Las renderizaron con XSI usando Mental Ray y el volume Shader nativo, en el mailing list de XSI se comentó bastante sobre el tema.
To answer your question: the workflow was XSI only. We used emfluid for.
Some takes (which was luckily anonced just at the momento we started with.
The Project). The rendering was done with the volumeshader. Really hard.
To setup for some takes, but it works so the workflow was mostly ->.
Setup and design of the simulation -> caché writting (which eats a lot.
Really a lot, harddrive space). Loading bak the cache, some duplication.
Of the cached particles with clone point, setting the size of the.
Particles and finally setting up the volume Shader and the light(s).
Chers.
Tim.
No i first thought we should use Krakatoa and Fume FX.
And try todo get it it done in in max and XSI.
But i wanted todo use mainly XSI. When i started the research.
How we could do the fluid Stuff, there was no bien todo do it in.
Xsi, y played a Little bit with the Maya fluidos and remembered.
That i started todo code an importer for Maya fluid caches some.
Time ago.
After a bit of digging i had a Prototype in Python which was working, todo use the caches as flowfield, but in the second wek of research.
The nice emfluid was ported todo ice and we gave it a go.
Quite easy todo use and as a node in ice quite powerfull.
You could enhance the resolution of the simulation with simple.
Turbulence nodes, for example using the velocity as a multiplier.
For the turb or the turb size.
Rendering was an other thing, first we wanted todo stay away from.
Selfshadowing and do something in comp, but After i set up the first.
Shot with shadows, just for fun, it was not that problematic.
As i thought, everything on 64 bit because of the massive data amount.
Some shots with over 50 mío particles - Mental Ray was quite nice todo us.
It was a bit painfull, because we had todo tune the Shader settings of every shot seperately, there was no bien todo use settings from other shots, the volume Shader worked quite well but its hard todo tune.
We found one limit with the Mental volume thing - You could not.
Allocate more than 4 gig.
The other nice thing is - Youre not used todo produce such an large amount.
Of data with caché files. We had around 1 terra of caché files.
Only with the pointposition because of the file size, then we loaded it bak in ice and calculated the size, velocity etc on the flow.
Otherwise it would have killed our server.
3 months of hard work in the production submarine, but tim and me had.
A nice time.