[QUESTION] Clarification on surface FEM and rigid body motion under out-of-plane loading

Hi,

I’ve been using warp.fem for cloth simulation, utilising a corotational SVD-based model with Green strain over 3D triangle meshes.

While investigating rigid body behaviour, I noticed that the cloth rotates properly when external forces (e.g., gravity) act in-plane (e.g., along X or Y), but not when they act perpendicular to the surface (Z direction). Initially I suspected that fem.grad might suppress out-of-plane variation, but after inspecting gradients for tilted triangles, I confirmed that the test and trial gradients do include non-zero components in the normal direction.

Despite this, the cloth still does not rotate under Z-directed gravity, even in asymmetric setups. I’ve also tried adding bending terms (e.g., discrete shell-style energies), but the issue persists — suggesting that the problem lies deeper.

Are there any known limitations or subtleties in Warp’s surface FEM implementation — such as the way test/trial functions are defined, or integration weights are computed — that could suppress rigid-body torque under out-of-plane forces?

Thanks,
Chris