This is a quick demonstration of an in-progress implementation of this paper:
math.zju.edu.cn/ligangliu/CAGD/Projects/BMesh/
(note that their BMesh has nothing to do with joeedh's BMesh project.)

Currently, the code mainly handles the topology generation described in the paper (although with some extensions of my own to nicely handle almost arbitrary topology with loops.) I've coded some of the tricky mesh evolution too, but it's mostly non-functional right now :)

Notes on things in the video:
* Some parts of the mesh I'm generating have bad normals, causing the apparent material discontinuities.
* Currently I don't handle the case where two adjacent vertices have more than two unique neighbors, which is why in a couple places the model briefly appears disconnected, until the intermediate edge is subdivided.
* The node sizes are stored as a CustomData layer. Currently the only UI for these is a the mysterious little "+" button I click on in the beginning to add a new layer.
* In this video I used subdivision built into the Skin modifier, but it works just fine to add a Subsurf modifier after the Skin modifier, or indeed most of the other modifiers. (The skin subdivision is really there for testing some incomplete code.)

If I can get the mesh evolution stuff working, this could become a very powerful tool for generating forms; even in it's current immature state it's quite fast and easy to generate basic stuff. Certainly nothing you couldn't do with ordinary editmode, but a bit faster (at the cost of precision of course.)

One last note, please refrain from sending me links to ZSphere videos and saying "make it do x, y, and z!". Believe me, I'm aware there's more work to be done here, but I don't get paid to write Blender code. A little patience is needed :) Thanks!

Loading more stuff…

Hmm…it looks like things are taking a while to load. Try again?

Loading videos…