Search found 65 matches

by Karrok
Tue Jan 07, 2014 12:36 pm
Forum: General Bullet Physics Support and Feedback
Topic: btCompoundShape problem?
Replies: 6
Views: 11480

Re: btCompoundShape problem?

if the problem is in the drawing: I've recently updated my local bullet with a child-drawing function myself and it seems to work (tested only locally and with forklift demo) http://img202.imageshack.us/img202/7682/7tog.png if (colObj->getCollisionShape()->isCompound() ) { btCompoundShape * compound...
by Karrok
Wed Dec 04, 2013 2:12 pm
Forum: General Bullet Physics Support and Feedback
Topic: The equations Bullet Physics uses for solving physics system
Replies: 1
Views: 3835

Re: The equations Bullet Physics uses for solving physics sy

f = m * a On a more serious note, you will simply have to dig through the manual, wiki and source code. Bullet uses a number of different narrowphase solvers if I'm not mistaking. Bullet consists (among others) out of BulletCollision for the collision detection and BulletDynamics/BulletSoftBody for ...
by Karrok
Fri Nov 29, 2013 2:22 pm
Forum: General Bullet Physics Support and Feedback
Topic: Compound Objects: Get colliding child-shape's userdata.
Replies: 0
Views: 4568

Compound Objects: Get colliding child-shape's userdata.

Hi All, I have a compound shape with a bunch of child shapes, for each child-shape I also attach a small user data object containing some specific data for that child shape. As an (simplified..) example , imagine a compound shape made out of 4 tiles. Each tiles has some user data value giving a cert...
by Karrok
Mon Jun 25, 2012 12:37 pm
Forum: General Bullet Physics Support and Feedback
Topic: Randomization (repeatability)
Replies: 1
Views: 3399

Re: Randomization (repeatability)

Yes, it is deterministic (ie, given the same situation and input you should get the same result)

To randomize it, just randomize the input, in this case the forces applied to the dice
by Karrok
Fri Jun 08, 2012 8:20 am
Forum: General Bullet Physics Support and Feedback
Topic: convex hull collision shape is larger than model
Replies: 6
Views: 9363

Re: convex hull collision shape is larger than model

Nope, that has nothing to do with the scaling. Internally (in Bullet) but rather how osgBullet converts a bullet cylinder. In osgNodeFromBtCollisionShape( const btCylinderShape * btCylinder, const btTransform& trans ) the detail ratio is set to .2f You can change this to have the debug osg node ...
by Karrok
Wed Jun 06, 2012 3:27 pm
Forum: General Bullet Physics Support and Feedback
Topic: convex hull collision shape is larger than model
Replies: 6
Views: 9363

Re: convex hull collision shape is larger than model

the orientation is wrong because in your case you did not set it to suit your needs. OSGBCOLLISION_EXPORT btCylinderShape* btCylinderCollisionShapeFromOSG( osg::Node* node, AXIS axis=Y ); from the osgb collisionShapes. Just adjust the last parameter to X or Z instead of the default Y axis it takes. ...
by Karrok
Tue Jun 05, 2012 1:03 pm
Forum: General Bullet Physics Support and Feedback
Topic: convex hull collision shape is larger than model
Replies: 6
Views: 9363

Re: convex hull collision shape is larger than model

use the halfextend of your cylinder collision shape to create a bullet cylinder. ie: btCollisionShape * myCollisionShape = new btCylinderShape(osgbCollision::btCylinderCollisionShapeFromOSG( yourOsgCylinder)->getHalfExtentsWithoutMargin()); assuming your osg cylinder is the trimesh as shown in the p...
by Karrok
Fri May 25, 2012 8:33 am
Forum: General Bullet Physics Support and Feedback
Topic: Stability issue with btBvhTriangleMeshShape
Replies: 6
Views: 9832

Re: Stability issue with btBvhTriangleMeshShape

I've had some trouble with rapidly shifting contact points myself as well. Perhaps it is a good idea for you guys to submit a patch in the Bullet Issue list so Erwin can have a look at it and (if it pans out to be a good patch that doesn't break other things ;-) implement it for a new bullet version...
by Karrok
Wed Apr 25, 2012 8:04 am
Forum: General Bullet Physics Support and Feedback
Topic: General and building questions
Replies: 3
Views: 5802

Re: General and building questions

those are warnings, you can ignore them unless they are relevant to your project. You'll find the headers in the bullet/src dir for a regular build environment and the libs in the bullet/lib dir. If you make an install, the structure becomes slightly different and you will end up with a bullet_insta...
by Karrok
Mon Mar 26, 2012 8:33 am
Forum: General Bullet Physics Support and Feedback
Topic: Ghost Objects or Rigib Bodies?
Replies: 5
Views: 6948

Re: Ghost Objects or Rigib Bodies?

ghost objects retain their own contact manifold, whereas non-ghost objects have their contact information in the default contact manifold.
by Karrok
Mon Mar 26, 2012 8:31 am
Forum: General Bullet Physics Support and Feedback
Topic: General tip #1
Replies: 3
Views: 4471

Re: General tip #1

Objects that are relatively vastly different in scale are best avoided.

So modeling an entire planet and then other objects on or near the planet that can collide with each other is generally a bad idea. Either the planet would be too big or the other objects too small.
by Karrok
Wed Mar 14, 2012 7:42 pm
Forum: General Bullet Physics Support and Feedback
Topic: Ghost object internal bodies
Replies: 3
Views: 4946

Re: Ghost object internal bodies

you could do axis-aligned ray test from the object outwards. so 6 ray tests in +/- x, y, z. If all return the sphere as the first object encountered that it would be fairly safe to assume the object is enclosed in the sphere. Alternatively: Once an initial collision with the sphere happens, keep a r...
by Karrok
Tue Mar 13, 2012 3:59 pm
Forum: General Bullet Physics Support and Feedback
Topic: Soft body module virtual methods
Replies: 2
Views: 4506

Re: Soft body module virtual methods

Well, it is open source, so you can just modify it yourself. And if you feel like your changes are "production-value" (and you feel like sharing) you can submit them as patch proposal to the trunk. I also have some personal changes made to bullet which are application specific (minor thing...
by Karrok
Tue Mar 13, 2012 3:49 pm
Forum: Release Announcements
Topic: Bullet 2.80 SP1 release :OpenCL rigid bodies & Android
Replies: 17
Views: 213418

Re: Bullet 2.80 release: OpenCL rigid body pipeline & Androi

Nice release, looking forward to the full GPU implementation come 3.x!

On another note; will the GDC presentation be upped to the bullet download area soon-ish as well?

Cheers!
by Karrok
Tue Mar 13, 2012 2:26 pm
Forum: General Bullet Physics Support and Feedback
Topic: Bullet on the cloud?
Replies: 5
Views: 7089

Re: Bullet on the cloud?

the whole premise of dead-reckoning is that the client only gets position (1st grade), direction+ velocity (2nd grade) or direction + acceleration(3rd grade). On the server you run the actual physics engine + a copy of the client. Starting a simulation the objects are at the same positions. Somethin...