simple pendulum sticks

[supernewb]

I have modified the ragdoll demo so that a single capsule is hinged using a balljoint at one end to a fixed point above the ground. It works fine except that when the rod is held and released a little off the vertical it sticks and very slowly slips back to vertical. I've tried changing all friction/damping settings but I can't prevent this sticking. Any suggestions please?

[kalesony]

No promises, but you can start by providing the source for repro .

[supernewb]

Source is attached. The pendulum setup starts around line 20 in swing.cpp

Thanks

Supernewb

[supernewb]

Anyone seeing the same problem I'm seeing?

[kalesony]

I see this: https://dl.dropbox.com/u/39695680/swing.avi
Are we supposed to modify the code anyhow for something evil to happen? Or is it the way the pendulum finally stabilizes that you find awkwardly looking?

[supernewb]

Hi, thanks for the post. I get different behaviour! - the capsules stick at a much earlier point, when they are at the maximum reach of a swing. Could this be a machine issue? I'm on dual boot I'll swap into windows on mac later.

[supernewb]

https://www.dropbox.com/s/m0svuf30ztny2fz/swing_SN.avi
see how the stick on the right gets stuck then slowly slips down.

I'm using bullet-2.81-rev2613 / vs2010 / Windows 7 on a macbook pro

@kalesony : You can't get it to stick at all?

[kalesony]

I played around and - nope, it looks nothing like yours. Have you tried different bullet releases?

[supernewb]

Thanks, which version do you use?

[kalesony]

2.80.

Will try it out with 2.81 when I have a spare minute.

[kalesony]

You're right, I have the sticking issue using 2.81 and your code.

[supernewb]

Thanks. I guess this should be flagged as a potential issue.

[Erwin Coumans]

It seems to happen because you enable damping and additionalDamping.

If you comment out those lines, it works just fine here. Can you try it?

Code: Select all

//m_body->setDamping( 0.05f, 0.85f );
//rbInfo.m_additionalDamping = true;