funcky quad mesh generation

Dear comunity,

lont time I dont post a question, so here it goes!

I'm looking for an easy way to produce an irregular quad mesh similar to the one in the image. Any tips will be most appreciated

Attachments:

photo_2017-04-19_11-04-39.jpg, 344 KB

Replies to This Discussion

Permalink Reply by Andrew Heumann on April 19, 2017 at 8:40am

here's something that might be a start at least.

Permalink Reply by Pep Tornabell on April 20, 2017 at 1:57am

easy, thanks!

Permalink Reply by Anastasios Kokkos on April 19, 2017 at 9:51am

Maybe try starting with a normal quad mesh and give random length values for each edge in Kangaroo2? (example file) or if you have the final vertices try pulling the initial quad mesh vertices towards those?

Attachments:

IrQuadMesh.gh, 25 KB

Permalink Reply by peter fotiadis on April 20, 2017 at 1:50am

There's 2 "fast" ways to do that (among others):

1. Populate randomly a surface (NOT using David's excellent stuff since this yields "uniform" points), do a Delauney (and/or remove undesired Mesh Faces) and then do a tri to quad conversion (requires code that is not very simple, mind : Google that).

2. "Agitate" (up to a point) a grid (say, due to a surf division or using some other source), do a quad mesh and then use recursively Maelstrom Morph on random Mesh vertices (for the Morph plane) and maybe random R1/2 values. Shown a similar Morph in one generation:

Permalink Reply by Pep Tornabell on April 20, 2017 at 1:57am

awesome!

many thanks

Permalink Reply by peter fotiadis on April 20, 2017 at 2:56am

BTW: Plan A requires complex code (the quad conversion): if you are inexperienced avoid even trying.

BTW: Plan B: Try it and if you hit the wall notify - but the solution (as captured) would be carried over solely via code (not a thing that you want I suspect).

Permalink Reply by Pep Tornabell on April 20, 2017 at 3:07am

I know conversion from tri to quad mesh is tricky and a great field of interest.

by the way, cool "TheBigMessageOfTheDayIsThis" output ;)

Permalink Reply by peter fotiadis on April 20, 2017 at 3:17am

Indeed it is and since this IS something that is used in real-life by pros these days (tri meshes for facades/envelopes et al are out of fashion: blame "progress") I doubt if anyone would be willing to share a working C# solution on that matter. Additionally (facades/envelopes) you'll need working packing solutions (the critical bit) ... that I BET that you can't find anywhere (even for a price).

Permalink Reply by peter fotiadis on April 21, 2017 at 12:31am

BTW: Staying on the quad theme here's some looped Maelstrom results on meshes [random planes but steady R1/R2/Angle] derived from surface division grid points. Quads for facades/envelopes have the advantage that if things go tough (planarity) you can easily switch to Plan B (a controlled triangulation by checking angles etc etc). Also for creating a rigid economical load bearing support structure (i.e. a truss) triangles (actually in 3d: tetrahedrons) is the way to go ... meaning that a "compatible" triangulation is always on demand.

Permalink Reply by Pep Tornabell on April 20, 2017 at 1:57am

thanks, I will give it a try

Permalink Reply by Anders Holden Deleuran on April 20, 2017 at 3:28am

Hi Pep!! All those 3 and 5 valence vertices makes this looks more like a collection of closed ngons that have been subdivided and then fucked up a bit to me. Quick example using Weaverbird for the meshing/subdivision:

Attachments: