meshing

Hello everyone. I wonder if there are some mesh wizrds here.. I am looking to create a mesh that goes from fine to coarse within a given frame. as you can see in the pictire, I have a fine mesh, and I want to create a new mesh that is merged to it in the large triangles. Thing is the triangle should be coarse in its middle.. so there is some algorithmic logic here..

Attachments:

mesh.jpg, 117 KB

Replies to This Discussion

Permalink Reply by peter fotiadis on January 16, 2016 at 1:00am

Hi (long time to talk).

You have some kind of "canopy" (geo dome or other), yes? And that engulfs panels, yes? If so (a) why meshes? (are you redirecting things for FEA?), (b) why "merge"? (instead of separating LBS members from envelope members).

Or maybe you work with some subdivision modeller? (MODO, Max, Maya and the likes).

Other than that what you are asking is rather easily doable ... but it could be better to upload some "sketch" with regard the big picture.

Permalink Reply by kipodi on January 16, 2016 at 8:22am

hi Peter,back on the horse :)

You are right, its for FEA .. I am uploading 2 sketches to explain the problem.

The picture from further away shows the whole geometry. the lines highlighted in green are the are the closeup picture. I'm trying to create a merged mesh, but dob't need so many nodes on the triangle surfaces..

Attachments:

mesh2.jpg, 68 KB
mesh.jpg, 148 KB

Permalink Reply by peter fotiadis on January 16, 2016 at 8:43am

FEA eh? Hmm ...

Mail me some stuff (if it's sensitive don't bother: I'll sell the idea of yours to others and I'll give you a decent 0.12% [max]). Hope that you haven't lost my e-mail.

And this is a self supporting panel with perimeter reinforcement "ribs", yes?

BTW: In general since you can have the naked edges on hand (either via code or via some component or add-on) ... well you understand what follows (most specifically: a tri-mesh for your case or some other mesh for other cases > then mesh join etc etc). But ... well ... it's not that simple (have you ever worked with FEA in the past?).

Permalink Reply by kipodi on January 16, 2016 at 8:56am

Yes, I 'm working with abaqus. its possible to create an adaptable mesh but only if you define edge to edge.. its not possible to give an outline of nodes and ask to remesh the interior surface in an adaptable way.. I thought maybe kangaroo with mesh machine.. or maybe there is a polygon editing software with a brush to reduce nodes while rebuilding the mesh off these nodes.. the triangle is planar so it dhouldn't be impossible..

Permalink Reply by peter fotiadis on January 16, 2016 at 9:55am

Indeed ... that's why I've added the "...well..." comment. But it's rather very easy (but I'll do it via the Dark Side - you know that already eh? he he).

BTW: you want a policy that "fills" the gaps? (there's some "deviations" from straight stuff in your pics)

Moral: Mail something.

Permalink Reply by kipodi on January 16, 2016 at 11:22am

mailed. see you on the dark side.

Permalink Reply by peter fotiadis on January 16, 2016 at 1:00pm

Hmm (red circles are where the Dark Side found some missing stuff)

Lot's of possibilities: this, this or that? that's the 1M question.

From the stupid one (in order to indicate the "hard" work done and rise my fee [see below]).

to the less stupid ones:

Note: I'll charge you extra for that one (I hate meshes, he he): at least 34 cans of the finest sardines - Da Morgada - known to man (in pure mineral oil).

PS: what about the "ribs"? where are they? what shape? how many? (we can do a very controlled meshing with code).

more soon.

Permalink Reply by kipodi on January 16, 2016 at 3:18pm

you really are form the dark side!

very nice, I think something like this can do the trick. Is it possible to have more of a gradient from fine to coarse?

p.s: how about quads? they can't work from just a point cloud like delauny I assume..there needs to be some extra sauce in that..

and for the million dollars, do you have an idea of how to export that mesh from rhino? I couldn't find an appropriate format

Permalink Reply by peter fotiadis on January 17, 2016 at 12:29am

Well ... in an ideal world this is actually an "inverse" fractal logic problem (by-passing dalauney completely) :

You start from the perimeter points and then you combine user controllable "sets" (say 3,4 or something anyway) with some "next" (inwards) node (derived from some proper division VS the offset perimeter) and repeat until you arrive into a "core" triangle. A controlled de-triangulation so to speak. This could yield very uniform meshes (if this has any meaning).

Recursion in plain English (peanuts via the Dark Side) where a method (i.e. a thing that does something) calls itself (quite freaky I confess).

BUT ... well ... in order to apply this approach on the ribs as well (or the lip reinforcements if I got it correctly) AND by-pass any brep > mesh "auto" method I need the ribs: what is their shape? (i.e. profile).

Obviously I'll charge you extra for all the above (more yummy sardines - what else?).

Quads could complex things for no reason (and rise considerably the N of sardines).

Export to Abacus you mean? (STEP214 ?? that in theory can yield some assembly component schema: nested blocks and the likes).

Permalink Reply by peter fotiadis on January 17, 2016 at 12:41pm

Get some test meshes ... but the new policy is totally different - I just haven't found the time to write the finishing touches - spend the whole day waiting for some proper wind [windsurf] > nada + null + void + nothing + Beaufort 0.0001 > broken nerves.

Yesterday was gale day, mind. Moral: Karma

Some aspects of the new policy: there's 2 levels: the first is the shown where a rather complex logic samples the points "into branches". Then according the inner level of divisions further de-triangulation occurs until we meet a "core" triangle.

Kinda like doing a fractal tree the other way ... he he.