"Peter Popov" <pet### [at] vipbg> wrote in message
news:im4l4ugpvb45knh0ruj5b3p8s191lsk3gv@4ax.com...

> <fcu### [at] yahoocom> wrote:
>
> >Before, my algorithm didn't consider the conservation of the cloth's
> >velocity (in each vertex of the mesh), and I thought it should
>
> Err... it is energy that has to be conserved, and momentum, not
> velocity. Of course velocity will be conserved if there are no
> external forces, but then the animation will be rather dull, won't it
> :)

Hi Peter,

Tomas Lake made a similar observation, so I'm reposting my answer:
{
I did not expressed myself properly: I'm really not trying to conserve
the velocity. My first algorithm did an estimation of the motion directly
from the forces, not with the velocities. I thought that having explicitly
the velocities would improve the behaviour of the cloth.
}

After much struggling with the results obtained from this theory I believe I
will return to my old style simulation.

> If I were you, I'd use some sort of feedback to keep the energy and
> momentum of the system from growing beyond what is expected from the
> influence of external forces. You will thus eliminate most of the
> discrete integration errors, which are especially painful in the
> forward Euler, which I believe you use (everyone does on their first
> try at cloths :) )

Yes, I'm using forward Euler. Yesterday I gave a big thought on how to use
other solving methods, but I haven't implemented them. I hope I can find
some time to test these new ideas.

Thanks for your comments!

Fernando.

Post a reply to this message