algorithmic modeling for Rhino
Upon subdividing a directly produced quad mesh ( http://www.grasshopper3d.com/photo/mesh-to-nurbs-polysurface-conver... ) from a uniform Kangaroo 1 MeshMachine remesh, I noticed that little weave-ready ribbons appeared:
Though the Mesh+ plug-in has a few "weave" components, they only make local loops between adjacent mesh faces.
So I did the bookeeping required to create little construction ellipses and little stubby start and end lines offset in and out from each mesh edge to allow a curve blend and then rail sweep of locally rescaled ellipses along those curves, with lots of control.
There are occasional tube self overlaps due to kinks that would require fixing in other software to 3D print the output.
Interestingly, the result is sometimes a single knot curve, more often a mere handful.
Tags:
FYI: there is a nice collection of polyhedrons at : http://www.food4rhino.com/project/rhinopolyhedra?ufh That can be used with Nik's script
Now you are out of the bounds of the current script since the MeshMachine step required for it to work at all just makes a triangular mesh and then smoothes it out, so you might as well be inputting a blob.
The two strategies that make real sense are sculptural form with a relatively fine MeshMachine mesh, and manually produced triangulated mesh trusses that don't use a MeshMachine step.
If I ever generalize it to work with quads, pents, hexes on up, it will still need rather regular polyhedra, and there's no way to relax those easy since there's no way to actually remesh them, even conceptually, since they are of mixed face type. But manually produced trusses would be easier to play with, if not limited to triangles. Ah, yeah that's more motivating than being stuck with known polyhedra that I get to decorate once and go home.
Alas though Rhino 6 may support general meshing, Rhino 5 will only accept tri and quad faces, actually.
Hierarchical woven weave was obtained by avoiding failure prone mesh piping of the entire complex curve in favor of individual round capped mesh tube segments. Boolean union failed in Rhino, ZBrush lost detail even with the finest setting for Unified Skin, 3DS Max failed to import such a complex curve to convert to hair, Cocoon bogged down forever, but Geomagic Freeform automatically unioned the meshes upon import as voxel clay then nicely reduced the mesh to 97MB, just under the Shapeways file limit for 3D printing jobs. NURBS pipes simply crashed on a curve so big.
Caveat and better workflow: obtaining a perfect NURBS polysurface is tweaky especially the filetting step and in this wild curve blend system there is often kinking that ruins both the initial Boolean union and then the filetting, not to mention how self-intersecting tubes cannot be self-Booleaned in Rhino. So, simply outputting as a fine OBJ or STL mesh into ZBrush to create a unified skin that is then smoothed a bit rids any kinks and then the magical ZRemesher affords a new quad mesh that T-Splines can smoothly convert to an efficient if somewhat patchwork smooth NURBS polysurface. No worry, with albeit some loss of total elegance.
The kinking is inherent in the basic geometry of fat tubes along tight curves, seen best by arraying many balls along the curve.
Here I used Cocoon in fact, on the NURBS surfaces as meta surfaces, to obtain a smooth mesh for the ZBrush/T-Splines conversion to NURBS.
Boolean unioning separate, quickly produced individual round capped mesh pipes is also possible in relatively cheap 3D-Coat with its voxel system. It's slower than very expensive Geomagic Freeform on initial import as voxels, and import Scale is how you adjust the voxel size, but it's then very responsive and a 3D mouse Space Navigator works well too, better than Freeform. ZBrush lacks 3D mouse support, and cannot handle super fine detail with its own voxel import system that imports a unified skin, not retaining voxels.
Here is how jewellers are enjoying 3D-Coat, which also happens to have a normal Windows interface unlike bizarre "2.5D canvas" ZBrush and it has more straightforward and advanced brush tools than Freeform, similar to ZBrush. It's the epic output of a single programmer, and still lacks the magic of the ZBrush ZRemesher smart direction matching quad output, so useful for converting any mesh to smooth NURBS via smooth T-Splines conversion.
http://3dcadjewelry.com/forum/content.php?204-Boolean-and-Smooth-Rh...
