Well done

fsimerey: i agree and i think that this is better way than ODE

Few optimalization: http://vimeo.com/13484440 [vimeo.com]

next move will be convex and concave body, then advanced forces(field) and finally constraint :wink:

Look this particle physics framework. (done in softimage and ICE integration)

http://lagoatechnologies.com/ [lagoatechnologies.com]

vimeo channel > http://vimeo.com/channels/lagoa [vimeo.com]

Very great engine. I hope see it in houdini.

Way to go cybermax! I've been working on the DOPs solver, and we've talked about getting a SOPs version of Bullet over on odforce. Are you planning to share the code to this?

Well done!

Oh, btw, the multithreading for bullet only works with Sony Cell processors (PS3) and I think partially in 32-bit Windows. It's still a WIP, along with OpenCL acceleration. But those are slated to be done by version 3, which means we might start seeing more and more of those as time goes on.

fsimerey:

HOLY CRAP!!!
The first scene is really SICK ops:

How many objects can be in the first scene? :?:
I like try it, but I think that I dont have a lot of memory.

Really thx for this video - There is a lot of good ideas .


goldleaf:
In this moment I dont know If I will be share it.
I know Bullet DOP project and I used it few part of codes :shock:
Now I have box, spheres, convex, concave and many forces.
I am working on “btGeneric6DofConstraint” for using it in Multijoints.
I also created a lot of code which prevents the crash of plugin.

See you in another video…

There's a better way in doing glue type objects in bullet with btCompoundShape.

Next video!!!
Funny,Funny, Funny

Multijoints with Bullet is BIG FUN :shock: - http://vimeo.com/13585660 [vimeo.com]


Constraints are implement over “btGeneric6DofSpringConstraint”, so the user can set the general (PtP, Hinge, Slider …)

I think that ODE constraints are more unstable, that Bullet constraints - Actually the bullet is only stable

The biggest drag is multijoints, calculated via SOP nodes about 40 seconds (from 300 boxes is created 1080 constraints). I will want to implement it through HDK so I think it will compute less than one second. In video Constraints are not yet rendered, I will draw later When I will implement with the HDK

in another video, I will break constraints through max_force that bullet does not support default, but now I know how to do it.


Ratmann: Yes, I know about this function, but I think that with btGeneric6DofSpringConstraint I have more control and Its still fast.

aaaaand another: http://vimeo.com/13592218 [vimeo.com]

Its simple. If the spring force in test constraint is greater than the maximum force, that test constraint is deleted from the simulation.

Great job for now cybermax
interesting stuff

CeeGee: thx, until now everything was just a test

Now I am preparing a much more complicated scene and the first tests are simply awesome :shock: . But I will record final video, maybe this week.

How about a cheeky screenshot to give us some idea of what you're up to…

Its first WIP so There is lot of mistakes.

http://www.postimage.org/image.php?v=aVHVImA [postimage.org]

http://www.postimage.org/image.php?v=aVHVKRJ [postimage.org]

http://www.postimage.org/image.php?v=aVHVNlS [postimage.org]

http://www.postimage.org/image.php?v=aVHVPR0 [postimage.org]

http://www.postimage.org/image.php?v=aVHVSl9 [postimage.org]

http://www.postimage.org/image.php?v=aVHVUQi [postimage.org]

Incredibly awesome!

Good job. Waiting for final results ))

Update WIP:

I implemented drawing Constraints including changing color according actual_force and max_force. And its work fine.

Next thing is that I write some code for testing “Has Object Convex hull?”
because If I using a voronoi fracture, that a lot of objects is convex(fast for simulation), but If there is some concave hulls, there is error in the collision, so if this code return false, that I create concave object.

bool SOP_Bullet::testConvex(int *gIndices, int numIndices, btVector3 *gVertices, int numVertices){


for(int i=0; i < numIndices; i+=3){

btVector3 v1 = gVertices[gIndices];
btVector3 v2 = gVertices[gIndices];
btVector3 v3 = gVertices[gIndices];

btVector3 e1 = v3 - v1;
btVector3 e2 = v2 - v1;

btVector3 n = e1.cross(e2);
n.normalize();
float plane_d = -n.dot(v1);

for(int j=0; j < numVertices; j++){

float dist = gVertices.dot(n) + plane_d;


if(dist < -0.01 ){
cout<<“dist: ”<<dist<<endl;
return false;
}
}
}


return true;
}


But I have problem. Concave object has unrealistic big bouncing effect :evil: I am trying some combination with Restitution and Collision margin, but It still exist.

so any Ideas how this fix it???

ok, tomorow I try to break objects into a collection of convex parts and maybe It helps 8)

done, Its call Convex Decomposition and its fast.

In this moment I dont know about some bugs so I will continue in house scene.

maybe waiting for some WIP video today 8)

First WIP video: http://vimeo.com/13788796 [vimeo.com]

As you can see, I still using H10. When I will have more time, I look at on new H11 features like Voronoi breaks.

Enjoy it and stay tune for the next WIP video.

Hello cybermax. Nice simulation!
Small question new fracture(v.11) is for ODE engine only?
I tried to merge with Bullet phy.engine(rough recompiled to v.11.0):
merge node, under it - bulletrbdsolver with ode solver(for fracture pieces)…but without success.
Has anybody tried it?

ilay7k: I havent tried, yet, but Everything what I do I make in SOP level. So I will use SOP voronoi. I read some information and H11 has voronoi for DOP too. But in DOP system compute crack according collision point(s), If I understad right??