Curve Crease_Form Finding 2 surface with two different curvature

Hi everyone!

I am starting developing a personal project wich explore curve crease.

Online i found some example at http://www.curvedfolding.com/profiles/blogs/grasshopper-kangaroo-fo...

where they used a simplify code from the kingkong plugin to create curved crease.

What i am trying to do is ricreate something like that but using two different geometry with two different curved.(One fisical model explain better what i am trying to achive in the img. Planar and Curved).

What i try until now is try to use a different approach.

1)Start with the two surface, transform it in mesh and use the component spring from mesh to activate the internal force.

2)divide the edge with the 2 different curvature that i want to join and connect the points trough lines that they will be the springs.

3) Connect everything to the kangaroo component

The problem is that the geometry does not react the way i want.

Hope to catch your interest, i will be very happy to have some help on this problem.

Thankyou very much

Attachments:

Planar.JPG, 75 KB
Curved.JPG, 74 KB
Test_1.gh, 25 KB

Replies to This Discussion

Permalink Reply by peter fotiadis on October 2, 2015 at 10:40am

Hmm ..

Let's reload the Matrix:

1. Since these "segments" pictured appear as "surfaces" (i.e. not trimmed stuff) how about some wonder/mysterious capability that could allow you to modify on-the-fly the curve control points (on a per item basis), make (at will) some nurbs (sweep 2 , network etc), modify on-the-fly the nurbs control points (on a per item basis)... and store (at will) all these as a x variant? I.e. making collections of possible solutions "stored" in some parameter (or several) as persistent data (kinda like internalizing a tree that contains per branch a given variant of that thing of yours).

2. If 1 rings any bell ... its rather easily doable but only via code.

Permalink Reply by Luigi Olivieri on October 2, 2015 at 11:02am

Thankyou for your reply, but I'm sorry i didn't really understand what you trying to say.

Your process seems to manual, even via coding, rather then trying to simulate the behaviour with kangaroo.

but thankyou anyway

Do you have any solution using kangaroo?

Permalink Reply by peter fotiadis on October 3, 2015 at 12:18am

Manual is the way to go when attempting to sculpt/"shape" an artistic like form (or many). Kinda like doing this with Modo 801 (and enabling the "trace" steps capability).

BTW: if a form is due to a "simulation" or not who could distinguish the difference? BTW: if a form is serving as an envelope of some building (kinda like the Bilbao Guggenheim thingy) then ... lot's of other factors (let's call them "goals" in the K2 way) should constrain the solution.

Anyway I'll post soon a K2 based approach (and maybe the other way, maybe not).

Permalink Reply by Luigi Olivieri on October 3, 2015 at 4:18am

In fact what i would like to work on with kangaroo, is also integrate the material propieties in the form finding process.
Maybe using karamba?
Do you think K2 is more indicate to simulate an active bending behavior?

Well we will wait your example that im sure it will be very helpful for other people other then me. I am very thankful.

Permalink Reply by peter fotiadis on October 3, 2015 at 11:31am

Well ... I have good, ugly and ugly news.

Good: I have a (rather strong) feeling that you've misinterpreted what I was suggesting. These are good news because after reviewing some related stuff of mine ... I've discovered that most are classified as internal ... meaning that I should re-make some for this case ... meaning: yikes.

Ugly: this IS not the way to approach this problem: what are you going to do when the surfaces are not co-planar? What is the condition of "stitching" them? (the common edge, that is).

Ugly: The def attached is WIP in the sense that a lot more checks are required in order to locate the neighbor edges automatically. Additionally it doesn't work with collections (pairs of surfaces - that's the V2 (near ready) does).

I hear you: why the C# at the start and the other inactive ones? Because of the V3 (almost ready) my friend, he he.

Moral: a pile of worms, what else?

Attachments:

K1_join_surfaces_V1.gh, 142 KB

Permalink Reply by ng5 Alex on October 3, 2015 at 11:56am

lol

Permalink Reply by peter fotiadis on October 4, 2015 at 12:03pm

in fact lol ^ 2

V2A with auto detection (U/V), anti-gravity(C)(tm), Big Brother, less components (if any, he he), more bugs, slower, uglier => big progress.

more tomorrow

Permalink Reply by Artyom Maxim on October 3, 2015 at 12:56pm

Hi Luigi,

You can see your problem on the first image. The issue is that you haven't used the actual mesh vertices to create the stitching lines. Should be like on the image #2.

Ten there's another issue when you turn on the simulation, see #3. To prevent that you need to add some additional constraints to the mesh to keep its edge angles the same. A good idea would be to triangulate the faces of the mesh and add angle forces between all edges.

Good luck!

Attachments:

Test_1.gh, 29 KB

Permalink Reply by Luigi Olivieri on October 4, 2015 at 1:49pm

Thankyou to everyone! Hope you guys will share your definition, for everyone.
The work the everyone does is higly appreciated and very helpful for everyone.
I can only say thankyou.

Permalink Reply by Luigi Olivieri on October 5, 2015 at 5:03am

Hey everyone i achive to connect the two panel using Peter script, but trying to semplify a little

(img.#1).

The good news is that it bends, the bad is that the shape doesn't react exactly how i did in the phisical model, it deform on the side, that's because i think the spring applyed to all the mesh make it react almost like a fabric.

(Img. #2)

If some one have idea how to develop more, is welcome!

Attachments:

Test_2.gh, 24 KB

Permalink Reply by peter fotiadis on October 5, 2015 at 5:24am

Passion is the virtue of any Engineer worth his onions, he he

Get the V2A and have faith: V3 does a lot of things more: in fact does ALL the things.

Notes on V2A (the grey group is your target, the others are ... other WIP/help things, he he):

1. The main problem is to "match" U to U automatically because the scope is (obviously) to work with pairs of surfaces sampled into DataTrees. V2A does this with a clumsy way.

2. The other problem related with 1 is to find the prox edges (on first sight "simple"). V2A also does this with a clumsy way (see case 3).

3. Manual control is available if you set autoTranspose = false. WATCH what the various panels have to say.

4. The "bend" is achieved via "gravity" (watch the related panel): just another option.

5. V3 may switch to K2 AND C# entirely meaning no components at all ... even the K2 related ones: these are the bad news, he he.

Note: Load R file first (nothing related with test cases is internalized).

Note: This IS NOT the way to do thing like these ... but anyway.

May the Force (the dark option) be with you

Attachments:

K1_join_surfaces_V2A.gh, 166 KB
K1_join_surfaces_V2.3dm, 152 KB

Permalink Reply by peter fotiadis on October 5, 2015 at 5:39am

BTW: by combining the (a bit chaotic, he he) options available you can get a variety of results (Big Brother doesn't allow you to take too much liberties mind):

Of course in cases where the anchor pts differ from the mesh pts (sliders in yellow group) ... well there's another constrain missing: the peripheral tension (like what we do in tensile membranes in real-life).