Dynamic remeshing script with Plankton and Kangaroo

Replies to This Discussion

Permalink Reply by ng5 Alex on January 8, 2014 at 3:21pm

Hello 

does this need the latest rhino 5?

Permalink Reply by Mathias Gmachl on January 9, 2014 at 8:37am

Wow, just got this up and running. Will experiment and feedback asap.

Hope you are well!

Permalink Reply by Mathias Gmachl on January 10, 2014 at 7:52am

Would it also be possible to add an align force to the remeshing to improve the regularity of a mesh? 

f.e. in the attached image the edges feeding the exoskeleton mesh generation could be used to align one of triangular mesh sides, so that irregular vertices are pushed/limited to the nodes.

Permalink Reply by Daniel Piker on January 12, 2014 at 10:34am

Hi Mathias,

Yes, alignment with curvature directions would be a very nice thing to add to the remeshing.

In fact, the edge flipping in Plankton is written in such a way that it also works with quads (it is really a sort of edge rotation) and I had been thinking that such an alignment force could be used for quad remeshing.

With triangles though there might be other ways to reduce the number of irregular vertices by allowing the edge lengths to vary more...

Permalink Reply by jjsolly on January 13, 2014 at 8:19am

Daniel. I seem to get an error regarding my RhinoCommon version from the components in those examples. Any ideas?

I'm running Rhino 5 and GH 0.9.0064

I tried upgrading to 0.9.0066 but then the Mesh tab from the Kangaroo ribbon disappeared. 

Permalink Reply by Daniel Piker on January 13, 2014 at 11:59am

Which Rhino service release are you running?

Permalink Reply by jjsolly on January 14, 2014 at 2:04am

Apologies Daniel. I had missed that i was a release or so behind. Silly mistake to make. Thanks for the help. 

Permalink Reply by jjsolly on January 14, 2014 at 2:57am

Based on the Remesh Plankton circles example.. ignoring the circle packing part it seems like this could potentially be used to generate decent FEA meshes? i.e. an engineer could use points to define seed locations for denser mesh regions? This could be really very useful for those of us playing with the refinement of geometry within GH/Rhino to then link into full-blown FEA. 

Before I play around a lot more.. could there be a way to define a "general" target length and then a localised mesh defined by seed points? I guess there'd need to be another variable regarding the size of region affected by the seed? 

Permalink Reply by Daniel Piker on January 14, 2014 at 3:36am

Indeed, generating meshes for FEA was one of the applications I had in mind.

The target edge length can actually come from any function that takes a point location and returns a number - so yes, it would be possible to have some global baseline value and local deviations from this based on proximity to seed points, or distance from the boundary, or a grey-scale image, or surface curvature...

Permalink Reply by Harri on January 20, 2014 at 4:21pm

Nice example Daniel.  I added some csharp Plankton script to create an offset mesh "box" for each face.  A bit of subdivision and render and we have some nice looking blobs!

Permalink Reply by Andrea Rossi on January 24, 2014 at 3:14am

Thank you Daniel,

this is super useful!!!

One problem: I am trying to use the same script on a mesh with more than one loop of naked edges (simply, a surface with an hole), but the definition has problems in recognizing the two distinct edge loops, and so it uses them as as single one, connecting them with new edges and collapsing the mesh structure.Is there a quick way to solve this? I am not an expert in scripting with meshes...

I tryed to comment out from the script these two lines:

        if(P.Vertices.NakedEdgeCount(i) > 0)
        {P.Vertices.SetVertex(i, boundary[0].ClosestPoint(Point));}

and managed to avoid the collapse of the whole mesh, but of course the boundaries of the mesh are broken. Any idea on how to solve it?

Thank you!

Permalink Reply by Daniel Piker on January 26, 2014 at 10:42am

Hi Andrea,

I did develop a workaround for this problem - if you input a list of boundary curves, and then for each naked vertex, get the closest one from the closest points on each curve in this list. I'll post the script later.

(This still isn't an ideal solution - generally I'm trying to develop a way of keeping mesh boundary edges associated with particular edge or feature curves (such as sharp creases) given as input. However this needs to be integrated with the topology operations in Plankton itself - so for instance if a mesh edge on a feature curve gets split, then the 2 new edges also become attached to the same curve.)

• ‹ Previous
• 1
• 2
• 3
• 4
• …
• 8
• Next ›
• Page