Bullet, more constraint problems

[Juzwa]

Hi,

 

I decided to go further with my destruction and did some changes. Before I had just few balls hitting the building, and that was all that triggered the constarints to break.

 

Now I removed the balls and added a magnetforce, to make the building explode from the inside, and added a drag. Thats all the changes I made. This is how it looks now:

 

 

I dont like it, it's not very natural. The problem is that I have set cone twist constraints for all the parts to keep them together in bigger chunks, however the cone twist looks like is not beeing even calculated. In this particular expample I didn't even use anything to break the cone twist. So the sim should not break at any case, but just stay in one piece and act like a jelly. You can see that the building just collapses, and looks like just glue beeing broken.

 

However in details view I can see that cone twist is beeing calculated, all the angles, torque. But the constarint does not affect the sim. 

 

There must be something wrong with my dopnet. Maybe magnetforce is not beeing supported by bullet? Maybe I must "multiply" the cone twist effection? I really dont know

 

Here is the file (is pretty heavy because of all the fragmentation and constraints)

 

ConeTwistIssue.hipnc

ConeTwistIssue.hipnc

Edited February 10, 2015 by Juzwa

[Netvudu]

with so many pieces, wouldn´t it be a better strategy to go with Glue but different strengths? so that some pieces remain attached?

I still haven´t checked your file though....I will....

[rayman]

You have some really small pieces that causes instability in your simulation (At least in the build I'm making tests 221).

 

Next thing you have to worry is the glue network.

I'm not sure if it is possible to break glue constraint with any kind of force, as you need physical collision to happen to make it break. Correct me if I'm wrong but all your active pieces are connected with one Glue network, so you have one huge cluster. If glue net starts to break it means something is wrong.

 

I cleaned tiny pieces, disabled glue network and everything seems to work fine.

 

 

Btw - I don't like this pre-simulaton setup. I have seen it before and I did't like it. It is a real pain to make any changes in the network. I don't see any real benefit of separating objects packing from constraints creatoin. To isolate some part of the simulation I had to delete not only the input pieces, but also constraints - each one individually. And then to use a wrangle to insure that there are no deleted objects left in the constrains. Otherwise I have to delete input geos and wait for the whole network to recook.

Why not just packing and then create constraints?? I'm not saying it is wrong - it's just my opinion.

[Juzwa]

@rayman - you might be right about that I can't make the glue break by a force. I just assumed it will work, but I have made some test and it looks like it has no affect on the constraint itself (breaking glue).

 

So i removed the forces, because I dont need them. However I still cannot make to work the Cone Twist constraints. I had a very similar setup which worked well, but it didn't have all the objects. When I added all the objects the cone twist stopped working?

 

What do u mean that tiny pieces causes instability? Are there too many pieces in the whole sim? Does this make the bullet crazy and don't import cone twist data into the sim?

Edited February 10, 2015 by Juzwa

[Juzwa]

Ok I found the issue. Somehow I had double connections in Cone Twist Constrants. I started the project on H13 and now I switched do H14. And when I opened the H13 file in H14 I had double connections.

[squardis]

Im not sure but your SOP solver wrangle doesnt have any bracket in the if statement also

if (@angle > 1)
{
    removeprim(0,@primnum,0);
}

 

instead of 

if (@angle > 1)
removeprim(0,@primnum,0);

 

// not sure if houdini needs that