Rigid bodies get stuck when setting mass > 12 after creation

[kate]

Hi,

I'm getting a strange result when I set a rigid body's mass to a value greater than about 12 (or 12.3534, to be more exact!) and apply an initial upwards velocity: after a few simulation steps, the object just gets stuck at the same position (in "mid-air"), with a velocity of 0. I wondered if anyone else has seen this?

It doesn't happen if I set the mass during the creation of the rigid body, which I realise is the proper way to do things; however, I'm a little concerned as I will need to be able to support switching rigid bodies between dynamic and kinematic states, which will require re-setting the mass (as discussed in this topic: http://tinyurl.com/83agdm - unless I work around it by re-creating the object, though that seems kinda clunky to me).

I've attached a simple program to demonstrate the problem; it's essentially copied from the "Hello World" demo, with the addition of the following code after rigid body creation/adding to the world:

// now set the mass to a higher value
mass = 12.3534;
rb->getCollisionShape()->calculateLocalInertia(mass, localInertia);
rb->setMassProps(mass, localInertia);
rb->updateInertiaTensor();

// add an upwards initial velocity
btVector3 vel(0.0, 1.0, 0.0);
rb->setLinearVelocity(vel);

When I run it, it's fine until the 135th simulation step:

...
i = 132, rb position: 0 0.238125 0, velocity: 0 -0.773519 0
i = 133, rb position: 0 0.225012 0, velocity: 0 -0.786754 0
i = 134, rb position: 0 0.211679 0, velocity: 0 -0.799989 0
i = 135, rb position: 0 0.211679 0, velocity: 0 0 0
i = 136, rb position: 0 0.211679 0, velocity: 0 0 0
i = 137, rb position: 0 0.211679 0, velocity: 0 0 0
... etc.

Maybe I've missed something out in resetting the mass? It just seems odd that it works up to a certain point, then stops...

Thanks,

Kate

P.S. Sorry if this has been posted previously; I've only recently started using Bullet and following the forum.

P.P.S. The attachment is just a text file - I had to add the .zip so it would let me upload it!!

[ola]

To me it seems like your rigid body is falling asleep (automatically deactivated) due to a low velocity. Can you try to disable deactivation on that rigidbody and see if that helps?

body->setActivationState(DISABLE_DEACTIVATION);

The default linear threshold for sleeping is 0.8, so your velocity is lower than that. You can also set the threshold to something else (lower), instead of disabling it completely: body->setSleepingThresholds(btScalar linear,btScalar angular)

Best regards,
Ola

[pico]

Hi,

you also have to remove the body from the simulation before you can change the mass property. Then you can add it again.

[kate]

Hi Ola,

Yeah, that fixed it, thanks... I guess I should have thought of that, I just didn't expect it to deactivate mid-fall! I think I'll reset the threshold; 0.8 seems kinda high to me

K.