SUBD Rabbit from Stefano Giovannoni

Here a first try with Legacy SubD. Before that I used mesh modeled/changed in Rhino then this mesh was subdivided using Catmull&Clark subdivision in Grasshopper.
I modeled Qeeboo Rabbit Chair/Lamp from Stefano Giovannoni


I first began using a quad sphere, then I suppressed an half and used SubD symmetry.

Using my old Perfect Maze generator with differents options



23 Likes

Nice work !!!

1 Like

Making better quad mesh from Maze and with more options.

Difference between Old Mesh and New Mesh


5 Likes

Great stuff, Laurent! Time to take my mesh maze script for another spin! :slight_smile:

For the topology of the first two images, do you produce many small individual mazes, instead of a big one?
Also how do you smooth the sub-meshes, all the while keeping their boundaries stuck together? Catmull-Clark tends to split the individual meshes while subdividing, much like shown in your last picture.

Also do you use recursive backtracking or some of the other maze algorithms (i.e. binary tree, sidewinder, etc.)?

I mixed many things, for the yellow red yes there are 2 mazes, here there are no gaps between the mesh because there just one frontier. For intersection of more than two frontiers there will be a gap when doing CatMull and Clark smoothing, except if you choose to not smooth the border. In order to overcome that I implemented a multimaterial CatMull & Clark, each face has a material (integer indeed), see this link.

For my maze I used Recursive Backtracking. It is perfect maze. See this link

I’ll be interested by what your mazes. I don’t remember seeing your work or an other name

1 Like

I’m currently trying to port my recursive backtracking script to work with n-gons. It currently only works with quad and tri meshes, since it predates Rhino even supporting n-gon meshes.
It currently accepts them, but they are triangulated.

Here’s a rough impression of what it can currently do, including drawing the maze for the mesh edges or face centroids, and meshing the resulting line network.

Thanks for the answer, I would like also to use nGons. From what I remember face connectivity in Rhinocommon is not good to treat nGons because I already have to treat triangles and quad differently and nGons leads to another special treatmen :-1:t. I’ll prefer to have per face a list of Vertex. If so algorithm is quite straight forward :grinning:

Yes, exactly! I currently try to triangulate the n-gon faces to get the neighbourhood information, parse that information and apply it to each n-gon face. The maze would effectively run on a tri mesh, but certain edges or centroids would be omitted from the evaluation because they belong to the triangulated n-gon and not the n-gon boundary.

I don’t have a lot of time at the moment though. It’s a weekend project at best. :wink:

This thread is mind blowing. love it