correct implementation of quaternion interpolation in java3d

[java3d-interest]

hi all,

The source code of Quat4f contains the following comments just after the method signature of both interpolate(Quat4f q1, float alpha) and interpolate(Quat4f q1, Quat4f q2, float alpha).

        // From "Advanced Animation and Rendering Techniques"
        // by Watt and Watt pg. 364, function as implemented appeared to be
        // incorrect.  Fails to choose the same quaternion for the double
        // covering. Resulting in change of direction for rotations.
        // Fixed function to negate the first quaternion in the case that the
        // dot product of q1 and this is negative. Second case was not needed.

I was wondering if the current implementation of interpolation is the correct one. It is implemented currently as followed:

     double dot,s1,s2,om,sinom;

     dot = x*q1.x + y*q1.y + z*q1.z + w*q1.w;

     if ( dot < 0 ) {
        // negate quaternion
       q1.x = -q1.x;  q1.y = -q1.y;  q1.z = -q1.z;  q1.w = -q1.w;
       dot = -dot;
     }

     if ( (1.0 - dot) > EPS ) {
       om = Math.acos(dot);
       sinom = Math.sin(om);
       s1 = Math.sin((1.0-alpha)*om)/sinom;
       s2 = Math.sin( alpha*om)/sinom;
     } else{
       s1 = 1.0 - alpha;
       s2 = alpha;
     }

     w = (float)(s1*w + s2*q1.w);
     x = (float)(s1*x + s2*q1.x);
     y = (float)(s1*y + s2*q1.y);
     z = (float)(s1*z + s2*q1.z);
  }

I am using RotPosPathInterpolator to animate an object, but I am not getting the animation I am expecting. I am not sure if this is due to a bug in my code or due to quaternion interpolation used by RotPosPathInterpolator class.

Thanks,
Rayhan
[Message sent by forum member 'rayhan_rahman' (rayhan_rahman)]

http://forums.java.net/jive/thread.jspa?messageID=292527

---------------------------------------------------------------------
To unsubscribe, e-mail: interest-unsubscribe@...
For additional commands, e-mail: interest-help@...

[java3d-interest]

forget about the RotPosPathInterpolator. it's to difficult to use (for me).
try to combine a rotationinterpolator (one transformgroup) and a positioninterpolator (the other transformgroup) to get rid of this problem.
[Message sent by forum member 'optimusprime1982' (optimusprime1982)]

http://forums.java.net/jive/thread.jspa?messageID=292563

---------------------------------------------------------------------
To unsubscribe, e-mail: interest-unsubscribe@...
For additional commands, e-mail: interest-help@...

There has been one issue:
https://java3d.dev.java.net/issues/show_bug.cgi?id=73

But this got resolved. I use it w/o issues these days.

- J

java3d-interest@... schrieb:

> hi all,
>
> The source code of Quat4f contains the following comments just after the method signature of both interpolate(Quat4f q1, float alpha) and interpolate(Quat4f q1, Quat4f q2, float alpha).
>
> // From "Advanced Animation and Rendering Techniques"
> // by Watt and Watt pg. 364, function as implemented appeared to be
> // incorrect.  Fails to choose the same quaternion for the double
> // covering. Resulting in change of direction for rotations.
> // Fixed function to negate the first quaternion in the case that the
> // dot product of q1 and this is negative. Second case was not needed.
>
> I was wondering if the current implementation of interpolation is the correct one. It is implemented currently as followed:
>
>      double dot,s1,s2,om,sinom;
>
>      dot = x*q1.x + y*q1.y + z*q1.z + w*q1.w;
>
>      if ( dot < 0 ) {
>         // negate quaternion
>        q1.x = -q1.x;  q1.y = -q1.y;  q1.z = -q1.z;  q1.w = -q1.w;
>        dot = -dot;
>      }
>
>      if ( (1.0 - dot) > EPS ) {
>        om = Math.acos(dot);
>        sinom = Math.sin(om);
>        s1 = Math.sin((1.0-alpha)*om)/sinom;
>        s2 = Math.sin( alpha*om)/sinom;
>      } else{
>        s1 = 1.0 - alpha;
>        s2 = alpha;
>      }
>
>      w = (float)(s1*w + s2*q1.w);
>      x = (float)(s1*x + s2*q1.x);
>      y = (float)(s1*y + s2*q1.y);
>      z = (float)(s1*z + s2*q1.z);
>   }
>
> I am using RotPosPathInterpolator to animate an object, but I am not getting the animation I am expecting. I am not sure if this is due to a bug in my code or due to quaternion interpolation used by RotPosPathInterpolator class.
>
> Thanks,
> Rayhan
> [Message sent by forum member 'rayhan_rahman' (rayhan_rahman)]
>
> http://forums.java.net/jive/thread.jspa?messageID=292527
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: interest-unsubscribe@...
> For additional commands, e-mail: interest-help@...
>
>
>  

--
Joerg Plewe
http://www.hardcode.de
'Life is hard and then you die.'


---------------------------------------------------------------------
To unsubscribe, e-mail: interest-unsubscribe@...
For additional commands, e-mail: interest-help@...