delaunay meshes from (poli)surfaces

hello,
i need delaunay meshes from (poli)surfaces with compound curvature for simulation cloth in max or maya.
can anybody think, if and how this could be solves with grasshopper?

any comment would be appreciated.

thanks in advance roberto

Replies to This Discussion

Permalink Reply by Vicente Soler on August 27, 2009 at 3:50pm

oops. The horizontal surface has less point density because i calculated the point density based on the trimmed surface's area, but the points are placed along it's entire uv space and the filtered out. To fix this use the untrim surface component and calculate the areas from these.

Permalink Reply by roberto gagliano on August 28, 2009 at 12:19am

hi vicente,
thanks a lot for your effort!
for lack of time i could not entirely scrutinize your definition, especially as it seems quite complex and i am really a novice to gh.
but please let me add some comments:
the mesh looks delaunay but the triangles differ very much in size and aspect ratio.
lately evert pointed me towards a delaunay definition on the cgal.org-site, which guarantees a more homogeneous distribution of vertices, which is crucial for a realistic cloth simulation. basically it states that within a circle that connects 3 adjacent vertices there lies no other vertex.

Permalink Reply by michael sprinzl on August 28, 2009 at 1:43am

I think this could lead to a mesh for calculation purpose. I tried to relax some points. Maybe this helps. ( no idea )

Attachments:

relax.ghx, 234 KB

Permalink Reply by Vicente Soler on August 28, 2009 at 5:16am

The algorithm is still delaunay, if you have the point distribution of that image over the brep, the second example would be created. The problem is the point placement in the first place, that is completely random. As i said, maybe try subidividing the surface in a grid of points, and then randomly move those points just a small distance so it's still an even distribution.

Btw, why is the bad example aliased and the good example antialiased? Reminds me of those infomercials where the guy using the wrong product is always uglier. :P

Permalink Reply by michael sprinzl on August 28, 2009 at 5:40am

... maybe look at the script...so you get it

its just a idea .... I try to find the nearest points and recalculate the original point. sorry that i did not try this on top of a surface. So my first trys are only 2d.

...and by the way...
look at 2D Range and Neighbor Search at www.cgal.org
kinky edges are bad for calculation...

Permalink Reply by roberto gagliano on August 28, 2009 at 5:57am

hi vicente & michael,
i cant really contribute technical input here.
just want to post a screenshot to illustrate ideal vertex distribution (for my purposes).
its a screengrab from garment maker/3dsmax. the density (iterations) is controlled parametrically.if you count the vertices on the circle´s perimeter, you will find, they correspond with
the density-value (highlighted in red)

i also posted this in the rhino-newsgroup, where i started a similar thread some days ago. maybe you want to see it, david is also there.