algorithmic modeling for Rhino
I have a model made of a poly mesh that I would like to convert to quads.
Converting to quads is needed, because I'm also using T-Splines (TS) and TS requires a quad mesh.
I thought the Quad(quadrangulate) function was going to do the trick, but I'm missing something.
I'm not a "poly" or "quad's" expert by any stretch, and I realize there's allot to converting. Saying that, is it possible in Grasshopper to accomplish this function?
I've attached a screen shot of my model to make sure I'm describing my model correctly. GrasshopperQuestion.JPG
There's a good webinar over at T-Splines about converting meshes. The point is: Your mesh looks pretty irregular. It's probably hard if not impossible to convert it to a workable quad mesh.
Thank you Hannes,
I'll watch the webinar. I have a feeling I'll be buying another program.
Dang, I really wanted to use GH.
Looks like some kind of a mesh object created from laser survey or point cloud data... Just a guess.
You may need to reverse engineer your mesh in NURBS (but that will depend on your mesh or source data) and then remesh (if you need to)...
I've use a trial version of some of these stand alone Resurf programs when the need arises (like below). Good luck.
Its a model from a photo that a 3rd party made for me. I should, and will in the future, want a quad output. I know Zbrush and some other sculpting programs convert poly's to quad's nicely.
I'll have a look at those rsurf programs.
The best soft for you is EVOLUTE
Automatically remeshing a surface with quads is tricky. (It can be done, but it is cutting edge stuff. See this paper for one approach : http://geom.mi.fu-berlin.de/publications/db/KNP07-QuadCover.pdf)
However - if you don't mind doing a tiny bit of manual setup, then an easier option might be to shrinkwrap your geometry.
So you start with a coarse quad mesh that very roughly approximates your surface (this might be just a simple box, sphere or plane)
Subdivide it (using WeaverBird) until you have the number of faces you need
Then pull that mesh onto your surface, and to keep the quads from getting too uneven and distorted, apply some smoothing or relaxation of the edge lengths or angles.
I'll post an example later of doing this with Kangaroo.
Thanks Igor and Daniel.
Daniel - Thanks for paper. I probably won't try anything, but I sure enjoy reading material like that. Makes me understand the problem allot better.
Shrinkwrap, that's a good thought too...
If there's any flip milling involved (for a 3d object) there may need to be a way to shrinkwrap 2 halves of the whole.
Or maybe shrinkwrap a pill shaped mesh (cylinder with spherical caps) to a 3d object.
...jus' thinkin' out loud.
I dont know how exactly you make this coarse aproximation( which is marvelous), but I though about some kind of "anchor" points (where subdiv and shrinkwrap is applied) based on 3d convex hull of initial mesh ("anchor" points are those which create the hull)... this would presereve details like ears ;)
The course approximation is can easily be modeled by hand from basic shapes...
Manually is probably the quickest way to do it.
I agree. Tsplines makes it very easy.
My problem is, once I have a rough quad plane (I just want to drape a plane over a mesh), what's the work flow for moving/applying/whatever the plane surface to the mesh?
Converting tri vert surface to quad is mind boggling. Sheeeez :)