Grasshopper

algorithmic modeling for Rhino

I've been experimenting with some geometry in GH, and have created a twisting tube spiral shape. For my project, I need this to be one solid shape, but the way I have created it means it passes through itself multiple times. Is there any way in grasshopper to create a solid form just from the overall external shape, rather than having it intersect itself (almost as if I were making a mold). I'm really stumped on this one, although I admit I am not an expert by any stretch

I hope this makes sense!

Views: 7798

Attachments:

Replies to This Discussion

Sequential Booleans is brilliant, with the advantage that the form could also change to any distorted obloid along the way. However, you are leaving behind a trail of little centipede segments any which might kiss the new sphere too closely and kill the Boolean. I guess if it fails, so what as long as you can ignore one link and increment further to get lucky again.

Also as new spheres approach existing form, if it merely kisses *anywhere* instead of boldly overlaps everywhere, that step will fail, so indeed a smart loop is needed, not a brute force one.

Rhino already uses a loop to merely automate repeated splits and joins, and its mainly the kissing factor that kills entire Booleans.

By the way, Cocoon would also work with many spheres or distorted spheres at once, affording an outer overall mesh surface and many isolated inner surfaces, as long as you keep the radius/power set to only slightly offset away from the sphere surfaces. This is redundant though, since his original tube will also work with Cocoon this way and you'd merely need to delete all but the outer shell mesh surface. This will offset it all a bit bigger, but you can offset the mesh back down, or run some smoothing on it which does the same thing.

Bizarre interface thus high learning curve ZBrush has a Unified Skin command that uses little cube voxels to create a surface mesh no matter what complexity lies within. I can tell it uses cubes since the lower settings results in a little cubes inside the form!

ZBrush then also has a brand spanking new vastly superior to anything else ZRemesher that smartly (follows topology well) creates quad meshes that can be readily converted in Rhino back to NURBS polysurfaces, albeit large ones.

The 3D Print Exporter script is where you import/export STL or OBJ files. You must use OBJ to export to not lose the quad mesh faces that convert so well to Rhino quad faces. And the mesh is ALL quads, by the way.

Here it is imported into Rhino and converted to a NURBS closed polysurface/solid:

ZRemesher is the reason I climbed the steep ZBrush learning curve. It is actually a 2.5D canvas program, for making illustrations using pseudo 3D shading. All the 3D modeling ability is thus hidden in the tools and subtools palette, not in the main program view. Upon importing an STL, it only imports into a tool you then need to manually drag out onto the canvas and then remember to immediately hit the Edit button to get out of 2.5D canvas mode. There are no units either, which is something I'm still learning about how to retain.

This tells me something. Cocoon is missing a crucial feature, a rather simple feature! To fill a solid with cubes is merely a yes/no inside/outside criteria, for a regular XYZ stack of test cubes, but there's no test, no isofield, just where cubes go or not and then marching cubes could surface that.

Attachments:
By the way, the real power of ZRemesher that turns any solid model into quads is that the efficient quad meshes can be converted to highly editable T-Splines in Rhino, rather than directly to NURBS solids. The T-Splines soft selection option makes up for the high face count.

It does look like a near perfect solution, such a shame it has to use another piece of software! I'm looking into some of these ideas now, all very interesting! I'm surprised there isn't this functionality in Rhino/GH already

Here is how to use Cocoon alternatively on your existing or any other NURBS solid, and indeed in this case of using NURBS as a Brep input requires end caps in order to isolate an outer mesh surface from a bunch of inner mesh stuff. It creates a field in space around the surface then creates a surface along a constant field value (iso value).

Use a tiny sample piece to set it up before an extended run for the whole object.

I used MeshMachine since the Cocoon refine component that uses some MeshMachine code inside is actually rather tweaky and only refines instead of really remeshes adaptively and doesn't smooth much.

Attachments:

This is a great definition!

Note that allowing MeshMachine to simply collapse the mesh via a zero (0) Pull input makes it much faster, so you then control it from damaging collapse by only using a few Iter steps. Also note however that mesh target edge length affects how fast collapse takes place too, so preview with faster bigger mesh target won't match final result with fine mesh target until you then increase the Iter steps.

The Frames input is hard set to 2 frames, just enough to trigger MeshMachine to do its thing, but this outputs two data tree branches, thus the filter on the MeshMachine output.

MeshMachine only outputs Plankton meshes, and the Plankton file may be missing from Kangaroo, depending on the version, so search the forum for it as needed.

I have used Cinema 4D with success in the past to remove interior geometry.
I wanted to try it on Spiral SHape.3dm, but I need to export in .OBJ format from Rhino. I have tried several options from the OBJ Export menu, all without success......Yikes!

Which options should I use?

I'm convinced that because most dedicated computer science majors are lonely dorks who play lots of video games, that BOOM, on schedule, no matter what, getting things done on computers is never simplified, but always made into a strategy game. It's just how they think. It must "default to fail" or else what's the point? Where's the drama that makes up for their lack of a (good) sex life?

Clearly the media is running out of names for the next generation......Might I suggest Gamers!

FYI: Amazon has just released Lumberyard based on the CRYENGINE. The D/L is FREE for anyone. You will be charged if you intend to test drive your game on the Amazon Cloud.

I've just seen the future: Its a  F**ken Game! ....Yikes!

https://aws.amazon.com/en/lumberyard/

This problem has been solved by parallel processing in Python, using a piece jostling strategy to fix any and all failed Booleans, operating on pairs of pieces, from balls on up to the full segment, doubling segment size each step:

http://www.grasshopper3d.com/forum/topics/help-creating-a-smart-loo...

RSS

About

Translate

Search

Photos

  • Add Photos
  • View All

© 2024   Created by Scott Davidson.   Powered by

Badges  |  Report an Issue  |  Terms of Service