Grasshopper

algorithmic modeling for Rhino

# Force driven metaball meshing

Metaballs are contours of a 3D scalar field. As an alternative to the typical marching cubes I am trying a force driven approach.
We can differentiate the function for the scalar field to get the gradient vector field, which can be used to push the particles towards a particular threshold value. The particles also repel each other, causing them to distribute evenly across the implicit surface. The particles are then connected according to proximity, the triangles in this graph are found and a mesh is drawn (Still a few issues with this step).

Views: 1881

Comment

Comment by ĐỒNG VIẾT NGỌC BẢO on June 8, 2012 at 4:20am

my graduation project use metaball, i make surface by create point (divide curve x,y,z axis)  and the mesh result was not good.I think your way to create mesh surface so great. so, could you share this definition? my email: dongvietngocbao@gmail.com. thanks !

Comment by Hrvoje Petrovic on October 10, 2011 at 1:25am

Great!

Comment by Daniel Piker on October 9, 2011 at 2:30am

Jacek - It's certainly faster than my attempt at MC. It still takes a bit of tweaking to get decent particle spacing, but I think potentially this could be pretty fast.

I can't take credit for coming up with the idea of this approach - the first place I think I saw it was this paper by Heckbert and Witkin:

http://www-2.cs.cmu.edu/~aw/pdf/particles-reprint.pdf

some other references here:

http://http.developer.nvidia.com/GPUGems3/gpugems3_ch07.html

http://zurich.disneyresearch.com/~wjarosz/publications/particles.pdf

http://www.cs.utah.edu/~miriah/publications/particle_sampling.pdf

https://wiki.engr.illinois.edu/display/graphics/Wickbert

Comment by Pieter Segeren on October 8, 2011 at 1:39pm

Way cool! Even wíth the gaps for now.

Comment by Daniel Kautz on October 8, 2011 at 12:48pm

Great idea

Comment by Astrid G on October 8, 2011 at 11:52am

Amazing!

Comment by Andy Payne on October 8, 2011 at 11:18am

Very nice!

Comment by Jacek Jaskólski on October 8, 2011 at 9:25am

Brilliant! (as usual)
It's much faster than MC, right?

Comment by Daniel Piker on October 8, 2011 at 9:21am