Compass Method applied to a mesh

[Elena_Casolari]

Hi! I am looking for a component to generate a grid through the Compass Method by Frei Otto on a mesh. I found already a custom component for surfaces, but I really need to apply the process to a mesh and I am not able to write it by myself. Any suggestion is more than welcome

[PeterFotiadis (Pfotiad0)]

Like this … but using a Mesh?

Frei Otto's Compass Method - structual engineer

BTW: 1 to 10 what is your level in C#? (if C# is your game). And … are you fully familiar with Connectivity Trees?

BTW: What is the goal? provide some sort of sketch. And given the opportunity do you know what K2 is? (kinda a physics engine, so to speak).

[Elena_Casolari]

It is exactly what I was referring to.

My level of C# is zero and I don’t know what are Connectivity trees.

The goal is to define on the mesh the orientation of the warp and weft directions of a textile material. In the image you see the starting point and the starting curves. I don’t need to remesh, I just need to analyse the inclination of the mesh edges with respect to the warp and weft.

If you mean Kangaroo 2 yes, from there I tried to use the WarpAndWeft component, but of course it just takes the edges of the already existing mesh.

[PeterFotiadis (Pfotiad0)]

Well … I could - theoretically - write a line or two of C# on that matter (job almost done exploit link below > see getAllGridPoints Method and imagine doing this using M Vertices and VV Connectivity). But that would be a black box for you (meaning more or less a useless thing).

So your best option is some good Samaritan who likes components (I don’t).

BTW: What you are after is called a Cheby Net:

Generation of a Chebyshev net on a surface

I’d like to share my GH Def / Python code for generating a Chebychev net over a given surface following the algorithm by Popov[1]. A Chebychev net can be seen as a quad mesh with all edges being the same length. The only things that change are the internal angles. chebychev.gh (104.2 KB) It’s my first time scripting more than a couple of lines, any comments on how to improve the (probably inefficient) code would be most welcome. [1] Popov, E. V. (2002). Geometric Approach to Chebyshev Net Gen…

GitHub - AlanRynne/ChevyshevNet-GH: Grasshopper component for developing a...

BTW: I can barely see why not use K2 Methods (driven obvously via some C#) on a steady or evolving Mesh : it’s just a matter of appropriate Goal(s) - A Chebychev net can be seen as a quad mesh with all edges being the same length … so if the Mesh is out of some Surf one should extend that (acting as a template) for the OnMesh Goal to work,

[Elena_Casolari]

Yes I found that nice paper about Chebyshev nets, but I am not really able to modify the component.

I will try the Kangaroo edge length + onmesh, thank you very much!

[PeterFotiadis (Pfotiad0)]

Your main Goals : OnMesh, EqualLength (use diagonals as well in order to control - up to a point - the quad shape). But as I said take care of what in fact means an “expanding” collection. Plus the Anchors policy is a bit tricky (try Plastic Anchors).

Play with a tri Mesh as well (not a C Net by the book - but who cares?).

Tip: In general remember that K2 when fitted with bananas … bananas is what it would deliver. This means that depending on Topology/Goals/Values … you may (or may not) get a solution (life sucks).

For intance (general case) : assume that (a) you have a “suitable” Surface (meaning that the expansion in N/S/E/W yields a non “freaky” result) and (b) you have some sort of ortho/diag/other pts/lines Grid (no need for a Mesh since the C Net is a Graph). So … while the template to follow is OK … If you set as static Anchors all the perimetric pts (or pts per side pairs [N/S - E/W]) … then maybe you’ll get bananas (IF and only if there’s no space for the relaxation). And what happens if the Surf template is closed in U/V? You tell me (life sucks).

[DanielPiker]

Here’s an example of generating a Chebyshev net with Kangaroo

Hanging mesh / Similar meshsize

To make a quad net with equal edge lengths (also known as a Chebyshev net), it is easier to approach by draping a full grid then cutting off the parts beyond the plane: drape_sphere.gh (8.1 KB) [image] Cutting a flat mesh to a circle then anchoring the points around the boundary as you were doing will result in crumpling, because it prevents the angles from changing as they need to to form a smooth curved surface.

[laurent_delrieu]

And this one

Microphone Mesh

Hi all! I decided to write this post because I want help with something about “how-to-model”. See the reference picture: [replacement-2pcs-lot-wireless-microphone] I found a solution using displacement maps but first, I tried some ways to model this in Rhino without luck. Any ideas? Thanks, Luis

[PeterFotiadis (Pfotiad0)]

@Elena_Casolari

I will try the Kangaroo edge length

In fact I discovered a C# written a zillion years ago that yiels a C Net Graph (or Mesh). Using a template Surface It does a K2 relaxation within a Loop - meaning that the U/V divs are “automated” (especially if the Surface is closed in U or V) since it attempts to find an ideal (kinda) solution taking into account the rational expansion of the template (for the OnMesh Goal). All that IF we talking about an expansion (depends on your target Length Goal).

Think the whole approach as a kind of MOO - so to speak.

However the whole thing has no meaning notably for AEC envelopes: why the edges [of the Graph or Mesh] must be equal? for fabrication? (pointless) due to some freaky rule? (pointless) due to aesthetics? (pointless).

The only rational reason to do that is to attempt some sort of clash free solution (related with, say, MERO KK sleeves, cones, tubes, cats and dogs).