Download

Linear Limits "ignored" by PHAT in 4.24 ?

Well, not ignored exactly.
If you set them to breakable they break.
If you set a lenght and a restitution, they stretch and then spaz out
Hoever if you set them to Locked and grab the mesh during simulation they stretch and spaz out.

Projection is enabled. Projection Linear Tolerance is set to 0.

I have hte same issues when swapping to any preset Ball&socket, hinge, prismatic or back to skeletal.

Adding a linear motor does nothing.
Setting a soft constraint does nothing.

If i select all bodies and disable gravity then simulate, nothing moves, so I know my setup is good and nothing is spazzing out on it’s own.

Can anyone else confirm this? Guess I need to try the new engine version too sigh…

Same issue in 4.25 - issue still present with 50 iterations on the simulation.
How are people making simulated chain links if you can’t pin or grab one of the sides?

Max sub-step isn’t doing anything for it in .24 or .25.

Alrighty, you tried selecting only one bone at a time and simulating that with no gravity. You might find just one of the bones spazzes out by itself but seems ok when everything is simulated in phats together. Took me bloody hours to get a custom skeleton to work in an acceptable way.

I hate phats…

I’ll go through it again, but the issue only really happens when you apply force to the last bone in the chain.
if you grab the one before the separation is much less.

It seems to me like the physics code is just badly implemented onto the skeletal mesh.
I do not recall having this issue in 4.19 or was it it .18? Which ever one has the new changes to phat come live.

This is driving me nuts.

I created physics profiles and created the whole system, everything works around 98% of the time, until the linear limits just decide that they no longer apply…

I just don’t get it. It’s not like there is any option to change - and sub-stepping is on.

Hey… Now that I remember, This **** was also happening randomly on ARK… now it makes a little more sense and I’m 100% sure i’m doing something wrong …

So, not going to say “in the end” because I’m still nowhere.
but ramping X on Inertia Tensor Scale to about 4 or 5 “solves” the linear limits problem. On the other hand it makes the mesh simulate like molasses too, so still nowhere.

For now, I ended up using that to systematically scale the skeleton.
4 at extreme extremities, 3 @ forearms/calves/neck, 2 @ upper arm/thighs - rest of constraint except pelvis.

This makes the simulation a bit leggier then one would want, but it seems to be void of error and weird displacements. As soon as I re-do my whole IK system for aligning the get up animation to the floor I’ll probably release some footage of before/after.

Meanwhile, if anyone happens to have a different/better solution - inclusive of opening up the engine source and tweaking physx (which I still think has some issues within it’s implementation but haven’t bothered looking into) I’m all ears.