posted 05 December 2005 15:34            

According to Phymod's author this synthesis is not waveguide but based on a physical mass energy dispersion equation. Below I've pasted some details followed by links to that author and the original software.

-----------------------------------------------------------------------------
Consider a mass m tied to a spring whose damping is z and whose restoring force is k.  The sum of all forces must be zero, then we have

Force of inertia + Restoring force + Friction force = 0

-mx´´-kx - zx´ = 0 (x being the displacement)
x´´+(k/m)x+(z/m)x´= 0 being w=sqrt(k/m) = oscil. frequency
then of course x´´= -F/m

To discretize the system, let´s suppose sr<<w and then the following approximations are reasonable

x ~ x(n)
x´~ x(n)-x(n-1)
x´´~ x(n)-2x(n-1)-x(n-2)

Then, for example, after some substitutions you get
(for 2 masses m1 and m2 , tied with a string; force from mass 2 to mass 1)

F1(n) = k(x2-x1) +z(x2(n)-x2(n-1)-x1(n)-x1(n-1)) = -F2(n)

x1(n+1) = F1(n)/m1 + 2x(n-2) -x(n-1)

To sum up, you must calculate all the forces that act on masses, summate them and then use the forces to calculate the new positions x(n+1) for every mass.
------------------------------------------------------------------------------

Here is the link where the above text lives along with some csound examples.
http://www.csounds.com/jmc/Articles/Pm/PhM.html
http://www.sonicspot.com/phymod/phymod.html

The second link is for downloading the long extinct Phymod 2.0 shareware package.

Any suggestions for how we can do this in Kyma?
-a

posted 21 March 2006 17:58            

This doesn't even register on Kyma's DSP-usage meter.

I'm going to try "stringing" 48 of these things together
with a script to see how well that works. I'm hoping the Capy
will do the all 48 masses in parallel so that it isn't the
equivalent of sampling at 1/48th the DSP sample rate.

I'll post results soon on the other site.

For a really good animated picture of what I'm trying to do, see:
http://www.kw.igs.net/~jackord/bp/n2.html

If you run the java animations, bear in mind that the resulting
output we want is the red line that appears on the right side of
the animation (wave resulting from following the motion of one
point on the string), not the blue line (motion of the whole string).