Page 1 of 1

Calculate Normals oriented to TLS locations

Posted: Wed Feb 05, 2020 8:59 am
by saintpat
Hi,

I'm quite new to Cloud compare and lidar pointclouds (though I have previous experience with photogrammetry). I'm looking for advice on a workflow for creating a mesh from an .e57 point cloud created from about 18 Terrestrial Lidar Scans. I've gotten as far as merging the pointcloud and extracting the Scanner locations with CloudCompare and I want to use these locations to calculate point normals (oriented towards the scanners). When I use the menu option to create point cloud normals I get the option to use camera positions. All good so far.

The problem is; there are over 400 million points in the original cloud. I want to calculate normals on a reduced size pointcloud, but when I subsample the pointcloud, it appears as a seperate entity in CloudCompare, without any TLS information attached to it. Is there any way to copy or move the TLS locations over for the purpose of calculating point Normals on the reduced cloud?

I'm looking to use the oriented point cloud for mesh reconstruction down the line. Thanks!

Re: Calculate Normals oriented to TLS locations

Posted: Thu Feb 06, 2020 5:57 am
by saintpat
I suppose my question boils down to- can camera/scanner locations be transferred between clouds.

I can currently do this by manually creating a new camera at 0,0,0 and copying over the transform information manually.

It would be great if there was an easier way to do this, such as drag/dropping (ideal) or saving multiple scanners in a text file and allowing the import of coordinates from a text file in the creation of new cameras.

Re: Calculate Normals oriented to TLS locations

Posted: Thu Feb 06, 2020 9:20 pm
by daniel
Yes, I believe this is something we should fix in CloudCompare (the transfer / duplication of sensor information when subsampling a cloud)

Re: Calculate Normals oriented to TLS locations

Posted: Sun Nov 06, 2022 8:44 pm
by framed
Hi Daniel. Just giving this one a polite bump. Retaining the sensor info when subsampling would be a very useful enhancement :)

Re: Calculate Normals oriented to TLS locations

Posted: Sun Nov 13, 2022 10:23 pm
by daniel
Ok, so I've gave a try to this idea, and it became a much larger refactoring.

Anyway, you can try the latest 2.13.alpha version.

(I also realized that with the previous version you could actually simply move around the sensor objects from one entity to the other... but now it should be done automatically)