Simulation Instability When Attaching Deformables to Rigid Bodies at Smaller Scales

[choiwanuk]

Hello, I’d like to ask about a simulation issue that occurs when using the Attachment feature in Isaac Sim to connect a rigid body with a deformable object.

I’ve uploaded two test videos demonstrating the issue. Both recordings use the same fixed camera setup and contain no external forces other than gravity. The only difference between the two tests is a slight difference in object scale.
[well-did]
https://github.com/user-attachments/assets/f2a55d1f-b272-4502-8bb4-a8e80b2cb88e
[failed]
https://github.com/user-attachments/assets/01e37acc-4e49-470c-a91a-88ab52af30c2

In the video with the smaller-scaled object, the attached deformables unexpectedly fly off or become unstable during the simulation. I’m wondering:

• Why does this instability happen at smaller scales?

• Is there a recommended way to stabilize the simulation for smaller deformable attachments?

All physical properties are left at their default settings, and the issue persists across Isaac Sim versions 4.1, 4.2, and 4.5—so it doesn’t seem to be version-specific.

I’d really appreciate any insights or suggestions. Thanks for reading!

[RandomOakForest]

Thanks for posting this. Although, a change in scale is something we can easily visualize, simulating its physics is a more elaborate process. In Isaac Sim, you need to apply the scale change to the mesh, not the rigid body of the object, to have the effect you are looking for. Let us know if you need further help.

[choiwanuk]

Thank you for your response, RandomOakForest.

Based on your suggestion, I conducted the same test using a mesh that was downscaled in Blender before being imported into Isaac Sim. I also included measurements in the video to show the actual size of each mesh and confirmed that all the meshes have their scale set to 1 after import.

sim_attachment_error.mp4

Unfortunately, the issue still persists. Minor downscaling seems fine, but once the object's size approaches around 0.1 meters, the instability consistently occurs. This makes it quite difficult to simulate objects at centimeter-scale dimensions.

It seems I’ll need some additional guidance on how to address this. If there are any further recommendations, I’d greatly appreciate it.

Thanks again for your kind.