Im creating shattered wall and applying RBD Glue from shelf.
im trying to create simple setup where ball passing through thw wall breaking it.
with Houdini's standart solver i can get appropriate result, but with the same scene and Bullet Solver a cant.
basically the sphere is smashing the whole wall and only after that the Glue constraints are starting to slowly breaks apart.
can someone provide simple scene where Glue constraints with Bullet solver are working fine?
Also can someone explain how to achieve the same effect like on video?
https://vimeo.com/41057935 [vimeo.com]

nothing?

Hey qbick..

I'm not sure the same density and glue values will yield a similar result in both solvers so you may need adjust them - but I've found a number of settings can cause issues.

If you post an example file I can take a look.

Mikal79
Hey qbick..

I'm not sure the same density and glue values will yield a similar result in both solvers so you may need adjust them - but I've found a number of settings can cause issues.

If you post an example file I can take a look.

ive attached a file.
with bullet solver the whole wall bounce-off at collison and only after that it breaks apart.
with RBD everything is perfect.
can you help me?

Hey qbick..

I see the problem and unfortunately I don't know an easy way around that issue with bullet.

You can remove the bounce by increasing the number of collision supsteps in the solver, and also of the dopnet itself. But as it dissolves the glue constraints - the individual chunks don't seem to inherit the overall velocity of the main piece and so appear to pause before breaking apart no matter how many substeps I use.

Some clever person on the forums might know thou?

But is there any reason you want to use bullet if RBD works for you? For what you require RBD might be the most controllable even if it is a bit slower.

thats very sad.
can someone more experienced help us?

I believe the reason is the RBD Glue Object is one actual object, which then spawns the fracture pieces on impact. Using the default RBD Solver, the impact which activates the pieces, breaks the glue, and the collision response(s) all happen in the same simulation step, and in that order.

Bullet's impact, however, registers the collision response and then the impact (from what I understand, at least). Since that impact is what tells the glue to fracture apart, it isn't until the next step that the glue pieces are activated. But since the object bounced in the air, there is nothing to collide against, so it's still together. But when it comes down again, it hits the sphere, and breaks apart (if the internal glue strength is low enough, of course).

Does that all make sense? Using the Glue Adjacent shelf tool creates Glue Network Constraints, which only work with Bullet, and behave more like the RBD Glue Object does with the rbd solver.

goldleaf
Using the Glue Adjacent shelf tool creates Glue Network Constraints, which only work with Bullet, and behave more like the RBD Glue Object does with the rbd solver.

no its not.
fractured object bouncing off and only after that glue consts breaking.
check this animation [clip2net.com] - Bullet
this [clip2net.com] - RBD.
as i can see - there is no way to get glue constraints work correctly with Bullet?

You're right. I was looking at the constraints themselves (which do break immediately), sorry about that.

Here is a specific setup that accentuates the delay, so saying that it doesn't work at all isn't exactly correct, though ideally, Bullet+Glue would work in every circumstance imaginable You'll have to be creative, and break it apart some other way since the software doesn't do it for you. Perhaps in the SOP glue object?

Also, perhaps try reducing the bounce? Bullet is much more sensitive to bounce than RBD (as you have probably noticed)

goldleaf
You're right. I was looking at the constraints themselves (which do break immediately), sorry about that.

Here is a specific setup that accentuates the delay, so saying that it doesn't work at all isn't exactly correct, though ideally, Bullet+Glue would work in every circumstance imaginable You'll have to be creative, and break it apart some other way since the software doesn't do it for you. Perhaps in the SOP glue object?

Also, perhaps try reducing the bounce? Bullet is much more sensitive to bounce than RBD (as you have probably noticed)

im not a houdini adept(for now, mwah..).
so for me its looking definetly like a bug and it needs a fix.
can you help me with “Perhaps in the SOP glue object?”
by “help” i mean do all job for me cos i dont even know from where to start