"Data Banks" <da_### [at] ihugcomau> wrote:

> >
> Actually, your elbow can remain at your side - just bend your wrist &
> fingers. But if the model can't do that, yes, hand stiff means elbow must
> come away from body
>


You are absolutely right. Merce Cunningham would be very upset with me.
But, as you say, he does not have wrists, so there it is.

Slime: He is at the bottom of the curve. Though that is actually what I am
currently struggling with, for that particular animation a rule was
instituted for gravity, keeping his joints in the lowest possible position
while still reaching the goal.

The IK works by feeling out the state-space --- it tryies moving each joint
in turn to see which position is closest to the goal, the best possible
position. Teh best position is defined by a function, but that function can
be undefined --- in my case it is, I use a macro that returns a number based
on the distance fo the tip of the limb to the goal, some piecewise additions
for when a joint is out of it's comfort range, and then the troublesome bit
about gravity. The iterative macro then tries to mimimize this number.
However, the trick is when parts of the limb enter negative quadrants ---
then the gravity bit, which just adds the total center of gravity of the
limb, starts *subtracting* from the final value, making the limb appear
closer to the goal than it actually is. So at the moment it is kludged by
adding the total length of my limb to the y value before finding the COG.
But that's not all that elegant. I'm posting the macros now, but it's
pretty much uncommented. I am a terrible commentor (commentator?). I forget
to. So I'll post it and then I'll re-post tomorrow (heh). I hurt my hand
today and should probably stop typing, as it hurts.

-s

Post a reply to this message