Page 1 of 1

Mean distance with absolute value and Max relative error

Posted: Thu May 28, 2015 1:59 pm
by ripo
Hi Daniel,
I have two question for you.
1. I have a point cloud with a fitted primitive. I used “compute cloud/mesh distance” and it show the “mean distance” from point cloud and primitive. This value is calculated with positive (above the primitive) and negative (under the primitive) distance compared to primitive. I want to know if is possible an average calculated with distance in absolute value.
2. In “Distance computation->Approx. results” there is “Max relative error”. I would like to know what the numbers mean.
For example 3.77496 + 1.94995/d % (d>0.0194995)
Thank you

Re: Mean distance with absolute value and Max relative error

Posted: Fri May 29, 2015 6:39 am
by daniel
1. Don't you have a 'signed' distance checkbox on the Cloud/Mesh distances computation dialog?

2. One should never use the approximate distances (as they are highly approximate). The d is the cell index (at the current octree level). In your case it means that you have an absolute uncertainty of at least 3.8 units plus a variable quantity that depends on the distance of the point (so for very close points you have almost 6 units of uncertainty, and this uncertainty decreases for farther points (but it stays above 3.7 in any case ;).

Those approximate distances are only used internally by CloudCompare to automatically set the parameters (i.e. mainly the octree level).