Jittery edges on softbody

[Matthias]

Hey, I'm making a project where I have an attached softbody that is able to be deformed, but I'm running into some issues.

The goal is for the mesh in the image below to be attached to the table it is on, this table is a static object.
Currently the issue I've been running into is that when I use an Obi Particle Attachment and bind the bottom particles of the softbody to the table, the top edges of the mesh begin to jitter, as seen through the deformation in the screenshot below.

The mesh is set up similarly to the dragon from the sample scene, having split the soft body from the skinner and renderer, though the issue remains when I have it all as 1 object.



When the mesh is placed on the table without an Obi Particle Attachment it is fully alright, similarly if I enable Constraint Orientation the mesh is also stable, but does not store any deformation that occurs to it, bouncing back to its original shape the second the collision ends.

I've tried quite a few things already, such as increasing my substeps, all variations of the blueprint generation using Surface and Volume sampling, and I've confirmed that it's likely not an issue related to my solver since the issue also occurs when I place this mesh in the DeformableBarrels sample scene.

The mesh does currently stay deformed after any collisions, the jittering is the only problem I'm having.

Does anyone have any suggestions as to how I can get this jittering to stop?
I would greatly appreciate any input, thanks already!

[josemendez]

Hi!

Using “constraint orientation” is usually a must on oriented particles unless you want them to freely rotate and are geometricamly allowed to do so. Otherwise, small impulses from positional solves affect particle orientation, and may lead to jitter.

However the bit about constraining orientation no longer allowing plastic deformation is strange. What are your plasticity settings for shape matching constraints?

Kind regards

[Matthias]

Using “constraint orientation” is usually a must on oriented particles unless you want them to freely rotate and are geometricamly allowed to do so. Otherwise, small impulses from positional solves affect particle orientation, and may lead to jitter.

However the bit about constraining orientation no longer allowing plastic deformation is strange. What are your plasticity settings for shape matching constraints?

Hey!

I'll attach an image of the properties on the gameobject to this message.
With the settings in the screenshot I get a mesh that does not jitter, but also refuses to stay deformed.
Note that I set the "Deformation resistance" to an immensely low value for testing purposes, in hopes it would have fixed the issue, but sadly it did not change anything.



Thanks for the speedy response!

[josemendez]

Hi!

Constraint orientation has absolutely nothing to do with plasticity, so I'd say there's something else going on.

Would it be possible for you to share this setup by sending a project that reproduces this issue to support(at)virtualmethodstudio.com? This way we can take a closer look and diagnose this properly.

thanks!

[Matthias]

Would it be possible for you to share this setup by sending a project that reproduces this issue to support(at)virtualmethodstudio.com? This way we can take a closer look and diagnose this properly.

Hey!

I've sent an email to the provided email address with a minimal setup attached that recreates the issue.
It's titled the same as this forum post.

Thanks for taking a look at it, hoping to find out what the issue is soon!

Kind regards

[josemendez]

Hey!

I've sent an email to the provided email address with a minimal setup attached that recreates the issue.
It's titled the same as this forum post.

Thanks for taking a look at it, hoping to find out what the issue is soon!

Kind regards

Hi Matthias,

The issue is that your comb structure only has two layers of particles: top and bottom, the bottom one being attached.

When "constrain orientation" is disabled, these clusters have infinite mass but non-infinite inertia tensor, which allows shape matching clusters to register deformation.

When you enable "constrain orientation" on the attachment, the inertia tensor of those particles becomes infinite. As a result, the shape matching clusters no longer register any deformation as they're are effectively kinematic (since they have both infinite mass and infinite inertia tensor, resulting in zero DOF).

The solution is very simple: add an intermediate layer of particles that allows deformation to take place in the clusters between this intermediate layer and the top one.

Keep in mind that softbody deformation is *not* computed using linear struts/springs between particles (this may have led you to using only 2 layers of particles), but using clusters of particles and modeling volumetric deformation as a 3x3 matrix. Compared to simple springs, this has the benefit of allowing arbitrary affine deformations to take place.

let me know if you need further help,
kind regards