Hope someone can help with this. I am reading about quaternions being used as rotations in 3D space. Lets say I have a quaternion q (where q = s + v, s being the scalar part and v the 3 dimensional vector), and a point P in 3D space. The text I am reading says that apply q to P as qP yields the following: ((-v.P) + sP + (v x P)) [ where ''.'' indicates dot product and ''x'' indicates cross product] Can someone explain this. I can see how the term sP arrives but as for the cross and dot product I am not sure. Is it that multiplication of two vectors is vP = v x P - v.P ?? I would have though that multiplying the two vectors would have just used the cross product. Thanks for any help you can give.

quaternion multiplication is defined as:

[s1, v1] * [s2, v2] =[s1*s2 - v1.v2, s1*v2 + s2*v1 + v1Xv2]

where '*' means scalar multiplication, '.' means dot product, and 'X' means cross product.

The inverse of a quaternion is defined as:

q-1 = (1/||q||)2 * [s, -v]

where ||q|| = sqrt(s2 + x2 + y2 + z2)


To rotate a vector, v, represent v as [0, v].
v' = q * v * q-1

The represenation of a rotation of theta degrees around the an axis of rotation (x, y, z) is:

q = [cos(theta/2), sin(theta/s)*(x, y, z)]

Given a unit quaternion [s, x, y, z] the corresponding 4 dimensional rotation matrix looks like this:



(the spacing was messed up without the source tags)

These are just definitions and i don't entirely know what the derivation is in all cases, but they work. I actually think the definition of multiplication makes sense in a way, and all the other formulas are more or less derived from that.

[edited by - kdogg on March 17, 2003 1:33:53 AM]