Hi,

I'd be very surprised by an inaccuracy like you describe.

Are you sure that it isn't just a general 3D rotation where the symmetries no longer exist?

The true test of whether something is a rotation matrix is whether the determinant is 1.0 or not.

So you can always test the accuracy of the rotational part of the decomposition by calculating the determinant.

All the best,

Mark

On 12 Jul 2018, at 21:15, Jack Sadowsky <[log in to unmask]> wrote:

Hi,

Aside from any inaccuracy arising from the choice of a center of rotation, what degree of accuracy can we expect from the avscale decomposition?

I ask because I have noticed in my own use and in others' posts that even with rigid-body transforms the rotation and translation matrix output by avscale appears to be inconsistent. It is my understanding that, for example, in a 2D pure rotation transformation defined by

a  b  0

c  d  0

0  0  1

a = cos(theta)

b = -sin(theta)

c = sin(theta)

d = cos(theta)

the absolute value of b and c should be identical. (Of course in our case the transformation is 3D, but I don't believe that affects the off diagonal terms.) However, in practice I have noticed them to be off by more than a factor of 2 (or, in absolute terms, as much as 1 degree).

Best,

Jack

---

To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1

---

To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1