CloudCompare forum

Hi

I'm having an issue with C2C. I need to compare a road scan to an aerial scan.

The preliminary super quick distance computation is more than enough for what I need. Sadly there is no way to save it to SF.

Whatever other option for "real" computation, freezes computer, even if I manually choose lowest octree.



If you could add something like "preliminary only" option in ui and command line please.

So generally lowering the octree level is bad for the computation time (as it ends with the comparison of more points with more points). Also, you can try to remove duplicated points (if any - with the 'Tools > Other > Remove duplicate points' tool) as this can be very bad for the performances.

Also, setting a maximum distance can help a lot reduce the computation hit.

And no, it's not possible to export the approximate distances (as some people mistake them for real ones).

But when I create an octree with Edit>Octree>Compute In the DBtree when I set to visualize the octree, the lowest values are the sparse ones and as octree goes up there are more points, should then octree 1 be the simplest way to compute distances?

It is wild that the preview can´t be baked as an SF as certainly looks good enough. If not real distances may be could be aproximated since at the end of the day any cloud is confined in a bounding box with known dimensions and the gradient is smooth enough for a rough stimation.

I tried with highest octree and again computer froze.

edit: I'm doing 1m dist compares for another task with theese clouds with no issues, in command line, so likely what causes the struggle is the distance, but I need several meters not just 1 like now, something like 70 or so. I guess too much for fine distance computation. I may give it a shot in command line, just in case...

A level of 1 means basically that you only divide the cloud in 8 parts. Then CC will try to compare 1/8th of the cloud with the other 1/8th of the cloud (in the best case, can be more). So the square number of points divided by 64... Which is 10^12/64 = 15 billion comparisons for a 1 million point cloud! That's huge.

Clearly, the C2C tool is not meant to compute large distances, as the farther a point is, the more points in the other cloud you have to compare with.

And currently it is not possible to export the approximate distances because that's not what most users are looking for. I guess it's technically possible to get access to it by modifying the code...

thanks for the explanation!.

Yeah sure, totally understand that not mainstream features not to be implemented.