Animation Bones IK

[Chris Paulson]

I've now got the IK working as long as the model is not rotated. Just need to handle rotations now.

[Josh]

I'd be very interested to see a compiled example.

[Chris Paulson]

True,

 

but for a: -

 

hip, knee, ankle, toe

 

you are solving mutiple rotations which is where it gets tricky (for me at least).

 

Hopefully in a couple of days I'll have it working and I'll post all the code.

[flachdrache]

I knew quaternions once.

[TylerH]

I lost track of this thread.

 

There are helper functions that convert Quats to Euler Angles, among other things. I did not implement conversions because that is up to the end user. Internally it makes calls to STL math functions that take radians, however, the actual parameters a Quat stores aren't Euler Angles in either degree or radian form, so the difference wouldn't exit. The only time you need to deal with Rad/Deg conversion is with the Euler Angles To/From.

 

Glad to hear you got it working. I haven't been very involved lately due to some school stuff.

[Canardia]

I read this on the Raydium website (a great engine to learn how to write engines btw), and I kinda understood for the first time how to use quaternions:

2.27 void raydium_math_quaternion_multiply(float *q1, float *q2, float *result):

Multiply two quaternions and fill the result with the resulting quaternion.

Quite usefull for making rotations over quaternions.

Here a list of common quaternions:

W X Y Z

1,0,0,0 Identity quaternion, no rotation

0,1,0,0 180' turn around X axis

0,0,1,0 180' turn around Y axis

0,0,0,1 180' turn around Z axis

sqrt(0.5),sqrt(0.5),0,0 90' rotation around X axis

sqrt(0.5),0,sqrt(0.5),0 90' rotation around Y axis

sqrt(0.5),0,0,sqrt(0.5) 90' rotation around Z axis

sqrt(0.5),-sqrt(0.5),0,0 -90' rotation around X axis

sqrt(0.5),0,-sqrt(0.5),0 -90' rotation around Y axis

sqrt(0.5),0,0,-sqrt(0.5) -90' rotation around Z axis

TODO: Those could be added like defines

[TylerH]

They are more complex than that, since the W component can be imaginary.

 

The beauty of quaternions is:

 

x^2 == y^2 == z^2 == x*y*z == -1

[Chris Paulson]

Well last night I got it working.

 

I need to improve the joint constraints and pick the spaghetti out of my code and I'll post it.

 

Watch this space.

[Chris Paulson]

Hi,

 

I've updating my blog with a video of the IK and the source code.

 

Enjoy...

 

Blog: -

http://leadwerks.com/werkspace/index.php?app=blog&module=display&section=blog&blogid=19

[Niosop]

Awesome job Chris!

[VeTaL]

Okay, i worked around your IK (great, Chris), but i want to know discuss some details.

I see, that on the ground it works perfect, as on 1st picture... But on the not-plane surface, looking like IK ignores its shape.

 

On the left picture, it works good, but on the right - ignores surface.

[Chris Paulson]

Good job at getting my code working so far.

 

The IK stuff needs at lot more work on it to get it working. It needs to the following: -

 

predict the next foot placement

do a line pick on it

Work out a height to place model/hips

IK adjust both feet to meet correct floor height (determined by line pick)

 

I halted work on IK stuff to get the blend tree code working, as this would effect the next foot placement.

 

The first picture is showing a bug as it says the stance is "unknown", I have already fixed this bug in the animation analysis code.

 

 

I really didn't think you'd get your head around my code so fast, amazing.

[VeTaL]

Thanks, it wasn't good time to cut it from recast and so on

Anyway, i just launched your code, i still dont know, how 70% of it works, but i'm going to manage with it. Also, i'm going to search more info about IK: i have a thought that your dividing to cycles is a bit too complicated.

[Chris Paulson]

The dividing into cycles is simple compared the usual stuff. Have a look a aigamedev web site etc. They use complicated maths such as point clouds and interpolation, my project is much simpler.

 

Also read rune's thesis.

[VeTaL]

http://grail.cs.washington.edu/projects/styleik/

Smooth intepolation between three walking styles

 

Interesting video, but there is no comments, how it is done...

[VeTaL]

Hey, i have an idea... Maybe we can take the "next" frame and apply forces to the current joints to move them?

I still dont know, how your code works 100%, so maybe this idea would be a little stupid, but i just wanted to share it.