Developabilize Force

I am currently trying out the developabilize force. In the paper on which it is based, the algorithm is demonstrated with a sphere (among other complex shapes):

I tried to achieve a similar result, but this is what I got with Kangaroo:

Is there sth that can be done to get better / simpler results?

Thanks a lot,
Cheers Rudi

1 Like

Something which the paper sort of skirts around is that this technique is extremely mesh dependent.
So how your starting point is meshed matters almost as much as what shape it is.
What mesh did you start with for the above?

I also realized later that the results they show actually involve a fair amount of manual work (see the caption (c) of Fig 25). So in the end I feel like it’s a neat technique that can produce some interesting effects, which can form a part of a workflow of turning freeform shapes into fabricable developable pieces, but it’s not quite the fully automated thing it makes it seem at first.

I was initially going for sth more complicated, but then decided to try the “simplest” geometry first - a subdivided cube using the Catmull-Clark subdivision.

interesting… haven’t had the time study the paper yet

I guess you also had a look at the work done at TU Vienna?

Would be really cool if we could gain more info from this research - the results that they are presenting look incredible.

and btw: How were you able to grasp all the mathematical backgrounds?

True indeed!

hello guys, is there any example file or documentation of this node? (31.0 KB)

-edit - actually here’s a better one, with the plastic anchor updated to the new version: (148.5 KB)

I’d forgotten how much fun this thing is to play with :slight_smile:

The results remind me a little of the sculptures by Ryo Hagiwara



how would you approach the unrolling of the resulting meshes? I guess one has to smartly split the mesh into patches including the unwelded seams, but then? Is there some rebuilding that has to be done?

Indeed, with this approach there’s a bit more processing that needs to be done to turn the resulting meshes into something that can actually be unrolled, and I never made this last part fully automated.
I also found that with this technique, although it gives piecewise developables, it’s not always easy to control the size of these pieces or the smoothness of the borders between them.
So as it is, it seems to me more of a tool for stylization, or giving some help towards developable fabrication, but it’s not a fully automated freeform-mesh-to-flat-cutting-pattern pipeline.
There’s been some interesting work on developables since this 2018 paper though, such as

and I’d still like to some day figure out a smoother process for all this

1 Like

From the ye olden days of architecture school, this was the recommended tool for unrolling/unfolding. Seems like it still exists!

Yes, I remember Pepakura - great tool, nice to see it is still around.
I think it’s not much use for solving this particular issue though.

In the case of these developable meshes, even after splitting at the seams, the patches are only developable up to discretization error (because the triangulation is not aligned with the ruling directions). So the flattening step needs to allow a tiny amount of stretching while keeping things connected.