Balloon packing (voronoi?)

Hi i need a bit of help with this project

I'm trying to 3d model these plaster models (images below) inside rhino using grasshopper. I've seen a lot of work similar to this in the discussions but I can't work out how I should do this. I'm thinking the basic form is like Voronoi 3d but because the balloons when they are packed together they deform each other which is like a shelled or filleted voronoi. I started with this idea looking at the work of Neri Oxman, she uses a voronoi but i think it is 2d extruded. I'm still pretty new to grasshopper so if anyone can give me a hand to figure this out.

I have worked out how i can make a voronoi 3d using http://www.grasshopper3d.com/forum/topics/3d-voronoi-4

but i don't know how to shell the voronoi to get the surface quality from the balloon packing model.

maybe there is another way of doing this?

I'd like to be able to control the location (central point) and volume (which remains the same through deformation). The purpose to getting it into 3d is to have more control over the undeformed form and it's volume and control of location

any help is much appreciated

Daniel

Replies to This Discussion

Permalink Reply by Michael Pryor on October 10, 2011 at 6:03pm

Here is a start, play with attractors, scaling, and box morphing to deform the spheres further.

Permalink Reply by daniel koene on October 25, 2011 at 4:10pm

Hi Michael

Thanks for your help. I have been trying this for a while now with not so much luck but my knowledge of grasshopper has increased by 100 fold trying to solve my problem yet it remains unsolved.

I can do what you suggest here intersecting spheres within a box but i am still stumped in getting the spheres to non intersect (deform as the balloons do). I am not sure how the attractor thing works exactly. I've had a look at kangaroo but its complicated. also thought of rabbit. also thinking the way metaballs interact is interesting but it would be a reverse of that where they don't stick together they repell eachother. just not sure how to use them.

I was trying to work out if the spheresurfaces could be divided into points and then interact them. If the closest point on one sphere to another sphere would stick to the other... so it would create a plane along the tangent of the intersection, but i think that would create something similar to the above boolean intersection. could the sphere form deform like that?... not sure about that.

another way maybe along the lines of your suggestion ... i instead of using the spheres use boxes that get morphed using twistedbox and those boxes get morphed by the voronoi planes. so that the spheres will then be referenced back into the boxes using box morph. do you know if there is a way to use a brep(the voronoi forms that are produced from your solution) as the box for twistedbox (eg. a more than 6 sided(cube) form)? just not sure how to do this.

any further suggestions are much appreciated

Permalink Reply by Adam Blaney on April 1, 2012 at 1:56pm

Hi Michael

I am really interested in your above development in sphere packing, I have been playing around with my own definition but I am pretty stumped on how to Boolean intersect the sphere with the bounding box.... Could you please help with any advice, I have attached the grasshopper file I have been working on and the resultant model that I wish for it to produce. I am working with grasshopper version 0.8.0002

I would be very grateful for any advice.

Many thanks

Adam

Attachments:

spherepack TEST.ghx, 187 KB
Sphere test image.jpg, 1.7 MB

Permalink Reply by taz on October 25, 2011 at 7:17pm

You can't really simulate collisions yet with Kangaroo...

http://www.grasshopper3d.com/group/kangaroo/forum/topics/inflate-co...

It seems something like this is what you want to do, but you'd need animation-type software.

http://www.dfab.arch.ethz.ch/web/e/forschung/52.html

Permalink Reply by daniel koene on October 25, 2011 at 10:26pm

thanks taz

thats a nice solution to arranging spheres. I like the project. I wish i had a brick laying robot.

as for arrangement of the spaces i have had a go at random points within a brep being the original form. but i would ultimately want to control where the spaces are so random or simulated dropping won't work.

with kangaroo is it possible to deform forms like the balloons so the surface becomes like a solid form?

Permalink Reply by daniel koene on October 25, 2011 at 10:07pm

Ok some progress has been made in deforming the spheres into the 3dvoronoi. one problem with it is that the twistedbox component only works for 8 corners i need to vary the amount of corners to 6 or 10 or maybe more. it is quite a mess at the moment but i can clean it up later. Does anyone know if more corners can be added i think twisted box works fine with 6 corners it just drops the remaining. just need to know how to work with more corners.

The forms that come out are nice but go through severe deformation. would there be a way to inflate or i guess scale the ovoids...

any ideas...

Permalink Reply by daniel koene on October 25, 2011 at 11:01pm

this is what i have so far

Attachments:

sphere pack test 1.3dm, 1.2 MB
sphere pack test 1.ghx, 1 MB

Permalink Reply by daniel koene on October 31, 2011 at 5:48pm

Hi so I think i've got as close as i can get. I borrowed from the definition in this discussion http://www.grasshopper3d.com/forum/topics/inflate-collide-deform it uses kangaroo to inflate a mesh and collides with a brep. I basically modified it to inflate a sphere inside the voronoi forms and then intersect the two once the balloon is inflated.

here is a pic of what i have tested it on.

and here is the files if anyone wants a look. I like the outcome because what happens in grasshopper is similar to the process of making the plaster models.

This method compared to the others would be easiest for applying to a series of forms. where as my previous result with the twisted box was labour intensive to attach the vertices to each twisted box. all the voronoi forms points i was using were listed differently. also the twisted box only can only be applied to quadroid forms.

So this likely is my final result to this problem. I will post my final images for this project later.

Attachments: