soft bodies not loosing shape, voxel problem?

[houou]

Hello there, first congratulate Tyson for this great tool!


I'm trying to achieve a softbody simulation ( like soft flesh for instance ) that keeps the shape quite a lot but has some bounciness.

I checked the manual and I saw that there is two main methods. one is using cloth and adding the Binding volume stifness.
It works but it doesn't look like flesh in anyway. 

http://docs.tyflow.com/faq/cloth/

Then there is this other method that says that we should scatter points inside the desired object using the "spawn" node before the cloth node, and then adding the bindings.
It sounds good but for me didn't work at all, it created several copies of the original object and scattered them inside. But well then it efectively created the joints and I could simulate with something.

Then I thought i might be doing something wrong, and tried the birth voxels. and it also worked, but then I don't have mesh. its just dots.

and then ... there is this other method I just found by accident.


1. the proper method (or my try about it) described in the manual for soft bodies.
2. the birth-voxel method.
3. the binding volume stiffness.
4. the weird thing I found by accident that kinda works. ( its not the exact same result as the birth-voxel method, but close ) and uses two, not connected,  set of nodes... is this normal? 


here some pictures and max file.

[tyFlow]

The volume stiffness is identical to Houdini Vellum struts and isn't really that great for achieving stiff softbody forces. It's better in situations where you need a small amount of volume binding strength to help maintain the shape of a flexible surface (think: hot air balloon).

The best way to simulate stiff softbodies is similar to your #2. The steps I would take are the following:

1) spawn voxels in a particle and send them to event #2
2) convert the same particle to cloth
3) in event #2, use a particle bind to bind all your voxels together as well as to the particles that form the cloth

That will give you a deforming surface (cloth) that is strengthened internally by the matrix of voxelized particles within. Those particles shouldn't be made renderable and instead exist only to reinforce the internal structure of the mesh.

Keep in mind binding strength is tied directly to sim substeps. For a sim like this you'll want probably 1/4 or 1/8 substeps, with bind solver substeps probably being in the range of 20-25.

[houou]

got it working!

like somebody posted in the FB group, it's a bit hard to control nicely some of the material proprieties with this method.
Then there is this other method using the PhysX bindings that works very nice. The problem is that it's quite heavy to compute.

Well I don't know if there is plans that in the future the softbodies could be more optimized or better controls. I guess you might be really busy this days.
Thankyou!

[alessandro.d84]

Hi, I m trying to create lots of balls that bounce and maintain their volume but I m struggling to have nice collision between them. 
I m using the method above with particles bind and add particles physics for collision but is really not precise.

The other method is doing the same but with physx bind ? 

I ve tried also with inflate but not having the result that I want.

I thought also to Cuda but it seams not considering the volume particle binds anymore.


Which can be a good approach for balls not completely rigid but not squishy ? Like basketball for example ? 
Thanks