boolean ball collisions

Hey guys, I'm trying to make a pretty complex boolen, which get my computer stuck, wondering if there is a good way to get only the top surfaces if intersecting balls..

doing boolean operations and splitting doesn't really work for this one as its too damn large. thanx

Attachments:

erg2.jpg, 169 KB

Replies to This Discussion

Permalink Reply by peter fotiadis on October 19, 2015 at 6:35am

Diagnosis based on Image(s) is not (yet) invented, he he.

Make a "typical" sample of these balls (say: 30 balls not 1M) in the exact way (Tree/List) that are sampled and post it here.

Permalink Reply by kipodi on October 19, 2015 at 6:55am

hmm there's a start up :)

I will post a file shortly , thank you

Permalink Reply by kipodi on October 19, 2015 at 7:13am

there it is..

I'm trying to get only the top surfaces. its basically a 3d voronoi so I was hoping there is a smart way to do the boolean subtraction as they all have the same radius..

Attachments:

ball collision.3dm, 767 KB

Permalink Reply by kipodi on October 19, 2015 at 10:42am

attached is a file

Attachments:

ball collision.3dm, 767 KB

Permalink Reply by peter fotiadis on October 19, 2015 at 11:37am

Hmm .. another "simple" case ... eh? he he.

On first sight I thought to unite all that stuff of yours and then extract the non planar BrepFaces and control what to include in some list (for instance (a) non planar and (b) non "touching" the "base" etc etc) .

But ... GH Union fails (and the Rhino Method attempted as well) without giving a reason/clue > so I've spend some minutes for that (you know ... the usual).

Appears that solid at index 14 causes havoc (but is reported as valid and there's no difference at base Z (that usually causes issues)). Maybe some tolerance issue maybe "corrupted" solid (most unlikely) ... who knows?

Anyway ... play a bit with this attached: it does "arbitrary" user controllable union (up to an index, or up to all)... that way you can at least locate the bad boy. So what may you ask? Well ... there's always Plan B.

more soon (with Plan B).

Attachments:

balls_VS_Karma_V1.gh, 14 KB

Permalink Reply by peter fotiadis on October 19, 2015 at 12:04pm

Plan B (120+ times faster) trims each brep individually against the "proper" surface (shown just a delauney extruded grid of lines to outline the approach).

Permalink Reply by kipodi on October 19, 2015 at 12:34pm

oh now you're talking.. but isn;'t that the "dual" of the pattern if you are working off the center points?

Permalink Reply by kipodi on October 19, 2015 at 1:28pm

Hey, yes I see the faliure. I think boolean operations is a whole field.. and since I eventually wanna run it on a more complex array of spheres I need something robust.. do you know if there is a software which specializes in these operations? it seems as rhino is not too stable in this.

Permalink Reply by Joseph Oster on October 19, 2015 at 1:50pm

GH can be annoyingly intolerant of very small errors, which is why I gave up on the sliced spheres in your Rhino file and re-created them in GH. 'Solid Union' had no trouble with the whole spheres, and the resulting brep can be cut successfuly with a large sphere as well as the plane I used the first time:

(re-posted because I forgot to internalize the center point of the large sphere)

Attachments:

spheres_collision_2015Oct19b.gh, 11 KB

Permalink Reply by Joseph Oster on October 19, 2015 at 12:06pm

Attachments:

spheres_collision_2015Oct19a.gh, 13 KB

Permalink Reply by kipodi on October 19, 2015 at 12:35pm

thanks, looks good, I"ll give it a shot..

Permalink Reply by kipodi on October 19, 2015 at 3:28pm

thanks the method is gread, the problem is that when I take the complexity level up it fails.. and it technically shouldn't.. at least to the naked eye. I wonder if there is a good solution in the market for complex booleans,, even if it requires some tweaking of the original model. I here coin the term "booleanable" :)