Page 1 of 1

quantifying how good a mesh of a ball is

Posted: Thu Dec 04, 2014 8:32 am
by tamir.e
Hi Daniel,
First i would like to thank you again for the SW, it's very use-full.
I have several meshes created by different 3D scanners of the same ball.
Image
I would like to be able to quantify which mesh gives a better representation of the ball.
for example:
1.what is the valid percentage of the ball scanned
2.whats the STD for each pixel.
i used RANSAC algorithm to create a speculated ball, now i would love to extract the data above.
Image
is this possible using CC?
Thanks,
Tamir

Re: quantifying how good a mesh of a ball is

Posted: Thu Dec 04, 2014 9:15 am
by daniel
Yeah, there's several ways to do this.

The simplest: if you have enough vertices on your original mesh, simply select its vertices and the sphere primitive generated by qRansacSD (should be beneath the sub-cloud corresponding to this spherical part in qRansacSD output). Then compute the cloud/mesh distances between those two entities (Tools > Distances). This will give you numerical estimations of how well the original mesh 'fits' to a sphere. Afterwards you can also segment the vertices based on their distance to the sphere (Edit > Scalar fields > Filter by value). So you can determine how many points are close, far, very far, etc.

If the meshes have too few vertices, you can use the same workflow by sampling (a lot of ) points on the original mesh and use the resulting cloud instead of the vertices.