Vellum and RBD both have the newer packaged workflows, and while it’s nifty on their own, I haven’t found a way to set up such that they can mutually affect each other properly in a single simulation.
Usually, we’d set this up in a custom dopnet using the multi-solver, but because of the way the packaged node’s geometry and constraints are group together, there’s no straightforward set up setting this up.
So here’s a quick and dirty (but efficient) way to fake the interaction between RBD packed simulation and a Vellum simulation.
The idea is to first mimic the movement we’d want of the interaction using just the RBD bullet solver. We feed that output to serve as a collisions geometry for another temporary Vellum solve, feed this temporary Vellum output back to a new RBD bullet solver for the final RBD simulation, and once again feed it back to a new Vellum solve for the final vellum interaction.
This multi-phase simulation approach yields decently convincing results. While it’s not technically accurate, it’s artist-friendly to utilise the packaged simulation nodes.
This screenshot of the node tree shows how Vellum and RBD packaged SOP level nodes feed into each other to arrive at the illusion of interaction.