Before starting to work on a dense point cloud I suggest to lighten it, without losing information, getting rid off the “duplicate points“.
There could be points so closed one to each other than they do not actually add a real benefit for th 3Dmodel.
You can delete them with Cloud Compare, open source software for managing point clouds.
Just follw this path, after selecting the point cloud you’d like to edit in the db tree:
“Remove Duplicate Points”.
In the panel that will pop up you just have to enter the minimum distance you want to be kept between the points of the new cloud.
Default value is incredibly low.
But it will select some points for sure.
I’ll show you what happens on a structure from motion point cloud made of 15,200,000 points (5 cm of accuracy).
Here is listed the minimum distance i chose and the number of the points at the end of the process.
- min. distance 1mm – 15,000,000 pts
- min. distance 5mm – 14.500.000 pts
- min. distance 1cm – 11,800,000 pts
- min. distance 2cm – 5,200,000 pts
- min. distance 5cm – 1,100,000 pts
Personally I would stop at the 5 mm of minimum distance.
Going further would mean losing too much information.
Anyway, you could take away 5% of the points that shoud not be managed in further processing and calculations.
Remove duplicate points is a non destructive tool.
A new point cloud is added to the project without modifying the original data.
If you would like to subsample a point cloud i would sugget to use the “Subsample tool” instead.
It’s much more reliable and made for that purpose.