algorithmic modeling for Rhino

Delaunay mesh on different it possible?!?

Hi guys!

I want to use delaunay mesh component on different surfaces to recreate a caotic triangulation. 

The problem is that I can use delaunay only for one single this means that i can use lots of time this command for each surface and specially, I have to selection the final point to each triangles to join with other delaunay surface...

do you know a best method and specially faster to use this command on different surfaces?!? 

thanks guys

Views: 2752


Reply to This

Replies to This Discussion

anybody has a solution guys?!?

I'll do something for you but give some clue(s):

You want to create a  delaunay mesh in a List of surfaces? (or a Tree) yes? (and why this is "slow"?? - you said: lots of time this command for each surface ...). Or maybe you have lot's and lot's of points in mind?

Points (Lists) on a per surface basis should been different? (by count).

If yes, by what "criteria" these must being connected? (what means "join"?) maybe randomly?

Or you want a random to random to random ... entirely chaotic thing?

PS: Is this a facade of some sort?

Hi Peter,

thanks for your  help. I've tryed to explain by image and files!

I want to use the command delaunay surfaces rotate on different planes and concatenated in order to have a single single mass.
As you can see from the photo, at the moment I'm using the command delaunay on each face and then, find using the "points in curves" the common points so that it appears one piece.
My process is very long and hard to achieve.
There is a much more simple to realize this idea by simplifying and speeding up the process?

as you will see from the attached files is a very long process to get to the desired solution.

I would like to create this skin in a completely random but with triangles of uniform size if possible.

For "slow" mean that it takes a long time to create a single mass so as I am doing.

I hope to have been more clear. I attach files too so you can see better.

Thanks for the help


OK, I'll prepare something for you tomorrow (hopefully fast). Your definition is slow because it requires Jack the Ripper (but I'm the right person for that, heads, legs, hands, GH components, whatever comes first: massacre with no mercy ... he he).

Q: points should  have a certain Z "noise" as well? (distance from surface(s))? in order to create some "3d" effect. Or the whole system is solely "decorative"? And what system could be that? Some steel members? Some "flattish" truss? Something else?

Q: Are you after something remotely "like" this attached? (meaning a lot of other factors should being taken into account: node restriction(s), closed Breps as vertical beams etc etc):

Q: If yes, what is your experience with GH?

more tomorrow.


OK, This is NOT for your case (yet) but clearly demonstrates what GH can do (as regards speed: it's real time even when the 100+ blocks are invited to the party).

1. If you want "some" humans to populate the "base"  ... load Rhino file fist.

2. This greatly reduced demo creates a Del Mesh "in real-time" (it's a very complex project that I'm after these days: 10 C# scripts work together to do ... something, he he) AND finds convex hull points as well, So it works "twice" but still is fast. In fact it's about 15 times faster than the very same thing that I did with Generative Components.

3. Again: NOT for your case but good fun (I believe). Obviously doing this for a couple of surfaces of yours ... it could yield also a real-time solution.


loooool!!  sorry my definition is very caotic! :D

A: yes, points should have a Z noise to create a 3d effect. basically, it's a furniture and external part created by Delaunay it's a skin in steel. Behind, will be the flat surface always in steel. It's an object to realize only in steel.

A: no, basically I want to create something like a rock with many spikes. So, basically, with any flat form, I want to cover it with delaunay system to triangulate my furniture.

A: I'm using GH  from 6 i'm beginner.

AMAZING your definition in GH with incredible!!! yes, i'd like to obtain something like that!!!

thanks very much peter!!! :D

Furniture you said ... that greatly reduces the "constrains" as regards the nodes (if this was a building ... then ... anyway forget it, not our case). So this is some kind of folding (or not) panneaux? 


1. We need some random collections (pts) on a List of untrimmed surfaces. Homogenous "random" distribution is better than the method used in the Dubai abstract sketch attached (but at some CPU cost, not a big deal since pts a few: David Rutten did some very good jobin the GH component on that "even" matter). Adding random "noise" (with user controllable min/max values) is the easy thing: like the 256 does. 

2. We need some "similarly" defined random pts applied on each surface edge.

3, We "join" 1+2 and do the del mash thing (that's the easy part). Thus the "similar" constrain is critical here for aesthetic purposes (see 2).

4. For each subsequent surface (sharing common edge, that is) we repeat  by adding the pts related with the common edge.

5. Arriving to the last surface: we use either one common edge or two if the surfaces define some "closed" (in U or V) combo.

PS: Novice means nothing: we all are novices more or less: so don't let this "truncate" your aspirations.

OK: 5 minutes work (edge points NOT added, yet : it makes all the difference) just to indicate that GH can cut the mustard (time) rather easily. Test data (surfaces) are internalized (hopefully, he he).

I've deactivated any C# and added the simplest def possible (Z noise as well) using standard GH components (but I feel dizzy: I can't get "oriented" properly without code, is this an ominous disease?  he he)

Bad news: maybe in the final thing (with a myriad of options as ... er... tradition dictates)  I'll switch to C# (is this an ominous disease? he he)


Added some "edge" logic.

Very naive: we need another approach for properly "fuzing" edge random points with the main collections - IF Z noise is added). In other words: create "frames" in the surfaces and allow some smooth transition from Z noised points to the "flat' perimeter ones. Can I switch to C# to do it? (clash detection comes in mind first among other things)

He, he


And "finally" (just before removing all these and doing the highly politically incorrect thing for this Noble Forum: only one component that is not a cluster, he he).



many many many many thanks peter!!! I've discover lots of things today!! :D:D:D


Do you want the wild option? (for real men who fear no Evil - me what else?, he he).





© 2019   Created by Scott Davidson.   Powered by

Badges  |  Report an Issue  |  Terms of Service