adding thickness and smoothness to the curves

Replies to This Discussion

Permalink Reply by peter fotiadis on January 31, 2017 at 2:29am

Just spotted that one.

Get these 2 as an intro to the "get lines and make a thickened mesh" puzzle.

The first (the simplest) provides 5 modes to choose anchors thus yielding a variety of "vault" type of stuff. Then it works either with ExoW (black mesh) or IntraLattice (blue mesh).

That said ExoW is tricky: occasionally reports engulfing issues and stops playing the game. For instance in this (diagonal) anchor mode and with some U/V random values:

Whilst IntraLattice appears rather less temperamental:

The other def is more complex and works using the Proximity approach that makes more sense with regard random 3d line graphs (as an exercise: Add a gate and use IntraLattice as Plan B).

best

Attachments:

• Thicken_case2_V1.gh, 147 KB
• Thicken_line_graphs_23A_V2.gh, 140 KB

Permalink Reply by peter fotiadis on January 31, 2017 at 2:39am

That said ExoW yields aesthetically better (i.e. more "liquid") nodes .. if it works.

Here's the EMapped ExoW mesh (anchor mode: random) :

Note: Triangulation NOT used in the first example: obviously that could stress ExoW a bit more (with regard engulfing issues, true or not).

Permalink Reply by peter fotiadis on January 31, 2017 at 12:28pm

That said for this type of topology chances are that ExoW would protest (but sometimes it works - see next screenshot):

On the other hand IntraLattice (although works almost in most of cases) is not perfect (spot "twisted" nodes and the likes):

Permalink Reply by Anna Żuchowska on February 1, 2017 at 2:12am

OMG! this helped me a lot!! Thank you soo much :)

Permalink Reply by peter fotiadis on February 1, 2017 at 8:38am

Moral: ExoW is the Ducati (if it works is heaven) and IntraLattice is the Honda (it works but leaves you a bit cold).

Permalink Reply by Anna Żuchowska on February 1, 2017 at 6:48pm

I have another question to the "2" file. Do you know how can I make a mesh from a curves I attached to the mesh made in kangaroo? 

Permalink Reply by peter fotiadis on February 2, 2017 at 12:19am

You mean that you want the replace the lines (edges) with curves ?? 

For instance: get a given line define a perp plane at, say, 0.2-0.8 > get the x/y axis vector > multiply by some value > get a "mid point" and then interpolate [the line start, the" mid" and the line end].

Permalink Reply by Anna Żuchowska on February 2, 2017 at 3:06am

I meant from this curves

Attachments:

• Zrzut ekranu (334).png, 427 KB

Permalink Reply by peter fotiadis on February 2, 2017 at 4:44am

Ah that one:

1. You must split each curve VS any other (easy with C#: notify if you need a "standalone" thingy that does this).

2. Then use IntraLattice (but as we had seen: IL is not the holly grail of liquid things).

General approach:

1. Forget meshes.

2. Create a FLAT random collection of lines.

3. Get all the resulting segments (as discussed above).

3. Morph these into some user controlled surface (curves) OR use K1/K2 to create a vault type of stuff (lines). Don't forget to deal with the end/start pts [as in 2] as anchor points. Using K1/2 means: springs on the segments [as in 3] AND some unary force assigned to the free to move nodes (all minus anchors).

4. Use IntraLattice.

But the big thing is that IF this is some kind of AEC (or you are an Architect) I would strongly suggest to use lines instead of curves ... since in real life ... blah, blah.

If you are an Artist ... do whatever you want.

Permalink Reply by peter fotiadis on February 2, 2017 at 5:40am

And here's a rather stupid (and a bit kitsch) example on the split all with all theme:

Imagine a C# that does this "network" of random connections (with 6 different modes). Shown linear ones but you can have curves as well.

Then another C# makes the pieces due to Curve to Curve intersection events (all VS all or a collection VS another or some other options). Then the pieces + unary forces + anchors (no mesh around) are treated via K1 (we should use K2 for that ... but anyway) and a kitsch vault is made.The fact that there's no triangulation ... makes it more pointless (planarity + beefier nodes + ... > further and further away from any engineering neatness etc etc).

Obviously with some interactive ability (C#) to define specific anchor points "within" the middle of the network ... we can transform the kitsch to wow in no time ... meaning that the possibilities are endless.

If the network is made via curves we need some target surface to morph it and then use IL.

In fact these things in real life are made via some proper system the likes of MERO KK or some tensegrity truss or some complex hybrid involving planar glazing etc etc ... but that's another story.

Permalink Reply by yehezkielwm on February 6, 2017 at 1:29am

You can also use weaverbird! I got the image below from pinterest 

Permalink Reply by peter fotiadis on February 2, 2017 at 12:02pm

BTW: Did some tests (flat random curves > split all > map to some Surface > IntraLattice).

For instance (all Curve segments are split at any Intersection Event):

Problem is that this type of random approach can hide nasty surprises (unpredictably small pieces that make IL life rather difficult):

And if we are out of Karma (the norm) this is what IntraLattice does (an invalid mesh plus some oops stuff: all "mesh portions" starting from 0,0 [rather unwanted, he he]):

Plus "some" very ugly things here and there:

Moral: I hate meshes.

• ‹ Previous
• 1
• 2
• Next ›
• Page