How to Calculate @w to Aim RBD Object

[thompson.alt]

As I'm just starting to understand quaternions and orientation in Houdini, I'm not sure how complicated this is to achieve. I could be completely over thinking it.

Ultimately I would like to create a force in dops to rotate the packed object to its current direction of motion, or it's velocity. Modifying @w seems to be the best way to do that.

For a first step, I'm trying to make a setup that takes a rbd packed object and rotate it back to it's original orientation. While it technically does that, it doesn't do it logically. It flips around its axis in bizarre ways. As I understand, it's behaving erratically because it's trying to rotate around all of its axes at the same time.

This is the code I'm using in a pop wrangle node:

vector  qToE(vector4 q_value){
    float   q_0 = q_value.w ; 
    float   q_1 = q_value.x ; 
    float   q_2 = q_value.y ; 
    float   q_3 = q_value.z ; 
    vector  out = {0,0,0} ; 
            out.x = atan2(2*(q_0*q_1+q_2*q_3), (1-2*(q_1*q_1+q_2*q_2))) ; 
            out.y = asin(2*(q_0*q_2-q_3*q_1)) ;
            out.z = atan2(2*(q_0*q_3+q_1*q_2), (1-2*(q_2*q_2+q_3*q_3))) ;
    return out ; 
}


v@euler = qToE(@orient);
@w = -v@euler*1*{1,1,1};

This is where I got the quaternion to euler function:

I've attached my houdini file to this post. If anyone has any tips, I'd be very happy to hear them!

 

Thanks!

orient_rotate_v01.hip

[thompson.alt]

I've made some progress since yesterday. Here is the code I'm using now:

vector4 first_orient = @orient;
vector4 second_orient = {0,0,0,1};

matrix3 transform = lookat(@v, 0, {0,1,0});
//second_orient = quaternion(transform);

// Get the difference between the two quaternions
// https://stackoverflow.com/questions/1755631/difference-between-two-quaternions
vector4 transition = qmultiply(second_orient, qinvert(first_orient));

// Convert to Angle-Axis
// http://www.tokeru.com/cgwiki/index.php?title=JoyOfVex17#Convert_back_to_matrix
vector angle_axis = qconvert(transition);

//@orient = transition;
//@N = angle_axis;

@w = angle_axis*chf("strength");

For now they're trying to orient themselves with their velocity. This is working quite a bit better, but I think there's still some funny flipping going on.

[anim]

not sure if you are taking this as a VEX exercise, but if not, you can use POP Look At to easily do this

 

[thompson.alt]

18 hours ago, anim said:

not sure if you are taking this as a VEX exercise, but if not, you can use POP Look At to easily do this

 

Thanks Tomas! I had actually tried that earlier, but it wasn't working for packed rbd objects. It seems the problem was that they don't respond to @torque like particles do.

The solution I found was to convert @torque to @w after the fact. Does anyone know if there is a rotational force like @torque that work on rbd objects?

Now I can dig into how POP Look At and see how the math works too.

 

Thanks again!

[anim]

packed rbds indeed work with v@torque, try upping the Torque Scale multiplier if you are not seeing any results

also make sure sleeping is off if this is the only force acting on them