3
$\begingroup$

I have a very accurate mesh of my head, captured by photogrammetry (http://ir-ltd.net/)

enter image description here

I would like to rotate and translate it so that it is as symmetrical as possible on the x axis.

How could I go about doing this?

Here is a close up of an ear:

enter image description here

Now maybe I could position a sphere...

enter image description here

enter image description here

And then position one of the opposite side in the same place, and then take the average of their centres, translate that back to the origin, and then perform appropriate rotation.

I can write a Python script that gets as far as the translation, but I'm not sure how to do the final rotation.

Also can anyone think of any other techniques?

PS it is essential to mesh scale must be preserved!

EDIT: Another technique might involve selecting one ear and calculating the average of all vertex positions. Then doing the same for the other ear. And then taking the line between these two centroids to be the new X axis.

But then the problem becomes how to select the ear? Specifically, how to make sure the second ear is selected in a similar way.

$\endgroup$
2
  • $\begingroup$ So your scanned head is rotated somehow in 3D space, and you want to rotate it to be axis-aligned (with X or Y axis), and the center should be at the scene origin at 0,0,0? (center as in the middle of a virtual line connecting both ears for instance) $\endgroup$
    – CodeManX
    Feb 21, 2014 at 1:26
  • $\begingroup$ Yes! I can get it pretty close by eye, but not exact. $\endgroup$
    – P i
    Feb 21, 2014 at 7:16

1 Answer 1

3
$\begingroup$

A easy way would be:

  1. Select three vertices (right, left and middle) on the head;
  2. Create a empty based on that virtual plane using Precise Align.
  3. Select the empty, the AltR. You may want to rotate manually along Z axis if it's not expected after that.
  4. Select your object, AltP > Clear and Keep Transformation. Then delete the empty.

enter image description here

Similar usage like this.

One thing that is worthy mentioning is: it's a scanned head in your case. Since real human body can be just roughly symmetrical, but not exactly as it appears, so the result may not be perfectlly symmetrical as expected. Of course, you can still give it a try.

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .