MeshFromPoints Options Settings

Over here I have a point cloud of about 10.000 mm length. The cloud contains about 6 million points.
What would be convenient settings?

Rhino help doesn’t explain anything:

Hi Gerard - I have to admit this is somewhat of a black sheep command - I’ll see what I can find out - I have virtually no experience with this myself.

@Gerard - I’m sorry, but it looks like I was more or less correct… this command is in there as ‘better than nothing’ sort of band-aid. As you may recall, users used to, back in V3, V4, have to go out and download the plug-in themselves because it could not be installed with Rhino. Now it can but it is still only a band aid.


1 Like

I tried this command based on an existing mesh (extracting the vertices from it) and recreated with MeshFromPoints. The default settings seem to give “sensible results” (it seems based on an analyze of the indata).

I also tried to play around with the settings, and I think I can say that

NumOfContouringGridCells - … the NumOfContouringGridCells determines the size of a “cube” inside which the algorithm is trying to construct a surface. A smaller value would mean bigger cube(s), and thus better chance for the surface to “cover” small holes. Smaller cubes (bigger number of GridCells) tends to make the surface sneak into smaller spaces between the points, and so make the surface very detailed but also into a very “noisy” fine-grained topology.

AutoAdjustGrid - Set to No to manually set the NumOfContouringGridCells parameter.

SamplingDensityPlusNoise - determines the size of the final triangles. Combined with small GridCells (=bigger number of cells) you get a more detailed surface, but also full of small holes and peaks. Bigger value means bigger triangles.

There’s a balance between the two (density and gridcell size (or number of) which often seen the most sensible choice, resulting in the “best” surface (where “best” which of course depends on your use case).

? - This algoritm actually produces a "interesting results, although far from good enough in my own case, but perhaps this old algoritm could be enhanced a little bit and become a very powerful one.

In any case, I do agree with the idea that this command should be kept available as a “better than nothing” command.

// Rolf

1 Like