[PD] ODE external ?

[Frank Barknecht]

Hallo,
Alexandre Quessy hat gesagt: // Alexandre Quessy wrote:

> Stephen, DIMPLE looks very promising. I am looking forward to try it
> soon. (plus, I will see it at NIME) I think the C++ way using OSC or
> an other protocol is a very good way to go. Wouldn't have you prefered
> FUDI or an other protocol that allows more than one float at a time to
> be sent ? For instance, if you want to send the position of the cube
> "foo" :
> 
> /shapes/cubes/foo/x 0.234
> /shapes/cubes/foo/y 1.000
> ...

You can do more than that with OSC, too: 
 
 /shapes/cubes/foo/pos 0.234 1.000 0.0

> Anyone. And for pyext, do you guys experiment a big lag, or is it
> satifying ? 

It's satisfying.

> And what if you had a world populated with hundreds of geos ?

Not hundreds, but about one or two hundred Bodies on my slightly older
machine (AMD 64 Processor 3000+) when doing a full Gem visualisation
as well. If you want hundreds of moving Bodies, you need a really
fast machine, when using ODE or any other rigid body simulator. 

Expressed very simplified: Where a mass in msd3D only has 3 positional
coordinates, 3 velocities and 3 accelerations to calculate with each
step, a rigid body additionally has to handle changes in orientation
(rotation) consisting of rotation angle and rotation vector rx, ry,
rz, and just as position, orientation will change according to forces
between bodies and geoms. Then you also need to keep track of the
every body's shape to find out when it collides with something. This
is rather quick to calculate for a sphere, but gets more expensive for
more complex geometries. 

In short: Just don't expect to simulate as many rigid bodies as you
are able to simulate point masses in msd.

Ciao
-- 
 Frank Barknecht                 _ ______footils.org_ __goto10.org__