offset tool

[hd_surveyor]

Hi, Daniel

I want to know if you can add an offset tool for point clouds.
in other words, for each point of the original cloud, you have to create a sphere and then add up all the spheres and find a cloud that encloses everything

Thanks
Alberto

[daniel]

All the magic is in 'find a cloud that encloses everything'... How would you do that?

[hd_surveyor]

Thanks for the quick answer.
This is my current solution, is slow and inaccurate, I hope you can improve it.

regards
Alberto

[hd_surveyor]

another possible approach is to place a sphere with a radius equal to the offsett distance on each point, every 3 adjacent points there are 2 points at the intersection of the spheres, after eliminating the points nearest the offset distance, the remaining points represent (with a certain approximation) an offset of the original cloud. To function properly the offset distance must be much greater than the distance between sets of adjacent points.
in addition to the offset radius may be necessary to add to the input values the maximum search distance of point triplets.

regards
Alberto

[daniel]

The only thing I could reasonably add is a method that would expose the Distance Transform of the point cloud. This is a grid with the distance to the nearest point in each cell. You would be able to choose the grid step. And then you could extract all the points (= cell centers) that are at a given distance to the cloud.

[hd_surveyor]

thanks, I think that would be a big improvement.

regards
Alberto

[daniel]

Ok ,the latest online 2.6.3 version has a 'Distance map' method in the 'Tools > Sand-box" menu.

This will ask for the grid resolution and an optional margin. Then it computes the distance for all points in the bounding box (+ margin) around the cloud.

You can select the points you are interested in (i.e. at the right distance) with 'Edit > SF > filter by value. You can first reduce the SF display range dynamically so as to get the right filtering range.

[hd_surveyor]

Thank you.
The distances in the scalar field DT values are expressed in box dimension / distance map resolution?

Alberto

[daniel]

Oops, you're right! I forgot to multiply the DT values by the cell size :D

It will be fixed in the official release...

[hd_surveyor]

thank you, it is not a big problem.
I noticed that easily create clouds with 200 million points million points of which useful two million, there is a way to avoid the calculation of unnecessary points, for example by imposing a maximum even approximate distance (octree level?) in the input parameters to reduce the computation time?

Alberto