Limited Resolution and simulation values

[Eritar]

Hi, in the softbody documentation it is written that at max resolution (128, as I understand), one vertex would correspond one particle, but it doesn't seem to be the case? It seems to be governed by Shape Analysis Resolution value as well, and I cannot enter the value higher than 128 manually.

Is it possible to override the limit if I want to have actual one particle per vertex?

In a similar manner, I understand that having a lot of particles would make the simulation unstable, and what values can I tweak to achieve better shape matching? So far what I've gathered - low Mass Scale, Sequential Shape Matching, and Constraint Orientation (I use particle attachments and simulate parts of a mesh) works best.

In this model nose should be much more dense than the rest of the model, for example.

  suzanne_dense_nose.zip (Size: 155.2 KB / Downloads: 1)

[josemendez]

Hi, in the softbody documentation it is written that at max resolution (128, as I understand), one vertex would correspond one particle, but it doesn't seem to be the case?

Only when sampling mode is set to "vertices". In "voxels" mode it just states the 

It seems to be governed by Shape Analysis Resolution value as well, and I cannot enter the value higher than 128 manually.

Shape Analysis is unrelated to sampling density. It controls how particles are projected to the surface of the mesh and how they're oriented/stretched to follow its shape.

Is it possible to override the limit if I want to have actual one particle per vertex?

If your mesh is extremely dense, a vertices sampling density of 128 might not be enough to place exactly one particle per vertex. You can increase the range by opening ObiSoftbodySurfaceBlueprint.cs and modifying the range of the parameter, around line 82 in the file:

Code:

[Tooltip("Resolution of the surface particle distribution.")]
[Range(2, 128)]
public int surfaceResolution = 16;

To eg:

Code:

[Tooltip("Resolution of the surface particle distribution.")]
[Range(2, 512)] // up to 512 instead of 128.
public int surfaceResolution = 16;

In a similar manner, I understand that having a lot of particles would make the simulation unstable,

Obi is unconditionally stable, amount of particles won't affect numerical stability at all. It will affect convergence speed - that is, how "strong" constraints are perceived - and performance: more particles lead to poorer convergence and worse performance.

and what values can I tweak to achieve better shape matching? So far what I've gathered - low Mass Scale, Sequential Shape Matching, and Constraint Orientation (I use particle attachments and simulate parts of a mesh) works best.

"Mass Scale" simply scales mass - which has no effect whatsoever on shape matching, as long as all particles in the softbody have the same mass. In physics, mass controls how forces between bodies affect their acceleration - if all involved bodies have the same mass, the actual mass value plays no role at all. See: https://obi.virtualmethodstudio.com/manu....html#mass

Sequential evaluation improves convergence speed - which roughly speaking means constraints will be "stronger" - at the cost of order dependence.

Constraint Orientation is important if you're going to attach multiple particles and you want both their positions and their orientations to follow the attached transform, so yes this is something you likely want to enable if using attachments.

Now, the main value that drives simulation quality in Obi is the amount of substeps - found in the ObiSolver component. More substeps will yield higher quality simulation, which often results in constraints being able to achieve higher absolute stiffness values. The manual's "how it works" page explains stability, substepping and mass in detail: https://obi.virtualmethodstudio.com/manu...gence.html

kind regards,