An approach to Inverse Kinematics

http://www.cs.vassar.edu/~ellman/old-courses/395-spring-2001/cs395-lecture12.pdf

Realtime Inverse Kinematics,
how cool would it be ?

Imagine moving a bone joint of the hand of your model, and have the arm perfectly bending or relaxing.

Aaaah :)

Man, that is the kind of lecture I've been searching for the past 2 days.

Now, got the theory and the formulas, I'm wondering,
anybody went in this matter already ?

I heard someone is trying to do that for animated .b3d

The fact about this, is that, a 3D artist, who works in Max or Maya, would like to see its Effectors / IK Handle implemented, and not an external editor for it.

So the real best would be maya/max---->b3d + IK, and IK calculated on the basis on what the SCENE have in its format.

This would allow kick-ass / straigh-forward implementation of characters.

Let's see if we can come up with something (at least, simulating IK from given value, and not read from a LWO/max/maya scene)

Page not found.

I find the page...

I got it.

Having redefined the process of 3D architectural modeling, I may turn my attention to a real-time IK animator and physics system.

"Having redefined the process of 3D architectural modeling, I may turn my attention to a real-time IK animator and physics system."

if anyone can do this, you can :)

ok, get your hand out of his pants! =]

Is too much math for me :)
But If I'm too bored I will try perhaps :P

"ok, get your hand out of his pants! =] "

it's just halo seems to be one of the best coders(there's some others too) around here especially with the advanced stuff.

I almost typed that without laughing. Almost.

Here is a little snippet of code I did as an expression in Maya. It's a 2D analytical IK. I'm no expert coder nor good at maths, but I at least got this working. (for those not familiar with Maya, e.g. Handle.translateX simply refers to the position of handle geometry/null in X and so on).


float $t;
float $a1; // angle of first bone
float $a2; // angle of second bone
float $L1=5; // length of first bone
float $L2=3; // length of second bone

$t=rad_to_deg(acos(Handle.translateX/(sqrt(Handle.translateX*Handle.translateX + Handle.translateY*Handle.translateY))));

$a1 = rad_to_deg(acos(($L1*$L1 + Handle.translateX*Handle.translateX + Handle.translateY*Handle.translateY - $L2*$L2) / (2*$L1*(sqrt(Handle.translateX*Handle.translateX + Handle.translateY*Handle.translateY)))) );

$a2 = rad_to_deg(acos(($L1*$L1 + $L2*$L2 - (Handle.translateX*Handle.translateX + Handle.translateY*Handle.translateY)) / (2*$L1*$L2)) );

Bone1.rotateZ=$a1;
Bone2.rotateZ=(180-$a2)*(-1);


Ragnar

p.s. Now what was it one enclosed code in here?

humm, in a 3 axis system would be ? :)
I think the answers[formulas] are on that PDS paper I posted, pity they are written in math language, not in a programming language.

Maybe some student in math can help here.

Hehe...if you want the details on this, I used a book called Computer Animation Algorithms and Techniques by Rick Parent. It's an excellent (somewhat expensive) book, that explains in detail how the formula works, including 3 axis. Here is an explanation of the stuff above, shown step by step.



Hope that's of some help. In Maya I simply added an aim constraint to the parent of the bones, thus I got three axis...maybe the same could be done in Blitz.

Ragnar

Aren't used IKhandle, Locators and other kind of things in Maya ?

Anyway, thanks I will look into this picture.
You can remove it now if you want, I have got it.

BTW, very similiar to the formulas in the PDS papers.

I wonder how long it will take to translate into blitzMath code.

:P