[QUESTION] Model for VDB evaluate_dihedral_angle_based_bending_force_hessian

[CliffieVanR]

Hi, can I ask which paper is used for the bending force implementation in VDB? So far as I can tell, the direction vectors of the forces don't correspond to those proscribed by Bridson et al.

Also, are you aware of any bugs in this function? I'm finding I get undesirable results when using it with garments (ie meshes with non-zero rest angles), although it seems to behave fine with the cloth grid.

Thanks in advance! Cliff

[AnkaChan]

Hi @CliffieVanR , the bending energy is k_e * (angle - rest_angle)^2. Yes, we've noticed that currently the bending has some issues for non-zero rest angles. Currently, you can walk around it by calling: self.model.edge_rest_angle.zero_() which will set all the rest angles to zero.

[CliffieVanR]

Hi Anka,

Many thanks! Do you have a reference for the derivation of the force directions and the hessian?

Best, Cliff

On Thu, May 15, 2025 at 12:14 AM Anka Chen @.***> wrote:

AnkaChan left a comment (NVIDIA/warp#730) https://github.com/NVIDIA/warp/issues/730#issuecomment-2881800039

Hi @CliffieVanR https://github.com/CliffieVanR , the bending energy is k_e * (angle - rest_angle)^2. Yes, we've noticed that currently the bending has some issues for non-zero rest angles. Currently, you can walk around it by calling: self.model.edge_rest_angle.zero_() which will set all the rest angles to zero.

— Reply to this email directly, view it on GitHub https://github.com/NVIDIA/warp/issues/730#issuecomment-2881800039, or unsubscribe https://github.com/notifications/unsubscribe-auth/BPMNJMEOA5FUBISNZTQLSPL26PE4BAVCNFSM6AAAAAB5CSXFOCVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDQOBRHAYDAMBTHE . You are receiving this because you were mentioned.Message ID: @.***>