I am trying to solve the Orthogonal Procrustes problem inside Blender for a specific case but I am having some trouble. I have a face and a pair of glasses in my scene. They are not in a specific position, they can be anywhere.

enter image description here

I have then defined the following function using the numpy library, that is an implementation of the Orthogonal procrustes problem:

import numpy as np

def procrustes(X, Y):
  X = X - X.mean(axis=0)
  Y = Y - Y.mean(axis=0)
  (U, S, Vt) = np.linalg.svd(np.dot(Y.T,X))
  R = np.dot(U, Vt)
  Y = np.array(np.matrix(Y) * np.matrix(R))
  return X, Y

It finds the best translation and rotation for putting glasses on the face (it finds an orthogonal matrix which most closely map Y to X) and now I want to apply it. In my case the translation can be avoided and we can assume that it's already fixed. So we can consider only rotation for simplicity. X and Y should be two np arrays of points in the space.
Do you have any idea of how I can apply this function to my problem?


1 Answer 1


First of all, I would suggest using the letters A and B, instead of X and Y. Furthermore, you don't document what type they are; I assume those are matrices. You can call X.to_euler(...) to get the rotation from the matrix, and assign it to ob.rotation_euler.

For more info, see the mathutils API documentation.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.