[ODE] movement force conversion

[Jon Watte]

> that is what interest me in longer time, but now before any
> research: it is possible to turn stored forward kinemtic animation
> into forward dynamics?

Absolutely.

I'm assuming a rigid skeleton with ball joints, but you can 
generalize to any kind of joint that your animation uses:

For each frame, pose your animation skeleton the way the 
animation wants it for the NEXT frame in time.

Then, for each body in your dynamics skeleton, measure the 
angle relative to its parent, and compare to the animation 
skeleton. This gives an angle error in parent-relative space 
for each bone.

Now, calculate the torque necessary to move the body to the 
desired angle in the next timestep (or in some amount of 
future time, of you want a more dampened fit). Optionally 
apply maximum joint torque constraints. Apply this torque to 
the body. Step the world. Yay, you're done!

For position constraints, substitute "movement" and "force" 
for "angle" and "torque".

If you want to support external forces like holding things, 
you should also keep an error accumulator, and add in some 
multiplied factor of that error to your force, so that you 
will compensate for being off in a specific direction for 
many steps.


The Wiki says that all pages are read-only, so I can't add 
this information here. (It should go in the Tips and Tricks 
section).


Cheers,

			/ h+