Voronoi Sphere

Hey Everyone!
Any ideas about creating a voronoi sphere? 2d is fine.

Replies to This Discussion

Permalink Reply by Manuel on December 14, 2009 at 11:52pm

Vicente Soler made a brute force voronoi 3D vbnet component, is very slow but it works.

Here I made a sphere, then I took some random points on the surface and applied the voronoi3d vnet component:

Another method would be if you take some random points on half sphere, project those points on a plane, use the 2d voronoi and project back the voronoi curves to the sphere... then repeat that for the other half.

Permalink Reply by Giselle Dcruz on December 15, 2009 at 12:27am

thanks manuel...but im looking more for a sphere with voronoi holes like in the picture.

Attachments:

voronoi sphere.jpg, 41 KB

Permalink Reply by Manuel on December 15, 2009 at 8:11am

Here 200 points, took 25 minutes to calculate, then I keep only the curves (cull the lines, polylines, etc) and use pipe:

Permalink Reply by David Rutten on December 15, 2009 at 8:36am

25 minutes... wtf.

--
David Rutten
david@mcneel.com
Poprad, Slovakia

Permalink Reply by Giselle Dcruz on December 15, 2009 at 10:46am

Wow how did u calculate the 200 points? and could you upload the gh file as well? thanks so much.

Permalink Reply by Luis Fraguada on December 15, 2009 at 3:51pm

25 mins to make the definition or for GH to calculate it?

Permalink Reply by taz on December 15, 2009 at 4:27pm

Here's a related question... Open to everyone...

Since a NURBS sphere is created from a single surface when you take a random sampling of a large number of points on the sphere surface there is a clustering of the points around the "poles".

I'm guessing that is also the case with Manuel's Voronoi ball above...

Is there a clever way to get a uniformly random distribution of points across the geometric (not NURBS) surface of a sphere?

Permalink Reply by taz on December 15, 2009 at 4:39pm

To answer my own question, I just heard about this thing called the "internet"...

http://mathworld.wolfram.com/SpherePointPicking.html

Permalink Reply by Vicente Soler on December 15, 2009 at 4:50pm

I was about to comment on that. Another problem that might arise (I'm not sure if Manuel solved or just camouflaged it), is that it will create a seam in the sphere. To solve this using the available components, you could copy the points in uv space as if you where tiling a texture:

Original set of points:
O
New set of points:
OOO
OOO
OOO

Input all of them into the voronoi component and then use only the cells in the center (I haven't tried this, if there's an obvious simpler way please say).

Permalink Reply by Manuel on December 15, 2009 at 7:09pm

Taz is correct about the clustering thing...

The seam that Vicente Soler mention, does appear, but if you just need the wires (for the pipes or something like that), the seam appear as an arc curve that you can cull out, so there is no problem with that... (thanks for the vb.net :D)

Luis: 23 minutes took the vb.net component to calculate the voronoi 3d for 200 cells. The B output of the vb.net component can tell you the time (thanks again Vicente). Anyway I was working on another thing, so maybe it can be a little less.

Images:

Permalink Reply by taz on December 15, 2009 at 7:20pm

It may not be exactly the same thing, but I've used Facet Dome since it looks like Voronoi and I didn't have to deal with the seam issue...

It may not be true Voronoi patterning.

Permalink Reply by taz on December 15, 2009 at 7:38pm

Ahh... Satisfyingly random with Facet Dome.

Definition is posted, but there's a lot of duplicate geometry I didn't try to get rid of.

At least it didn't take 25 minutes...

Attachments: