Fine registration vs compute cloud/mesh results meaning

[Lucreziard]

Hi all!
I'm quite new to CloudCompare and I have a question about the fine registration vs the compute cloud/mesh tool final values.

I need to compare two meshes and their deviation. I aligned them roughly and then hit the "fine registration" tool, after the fine registration was completed the software showed a window with the results (RMS value and others),then I computed the mesh/mesh distance.
I was wondering how to interpret the results though, since I am interested in understanding how different the two meshes are from each other: what is the difference between the RMS values after the fine registration process vs the mean/standard deviation values after the computed mesh/mesh distance?

Thank you very much in advance!

[daniel]

So there are 2 differences:
- the RMS is a slightly different calculus compared to the std. dev. (RMS = root of mean squares / STD DEV = root of 'mean squares minus the squared average')
- the ICP tool only uses a subset of the total number of points at each iteration. Therefore the RMS is computed on this subset and not on all the points

Mind that once you have compute the C2M distances, you can use the 'Edit > Scalar fields > Show histogram' tool, and the RMS of all the distance values will be displayed in the console (if you want to compare it with the one from ICP).

[Lucreziard]

Hello again and thank you very much for the answer.

So I am also wondering which one of these two calculations would you consider more reliable/correct to assess the accuracy of the compared mesh..?

Thank you!

[daniel]

Definitely the C2M distances. And you can use either the std. deviation or the RMS... Maybe the RMS is simpler (and it includes a potential 'bias' of the distances, it's not only about the variations around the average).

[Lucreziard]

Thank you very much!