Max number of colliders on PhysX3-branch cloth is 32???

Hi all,

I am considering porting a PhysX project built on the latest SDK in the PhysX 2 branch (2.8.4.6) to 3.x branch and to my surprise with the latest 3.3 beta2 I get a warning when I more than 32 spheres collide with the cloth saying “warning : Dropping collision sphere due to 32 sphere limit”…

I had my 2.8-branch code collide with over 700 spheres with no problems! Could it be true that such a tight limit has been imposed on PhysX 3.x clothing???

Please tell me there is a way to drastically raise the number of colliders the cloth simulation can accept!!

Thanks for any hint… grasping at straws on this one!

Dan,

Edit to the above: I tried setting the spheres with both the ‘PxCloth::setCollisionSpheres()’ and as their own static actors using PxCreateStatic(…, PxSphereGeometry() …)

Both appear to have a 32 sphere limit, although I haven’t tested if I could insert 32 spheres within cloth and another 32 as external actors

Even with 64 or using capsules is too limiting to my app and the 700 capsule limit I was able to reach with PhysX 2.8.4.6 was far better.

Has anyone noticed this sharp decrease in the usefulness of cloth colliders with the PhysX3 branch?

Thanks!

It was said many times. Currently, PhysX 3 cloth is mostly oriented on clothing simulation for game characters. And you don’t need 700 spheres to represent character skeleton.

Hi Reticula, thanks for taking the time.

My current project is to perform fast & accurate cloth fitting around a complex shape like the torso of a character by approximating the body mesh with as many capsules as PhysX can take before choking. With PhyxX 2.8.4.6 I was able to get up to 700 capsules colliding with a 4K-vert cloth at over 24fps… pretty useful when Blender cloth sim on the same meshes would give me one frame per 6 seconds.

I’m just surprised to see a 20-fold reduction in supported collider. So if I read you right there are no way around this limit? The right SDK for my project is therefore PhysX 2.8.4.6?

Thanks again for your recommendation! Saves a lot of time.

Dan,

Hi everyone,

Is the PhysX2 branch the recommended path for cloth-related applications that need a high level of capsule colliders?

While I’d prefer to use PhysX3 branch because the greater flexibility of tapered capsules, the limit of 32 colliders per cloth is too restrictive for my app and I must return to normal capsules of PhysX2 branch if a workaround to the 32-collider limit of PhysX3 cannot be found…

Any thoughts / recommendations?