Freeform only two years ago enabled the always present *second* stylus button after a decade of it being a dummy button, freeing your hand from a keyboard rotate key, and pathetically still lacks a flatten brush, so I'm not very impressed with their team. As often as not the tiring 3D haptic arm fights me instead of aids my workflow, and the internals of each seemingly simple command are usually extremely uncooperative and arcane like they are all alpha release versions retained for years.
Animators are moving to 3D-Coat from ZBrush to be able to sculpt in voxels, free of the usual mesh topology chess game and ruinous facet stretching. ZBrush instead now offers various auto-remeshing schemes and a specialized brush to somewhat solve mesh stretching into crude facets, adding with ZSpheres to rough things out minus mesh topology. Neither one offers elegant subdivision modeling like T-Splines or MODO though so have been described as dealing in amateur hour "mesh soup."
Free Autodesk Meshmixer can also create solids but wasn't well behaved with these large arrays of round capped tubes, quite confusing actually too.
The ability of voxels in 3D-Coat and Geomagic Freeform to do outward surface offsets to bulk up volume is extremely helpful so I don't have to regenerate these pipes to compensate for volume loss during overall smoothing. ZBrush only seemed to have inflate which caused kinking self-overlap from valleys.
On second try, Rhino itself is quite well behaved offsetting the original individual mesh pipes to make them fatter as needed, so the slow Grasshopper step doesn't need to be done again just to tweak the design. Self-intersections get worse but the voxel and smoothing stage fixes those automatically.
The ZBrush Dynamesh system includes a Boolean union feature, automatically, but it failed to be reliable for weaves that had very close surfaces that confused it enough to cause unwanted threads of mesh crap between them in bad fashion, when it worked at all by using a high but slow resolution.
Materialise Magics command Shrink Wrap is OK, though slow, but the overall software isn't very friendly about smoothing or subdivision, and is quite expensive, not a good option for clients or a team on a tight budget.
Hi Nik,
I would like to ask if you use ZBrush only for this type of work or you also sculpt something there. I ask because I tried to learn ZBrush and goooooood the controls were awkward and no way to change bindings to your liking (aliases in Rhino is such an amazing thing, my top favoutrite combination is GH, sooo nice to have these letters sit next to each other, fate? XD)
I lately see ZBrush as a one trick pony accessory to make perfect quad meshes for Rhino T-Splines. I already own Geomagic Freeform though, but am actually moving towards much more Autocad Meshmixer as a series tool. It's just that it currently lacks background image import as a sculpting guide, whereas Freeform is akin to an engineering program that can not only do images but guide curves too. But there's also 3D-Coat and that also smoothly combines voxels with meshes, and has a normal Windows interface. ZBrush lacks 3D mouse support, being made for use with a tablet.
Meshmixer has a $79 video coarse available, to make up for the lack of a current manual:
http://education.honeypoint3d.com/courses/3d-printing-and-3d-design...
Meshmixer can also heal holes smoothly, unlike ZBrush or 3D-Coat, which is useful for blending things together simply by deleting faces and reblending. 3D-Coat has dynamic remeshing but it's not at the level of Meshmixer. Both ZBrush and 3D-Coat are mainly targeted at game and animation designers with a few jewelers thrown in. Freeform does work with a normal mouse finally but you can't buy it that way, so it's still about $8.5K plus maintenance fees of $2K a year for updates.
There's also Blender but that's mostly kids, and I never found it fun to use since the interface is so modal and also prone to sudden chaos in interface layout I can't fix.
Meshmixer beta updates:
Welcome to
Grasshopper
Added by Parametric House 0 Comments 0 Likes
Added by Parametric House 0 Comments 0 Likes
Added by Parametric House 0 Comments 0 Likes
Added by Parametric House 0 Comments 0 Likes
Added by Parametric House 0 Comments 0 Likes
© 2024 Created by Scott Davidson. Powered by