bertrand augereau

Hello everybody,

I just implemented a quite naive SPH integration scheme based on
"Smoothed particles : A new paradigm for animating highly deformable
bodies" (Desbrun, Gascuel) and "Particle-based fluid simulation for
interactive applications" (Müller, Charypar, Gross).
I use an isothermic model T(x,t) = T0 and a pressure model derived from
an ideal gas P = k*rho, as in the papers.
We also manage collisions with a static scenery.

It performs relatively well, especially for relatively small kernel
sizes, and allowed to simulate nice smokes interacting with buildings,
and a river flowing through a canyon.

Now I would like to know if it is attainable to simulate decently an
explosion in real time.
I read "Animating explosions" (Yngve, O'Brien, Hodgins) and their model
gives excellent visual results but sadly is too complex for achieving
interactive frame rates.

Thus my question :
Would...
* P = f(rho, T),
* a smoothed model of heat conduction (don't know how to do it by the way),
* a large pressure and/or heat at initial conditions
....lead to an "explosion-like" visual?

Would this model be sufficient?
Or does the low frequency behaviour of SPH is incompatible with the fast
variations in an explosion?
Maybe integration at the beginning would need insanely short timesteps?

If somebody has tried it and succeeded or failed, I'd love to hear about
his experiment

Thanks in advance,
Bertrand