Inflatables example?

Hello,

Where can i get a working example of an inflated cushion? Like the one in the video.

Thanks!

Replies to This Discussion

Permalink Reply by gabriel on January 20, 2011 at 3:11am

+1 !

Permalink Reply by Arthur Mamou-Mani on January 20, 2011 at 2:17pm

You might want to try this one:

Attachments:

LMU_DDW_SESSION5_Kangaroo_Inflate.ghx, 288 KB
LMD_DDW_SESSION5_Kangaroo_Inflate.3dm, 38 KB

Permalink Reply by Danijel Ivanusevic on January 21, 2011 at 1:14am

Thanx!

It's just what I needed to get started.

Permalink Reply by Roberto Maffei on May 16, 2011 at 3:09pm

dear, your definition is cool.

but i tried to add different meshes and it doesn't work! maybe i got lost with the values or my mesh is not really close? (it was, a cylinder) see attach.

thanks

Attachments:

LMD_DDW_SESSION5_Kangaroo_Inflate_other mesh.3dm, 44 KB

Permalink Reply by h on February 8, 2013 at 1:02pm

I had the same problem. Things that fixed it: scale geometries up, rebuild mesh normals, make the time step smaller in the kangaroo settings block

Permalink Reply by h on February 8, 2013 at 10:01pm

I'm getting some bugs...

In this one, the resulting mesh is skewed due to the transition from quad to triangular. Otherwise, the mesh just spins around...

Can anybody offer me any pointers here?

Permalink Reply by Daniel Piker on February 9, 2013 at 9:49am

Hi Harrison,

The definition below shows a different, more symmetrical way of triangulating the mesh (using weaverbird's 'stellate' component), that should help stop the mesh spinning around.

Also, I've added an example of colliding the inflating mesh with a solid like you asked about in your email.

Attachments:

MeshInflate_collide.gh, 26 KB

Permalink Reply by h on February 9, 2013 at 11:09am

Thanks Daniel. I expect this will help a lot. Right now its not running though - The first error is on the first 'Mesh' component: "Solution exception: Object reference not set to an instance of an object"

Is there supposed to be geometry going into the 'move' command upstream of that? I'm getting a translation vector, but no geometry.

Currently, believe that I have Kangaroo, WB, and Meshedit installed. Am I missing something else?

EDIT:
got it to run by using a list item to have just the naked vertices feed to the move command.

Wow, this is incredibly cool, and incredibly fast. How is it that your stuff always runs so smoothly and fast compared to mine? Are there any heuristics I could use to help reduce computation load and speed things up in my definitions?

Thanks,
-Harrison

Permalink Reply by Daniel Piker on February 9, 2013 at 11:30am

Glad it helps,

The error you mention sounds like it is caused by having an older version of Kangaroo.

Version 0.084 slightly changed the nakedvertices component to output both the index of which points are clothed/naked and the vertices themselves.

As for speed - using as simple meshes as possible helps (though too coarse and the behaviour will be compromised). If you want to have a smoother result, you can always subdivide the Kangaroo output with WeaverBird, but by keeping this subdivision downstream of the physics simulation you reduce the amount of computation significantly.

Permalink Reply by h on February 9, 2013 at 11:32am

Cool. I can definitely try that.

I think my issue is coming from my meshes then. I start with a mesh as you have created, then create cylindrical Breps, and turn them to meshes, and subtract them from the initial bag mesh (then unify normals). This creates the 'buttons' I need, but results in a lot of tiny faces. Then I have to use WBQuadSplit to return everything to quads, and things run very slowly.

Permalink Reply by h on February 10, 2013 at 7:20pm

Daniel,

Thank you again for your assistance. Should my project go anywhere, I will let you know, and give you the opportunity to accept or decline an acknowledgement before hand.

Permalink Reply by Gabriel Áh on April 12, 2013 at 2:13pm