Preliminary results


Root  Previous item  Next item

When I'm writing this, the plug is only one day old and there's little to say, however the basics are up and running as you can see here:

DelaunayPreliminary_01
A screenshot of Rhino with a collection of point samples. As you can see they are
the result of several distortions applied to an ordered set. The result is anything but ordered.

DelaunayPreliminary_02
The plugin has created a mesh surface that connects *all* samples without
holes, overlap or flipped faces. But since the input was rather coarse, some
extra steps are needed to get a smooth surface.

The sample points have been hidden and the mesh-points (except the ones
along the edges) have been selected.
DelaunayPreliminary_04
After a few smoothing operations, the result is a smooth surface that approximates
the original point set rather well.




If we start with a more ordered sample collection like below, the plugin output is
almost directly production ready:
DelaunayPreliminary_05
50% of the samples were deleted at random to prevent a completely ordered
input. Samples with regular intervals will result in a grid which is boring.
DelaunayPreliminary_06
The generated mesh surface. Basically it's finished but for the awkward
faces along the edges of the mesh. A simple -_Smooth distributes the
vertices towards a better balanced set:
DelaunayPreliminary_07


Plugin features as I'd like to implement them:
·Input from pointcloud objects and perhaps mixed input.  
·Multi-threading (very important, with multi-threading you will be able to continue working and even alter your points, while the triangulation algorithm chips away in the background.)  
·Thin triangle filtering. (to prevent odd faces along the edges of the mesh)  
·UV-delaunay triangulation using a Nurbs surfaces UV-space.  
(the internal algorithm is already UV-based, but there is no interface yet)
·3D-triangulation algorithm. (so far there is only a 2.5D algorithm)  
·Automatic smoothing of internal faces.  

 

For additional information.