Soft bodies?

[josephrocca]

Hi, I originally asked this on stackoverflow and was directed here (if it's not an appropriate question here it would be great if you could answer it on SO):

I'm told that soft body physics wasn't ported over to Ammo.js. So a few questions:

• Why? Would it be too computationally intensive for JavaScript?
• Are there any JavaScript builds which include the soft body physics part of the Bullet engine?
• Is Bullet's soft body physics reliable enough for me to be attempting to play around with a JavaScript version?

Actually, all I need is simple soft-body cubes that can connect to each other (maybe there's a simple hacky way to do this?)

Thanks a heap,
Joe

[kripken]

I think no one has written the IDL stuff to do that, that's the only reason this wasn't done.

[josephrocca]

Thanks for the quick reply! Sorry, what is IDL?

[kripken]

The ammo.idl file defines the interface between C++ and JS. Only what is defined there ends up exposed in the build. We would need to add the soft bodies stuff to there. See issue #60

[josephrocca]

Okay, thanks. I'm a big old newb with this sort of stuff - how long do you think until ammo might support the soft bodies stuff?

Also, what's your opinion on me trying to simulate simple soft body cubes with springs and masses in ammo? Feasible?

[kripken]

It depends on whether someone decides to do the work. Could be tomorrow, could be never ;) It's not too much effort though, see the progress so far on the IDL stuff.

I would expect ammo to run at about half the speed of bullet in a native build. If your simulation runs at 2x the speed you need it to run natively, likely it will run fine in bullet.

[josephrocca]

Okay (sorry for the late reply!), thanks so much for your help!

[ghost]

@josephrocca did you try your spring and masses approach? I did not know that spring constraints were supported in ammo. How did it work out?

[Kagetsuki]

It appears this has been implemented?
EG: http://kripken.github.com/ammo.js/examples/webgl_demo_softbody_volume/index.html
This issue should probably be closed?

[kripken]

Yes, thanks @Kagetsuki