7 replies to this topic

[krafty]

Hello,
I am trying to experiment with generating meshes out of vertex information only, is anyone having any ideas how one should go about it ...
There are a few techniques like generating a parametric surface and then triangulate ...

[Sol_HSA]

Connect the dots..

Without any other information but a bunch of vertices, you can end up with just about anything.

[krafty]

correct, but there are a few techniques that can generate a good enough surface ...

[krafty]

actually there is a lesser chance of landing up with an unexpected object, since it is going to be a surface

[geon]

If your data is pretty smooth, you could try starting from a random (reasonable) triangle and connect all edges to one more point, with the constraint of being close (decided by the distance and a threshold?) and g1 continuous (decided by the angle between the new triangle and the old?).

[Dim_Yimma_H]

Have you taken a look on the marching cubes algorithm for generating a surface of volumetric data? Here's a link to a description by Paul Bourke: http://local.wasp.uw...try/polygonise/

[z80]

Marching cubes is best when you have a clear definition of inside and outside for a given point i belive.

I think the term for "vertex information only" is a "point cloud".. Should make it easier to search.

Also this one might be helpful:
http://www.cs.sunysb...man_Vis2005.pdf

[krafty]

Yes,
great I was just about to post about Marching cubes and Marching tetraheds algo, its very good.

I am trying it out... will post if it works out good, another way btw, is Voroni, Delauny triangulation something on similar lines to what geon was saying ...

but Marching looks good at the moment

cheers