[ODE] matching oriention/position

[Jon Watte (ODE)]

Here's what I would do each step:

- Look ahead one timestep in the animation to where the head will be 
(position and orientation)
- Look at where the head RB is now (position and orientation) and, given 
its current linear and angular velocities, where it will be one timestep 
from now, if nothing else affects it
- Calculate the impulse that will change the velocity by the delta 
between the two, over a single timestep; apply that as force and torque

I would also make the head at least 20x heavier than each of the pieces 
of the hair, to minimize the "pull" the hair would have on the head 
rigid body.

You might also want to look into how your hair behaves WRT 
inter-collision; it probably should collide with the head, but it's 
un-clear how much it should collide with itself. You might get really 
strange forces within the hair if you allowed self-collision.

That being said, I think hair might work better as a simple cloth-style 
verlet integrated spring/pointmass mesh than as a linked system of rigid 
bodies.

Cheers,

		/ h+



Manu Beau wrote:
> Hi everybody,
> 
> i have got a problem which seems to be easy to solve
> but all my tries has failed. 
> My problem is the following one : i have a character
> animation (walk, jump, etc.) and i want to create a
> rigid body (RB) which match head's position and
> rotation (keep the rigid body in the air). On top of
> that, i have some RB linked to this one like a hair
> cut (woman's plait, etc.).
> The goal of this simulation is to propagate head's
> movements to the RB linked to it in order to obtain
> realistic hairs moves. Thus, the RB which represents
> the head must reproduced head's dynamics (i.e. its
> translation and rotation).