Normal Calculation not working as expected.

[__G__]

Hi All,

I'm trying to calculate normals on a point cloud, and it's not working as expected. Most points are either a Z normal value of 1 or -1 with very few values between - image attached.

2024-04-30_18-15-09.png (317.93 KiB) Viewed 412 times

I've tried quite a number of settings in the compute normal dialogue box, but nothing seems to make any difference.
The most recent attempt was using the quadric method, with an octree radius of 5 (previously 1.3, and 0.2) with other settings as per the screenshot below.

2024-04-30_18-17-27.png (18.94 KiB) Viewed 412 times

The other observation is that in flat ground areas, the normal calculation doesn't seem to be able to distinguish between up (Z=1) and down (Z=-1)

Any suggestions?

[daniel]

Not sure to understand... Does your scan have an associated scan grid (because in this case you don't need to use any of the other options.

If not, then you need a radius as large as possible so that enough points are captured in all areas of the cloud (normally the 'Auto' button works quite well to have a first estimate).

Last, and if the scan grid(s) are not present or are not working as expected, what do you get if you choose a 'preferred orientation' of +Z? (which should work for a large part of the cloud).

And don't hesitate to share your cloud with me (admin@cloudcompare.org) if you want me to look at this in more details.

[__G__]

Hi,

I'm not sure what you mean by 'scan grid'. That setting was checked by default, so I didn't change it.

[daniel]

If there's one (or several), you would see them in the cloud properties. And if there is, then you can probably remove all the other options.