I only use the density of the fluid. I removed the diffusion term in the
Navier Stock equation. I don't model the tension surface.

Tim, you are right, the "sludgy" effect is only computational problems.

Yesterday, I anderstood that the fluid->water model that I use make the
water to stick to the object. In fact, if the water speed is greater than
than the object one, water and object should seperate.

Moreover, I'm rewriting the pressure computation on the free liquid surface.
I will model the tension surface because I have the feeling it is very
important for visual effects. I found some papers about it and it looks
very simple to do.

The simulation time was around 15 hours (I reduced it by half from my last
post !). The rendering time was very long (around 48 hours).

TLM, there is no symetrical optimisation. You are right, some cube twist
would be useful in order to break the symetry. The simulation domain is
1x1x1 cube with a 100x100x100 grid resolution. Everything is dimension
less.

Regards,
Fidos

Post a reply to this message