Kangaroo - convert output mesh to surface (not polysurface)?

Hi Everybody

I've been playing around with Kangaroo a bit and successfully managed to get my head around inflated structures / minimal surface structures.

I want to take this a step further now to try and make a planarized triangulated structure.

My question is can you convert the output mesh into a surface to allow the use of surface divide (and the many wonders which stem from that simple operation).

At the minute i have a polysurface which is incapable of being divided. While writing this I am think that if you rebuild the original surface (which is then meshed to a suitable resolution perhaps the corners of the polysurfaces could be the pseudo-divisions.

But basically can a mesh be converted to a surface (not polysurfacee)

Many thanks you beautiful mass brain....

Replies to This Discussion

Permalink Reply by David Stasiuk on August 6, 2014 at 6:40am

What sort of input mesh are you using? If it is derived from a base polysurface, then it's fairly easy to rebuild your base surfaces from the outcome so long as you maintain their topologies (and, say, build your simulation-driving meshes from grids of their control points). It'd be helpful if you could post a definition or image as well...

Permalink Reply by J-P on August 7, 2014 at 2:58am

Hi David,

here are is the definition and variety of surfaces I am getting outputted.

I am using a surface as the input and using the mesh brep component to get the mesh for kangaroo,

I am converting it into a polysurface within rhino using the mesh/polysurface in the mesh tools.

When you say to use a grid of the control points to produce a mesh for simulation, is it possible to perform the reverse and produce a single surface from a grid of points, I've used patch before (many years ago) but I remember getting far more control points that tapered off outside the surface I was wanting

many thanks

Attachments:

outputs.png, 303 KB
def.png, 178 KB

Permalink Reply by David Stasiuk on August 7, 2014 at 6:14am

Here you go...if you use the mesh surface component instead and divide your surface using the same UV dimensions, it's easy to track the points and rebuild surfaces from them.

Attachments:

RebuildNurbs.gh, 19 KB

Permalink Reply by J-P on August 8, 2014 at 11:06am

Hi David,

Many thanks for your reply. I was under the impression that Kangaroo required a mesh input, I never realised the surface points could be inputted, and a surface built this is excellent ground breaking news!

However this has thrown up a very strange GH object ....

Firstly - I am not a fan of divide surface, as it can give uneven distribution of points (see image). Arthur Mammou-Mani, showed me a very neat little method of producing a grid, then using a bounding box and evaluating the UV co-ordinates of the points of the grid relative to the bounding box, then transposing them onto the desired surface which will correctly plot the UV values... this has proved to be invaluable!!

This little operation relies on reparametricising the surface, however the surfaces outputted from kangaroo is exhibiting some strange behaviours. If i reparametricise the surface it vanishes (or should I say becomes 1mm in size) I've never noticed this with other surfaces I have done this to (I thought this was an internal operation to do with the structure of the surface - which usually doesn't affect the surface (equally reparameterising a curve doesn't outwardly change its appearance).

I then thought it must be to do with kangaroo being 'live' and the surface being constantly updated. I thought I could bake the surface and then reference the baked surface back into the definition independantly.

So this is where I am confused, even when the surface is baked and then referenced in a new surface placeholder, if that surface is reparameterised, the surface within kangaroo is scaled to one unit (effectively vanishing). I was under the impression once an object was baked, it severed the links to the GH definition, however this seems to remain linked. The form no longer changes, the points remain visible, but it will not form a surface (again I realise a surface is being formed it is 1mm), but it is still affecting the Kangaroo object.

Similarly it doesn't matter where the surface is reparameterised 'downstream' of the definition, if it is reparameterised anywhere, it will disappear / become size one. I thought GH definitions only worked one way (aside from hoopsnake) and I would have thought any changes 'downstream' would not affect anything that came before it in the definition.

I do have one other question, but given the length of this essay like rant, I'll ask it some other time....

Any info would be much appreciated...

J-P

Attachments:

divide surface.png, 382 KB
KANGAROO SURFACE.gh, 44 KB

Permalink Reply by David Stasiuk on August 8, 2014 at 4:21pm

Yes, the point distribution can be odd in some untrimmed surfaces. However, if you set "interpolate" to true when you build a surface from points, you'll find the distribution to be even and consistent with the points you've used to construct your surface.

I can't reproduce the issue you're finding with the surface disappearing...one thing you can do is reparameterize your surface right in the beginning. Then you don't need to worry about the bounding box thing to give you your UV coordinates...your surface will run from 0 to 1 in both the U and V directions. (you also don't need to use the "explode brep" component, unless you're using a polysurface...a single nurbs surface works fine on its own!)

Something I do see that is giving you trouble is the anchor points you're trying to layer into your kangaroo execution. Those points you're trying to insert in the middle of the surface won't do anything because they aren't tied to any of the springs you're working with. In order for an anchor to work, it has to be attached to other particles in your simulation, so you either need to select points that are already in the simulation or you need to find a way to add them. I've put something together for you here that I think may be what you're trying to achieve...let me know!

Attachments:

KANGAROOSURFACE_ds.gh, 105 KB

Permalink Reply by J-P on August 12, 2014 at 1:21pm

Hey David,

cheers for the help. I was aware that the points on the circles, weren't being chosen from the surface points themselves, but wasn't really sure how to go about it. And other issues were more perplexing...

I've had limited success.... the front edge of the mesh not attaching is mildly irritating, but the main issue if you look at the image is that I want to pull the circles attached downwards, at the moment I get a constant base surface....

I've been thinking it through and I think a minimal surface rather than inflated is the way forward...

my file is getting pretty huge so i've started a little file to demonstrate the problem of making a minimal surface with multiple holes.

I will refrain from rambling and leave it at that....

Attachments: