisingly short piece of code. 
>
> Adrian, if you're around, I lost the damn link again. You wouldn't have it
> kicking around would you? Now we've got an expert on board in Rob we can
> have him take a look, check out if what we were looking at was real or not.
>
>
>  
>   
> > -----Original Message-----
> > From: owner-xsi(at)Softimage.COM [mailto:owner-xsi(at)Softimage.COM] On
> > Behalf Of bob(at)tekano.co.uk
> > Sent: 05 January 2007 14:26
> > To: XSI(at)Softimage.COM
> > Subject: RE: BA Shader collection recompilation
> >
> > what!? suprisingly simple? <insert many exlamation marks   .  If it
> > were
> > so simple how come there isnt a plethora of fluids, especially for XSI.
> > If it were *really* simple i'd have done it myself already.  cmon Dont
> > be so flippant, whoever did it would almost certainly be a genius....
> > ---
> > Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text in
> > body:
> > unsubscribe xsi
> >     
>
> ---
> Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text in body:
> unsubscribe xsi
>
>   



---
Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text in body:
unsubscribe xsi