hugo mine

Greetings wizards!

I am exploring this Pixar paper called "Directing Cloth Draping through Blended UVs" that you can find here, with a video presentation of the method at the bottom:
https://dl.acm.org/doi/10.1145/3721239.3734074 [dl.acm.org]

I experimented with it in Houdini and haven't come up with a proper result.
First, I am wondering if overriding the rest length is enough for maintaining original shape, because I struggle to figure out how ? Or is it involving Bend Rest Angle?

I made a test overriding rest length with UV, making sure they have the correct scale, it's indeed looking different in the way it's forming folds, but not preserving the shape especially.

I worked on what they call "fractured UVs" by matching the scale of each triangle on the flat "pattern" islands. The main issue is that, of course, Vellum Cloth will not generate bend constraints from individual triangles (the diagonal constraints), so I tried to transfer them back on "original" vellum constraints, based on fused and non splitted triangles.

Also the paper mention maintaining orientation and wrap/weft, so I feel that the correct scale is not enough and that I am missing something essential. The paper covers some technical aspects that my non so technical brain struggle to really figure out.

Hoping that some of you might have some ideas or inputs about this one, and if I come up with progress I would love to share it with you.

Have a wonderful day and happy new year !

Hey, just reading this post as I am also looking into it.
I agree that it's extremely complex, the more I dig into it the more I discover extra physic rules.

The easiest way I found for now is to make voronoi fracture on a sphere, keep only the surface, then simulate in vellum (hacking cloth to make it super stiff). Inside the solver, a custom pop wind allows to give the main motion, with a chop noise injected into 3 axis. You can consider using advect by velocity that sound more acurate for this effect. Gravity is killed, and sliding constraints allows to maintain the plates against the sphere. A very slight offset allows plates to slide on top or under each other, take collisions into account.

You can then add prefracturing with constraints that breaks based on stress, on any relevant rules (size of the plate for exemple). In real life, the new fracturing are super dynamic and triggered by the actual motion, so you might need to build a very custom solver to achieve this realism.

Also in real life plates are kind of semi rigid on this scale of time. Cratons tends to be super stiff, when other parts of plates are more flexible, as it forms mountains and relief. Magma is also generating new crust when plates are diverging, on ocean ridges.

You can also investigate Gplates that is very specific to this type of task, but the interface is not especially intuitive when you never used it.

Would love to hear from other people what they came with to tackle this kind of effect !

Alright I see, thank you very much for the suggestion. I tried the AdvectByVolumes without success, would you mind suggested me a link or a method for that ? I am not sure how I should convert the geo in vdb or volume, especially how should I set the velocity for making it work. In the context I will work from the geo and not the flip fluid cache.