Rigid body gaining energy

[Julio Jerez]

I finally determine the casue for these problem, and I made a ssampel mesh that expose every time even with a sphere.

now tomorrow I will made the fix and that will take care of the bug for ever.

[Julio Jerez]

Ok see if this is better. is not efficent yet I am testing a solution first.

[Bird]

Ok see if this is better. is not efficent yet I am testing a solution first

Good news is that I don't see any problems anymore in the Newton demo scenes. But I'm still having problems in my project.
http://www.hurleyworks.com/media/flash/Poppers2/Poppers2.html


-Bird

[Julio Jerez]

Ok, now taht teh collsion routine are all writen we nee to debug that bug again.

I can understand the tunneling because continues collsion is no on yet, bu the high bounce does not look right.
This weekened I am going to go ove the dgScene and the serialization so that you can save that scene, and I can debug it and see what that problem is all about.

Also need to optimize the post prosseing of teh contact that I just added, it is no very effeceint, I just wantd to see if it will fix the bug with doble faces with normals pointing the oposite direction.

[Bird]

This weekened I am going to go ove the dgScene and the serialization so that you can save that scene, and I can debug it and see what that problem is all about.

Cool. I have some other problems that I'd like to save and load into the Sandbox demo to test.

I hope you can finally put an end to this evil bug and move on to something more interesting again (like destruction )

-Bird

[Julio Jerez]

Al right let us do this,

in you applicatiopn you cna insert this call
NewtonSerializeToFile(scene->GetNewton(), "filename.bin");

and it will save the scene, to a binaty file, then you can import that scene in the snadbox demo and see if it works.
then when you see the Bug you can simply send me the saved file and I cna debug that.

it does no save joint yet I will do tah later,

[Bird]

vs2010 project won't compile in latest download (v 2558 )

dgBroadPhaseCollision.cpp
c1xx : fatal error C1083: Cannot open source file: '..\..\..\source\physics\dgBroadPhaseCollision.cpp': No such file or directory

-Bird

[Bird]

vs2010 project won't compile in latest download (v 2558 )

dgBroadPhaseCollision.cpp
c1xx : fatal error C1083: Cannot open source file: '..\..\..\source\physics\dgBroadPhaseCollision.cpp': No such file or directory

-Bird

Ok, I see it was renamed to dgBroadPhase but the vs2010 project wasn't updated yet. Also the new openCl classes (dgOpenclInstance and dgOpenclBroadPhase) need to added to the vs2010 project so that the Sandbox demo will link properly.

-Bird

[Julio Jerez]

Ha, ok I will update the VS2010 project tonight.

I was trying to make the OpenCL broaphase, so I renamed the Broaphase to be a super class common to all broaphases.

[Bird]

in you applicatiopn you cna insert this call
NewtonSerializeToFile(scene->GetNewton(), "filename.bin");

and it will save the scene, to a binaty file, then you can import that scene in the snadbox demo and see if it works.
then when you see the Bug you can simply send me the saved file and I cna debug that.

Excellent! I've got that working and it makes it really easy to export test scenes. But one problem is that when I import the serialized scene into the Sandbox, the camera does match my LW scene and by the time I adjust the camera in Sandbox, the simulation is over. Would it be possible to add a "Start" button in the demo player and postpone the start of the simulation until it's been pressed so that there's time to reposition the Sandbox camera to get the objects in frame?

Also it looks like the static bodies are serialized in their untransformed state, so some simulations won't play as intended in Sandbox.

-Bird

[Julio Jerez]

Ok I fixed the orientation problems, can you try again?

on the camera orienation, not serialization do no save the orientation, because newton world does not have a graphics represenation.

[Bird]

Ok I fixed the orientation problems, can you try again?

on the camera orienation, not serialization do no save the orientation, because newton world does not have a graphics represenation.

Orientation problem is fixed. No problem with the camera now, I can just set up the scene in LW so that it matches Newton's camera now that I realized that Newton's forward is X and not Z like it is in LW

I made a new scene that shows the high energy problem Here's what I see in Sandbox http://www.hurleyworks.com/media/flash/NewtonPoppers/NewtonPoppers.html
I think the falling boxes at the end are just because the container is full, but the 2 others in the middle of the simulation are definitely a problem.

Here's the bin file. http://www.hurleyworks.com/downloads/Poppers.zip

-Bird

[Julio Jerez]

Ok I have it, and it clearrly show severall bugs.

I will see whay the boxes in the midle gome out the container once they are inside and the velocity is small, that should not happens at all.
The good thing is that is happen teh same way each time.

how do you make the container and how do you make it shows in wire frame by default?

edit: oh I see, in a close up teh container is not a wire frame is made of descret primitive.
are those cylinder , capuses of polygons?

this is a good tes case for debugging.

[Bird]

They are thin cylinders. I thought that would be a good way to eliminate the tunneling problems in the first test scene I had made.

-Bird

[Julio Jerez]

They are thin cylinders. I thought that would be a good way to eliminate the tunneling problems in the first test scene I had made.

-Bird

I am assuming theye are a static collsion tree mas of small long cymders, is that right. That is what I see which is very good for debugging fo rteh simplicity,
I just want to confirm this asumtiontion is correct.