Search found 4221 matches

by Erwin Coumans
Mon Jul 25, 2005 2:46 pm
Forum: General Bullet Physics Support and Feedback
Topic: raycast through the sweep and prune space
Replies: 6
Views: 17686

raycast through the sweep and prune space

Although Gino van den Bergen's explanation of my idea is better and much more detailed, here is the original posting on comp.graphics.algorithms: http://www.erwincoumans.com/p/raycast_on_sweep_and_prune.pdf It is a very efficient way to do raycasting on dynamics scenes. It works best for very long r...
by Erwin Coumans
Mon Jul 25, 2005 12:40 pm
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: How to create the contact manifold ?
Replies: 15
Views: 47266

In my case during the TOI calculation. Then contacts are maintained or not from one step to the other based on the small positive distance between the contact features. Nothing fancy . sounds quite fancy to me though. Because how do you deal with the 'timestap' of the contacts ? During the TOI calc...
by Erwin Coumans
Sat Jul 23, 2005 9:35 pm
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: How to create the contact manifold ?
Replies: 15
Views: 47266

I think tesselation is not a bad thing. You automatically get rolling resistance. This was done with barrels in Half Life 2. Interesting. So how do you deal with all the contact points ? For a cylinder you can easily get loads. Do you have a post-processing stage to reduce them ? If so, how do you ...
by Erwin Coumans
Sat Jul 23, 2005 7:27 am
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: Constraint Solver approaches
Replies: 11
Views: 24651

In the limit of small time-steps the implicit scheme will converge towards the right behavior. The problem is that we graphics people use insane high time-steps. Try using something like 1e-3 - 1e-5 and the behavior would be much more convincing (of course you can forget about realtime interactive ...
by Erwin Coumans
Fri Jul 22, 2005 9:37 pm
Forum: General Bullet Physics Support and Feedback
Topic: New Bullet demo binaries for windows
Replies: 0
Views: 5553

New Bullet demo binaries for windows

http://www.continuousphysics.com/p/bullet0.1demos.zip The raycaster demo shows minkowski sum of a cube and a cylinder. The cube is rotating, so the minkowski sum deforms. Other demos show islands for deactivation, time of impact between two rotational+translational moving objects (*), gjk with visua...
by Erwin Coumans
Fri Jul 22, 2005 9:24 pm
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: How to create the contact manifold ?
Replies: 15
Views: 47266

Here's what I do for box-box collision: 1. Find the separating axis with minimum overlap. In this determination, I bias against edge-edge contact. 2. If the minimum axis is edge-edge, I just generate one point. 3. If the minimum axis is face-(vert | edge | face), I determine the reference face acco...
by Erwin Coumans
Fri Jul 22, 2005 9:10 pm
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: How to create the contact manifold ?
Replies: 15
Views: 47266

To device algorithms for case analysis I have from time to time found it very helpfull and instructive to think in terms of external voronoi regions. Once you have a good understanding of these, you simply need a method to prune away external voronoi regions, which are of no interest. I agree. I'm ...
by Erwin Coumans
Fri Jul 22, 2005 9:06 pm
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: Constraint Solver approaches
Replies: 11
Views: 24651

This is just an implicit time-stepping scheme. If you do not include a TOI kind of term on the right-hand side vector you will see objects bounce before they actually hit anything. I wouldn't be so sure about that: Set the restitution to 0. Let the collision tolerance where you start picking up/bre...
by Erwin Coumans
Fri Jul 22, 2005 9:33 am
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: How to create the contact manifold ?
Replies: 15
Views: 47266

In my opinion stable and robust contact point generation is a topic that could use some attention. I agree, its once of the reasons to set up this forum. continuous collision detection is another one, and all the extra bits that are hardly spoken out :) For algorithms such as GJK or VCLIP I have su...
by Erwin Coumans
Fri Jul 22, 2005 9:23 am
Forum: Research and development discussion about Collision Detection and Physics Simulation
Topic: Constraint Solver approaches
Replies: 11
Views: 24651

In your dissertation, you describe a method for applying velocity constraints to contact points that are not actually touching. Perhaps this would eliminate some of the jitter I was seeing from the overshoot. Erin Wouldn't this result in a visible soft collision ? When I tried the approach of apply...
by Erwin Coumans
Thu Jul 21, 2005 8:05 am
Forum: Physics authoring tools, serialization, standards and related topics
Topic: ColladaMaya physics export XML
Replies: 0
Views: 6275

ColladaMaya physics export XML

Although work in progress, there is upcoming support for Physics specification in the next COLLADA XML schema 1.4.0. The Collada Maya exporter has early supports for this: http://colladamaya.sourceforge.net/ Although not up-to-date, there is a document about Collada Physics: http://www.feelingsoftwa...
by Erwin Coumans
Wed Jul 20, 2005 8:17 pm
Forum: Physics authoring tools, serialization, standards and related topics
Topic: new bullet gameblender, 20 july 2005
Replies: 0
Views: 5793

new bullet gameblender, 20 july 2005

http://blendertestbuilds.kidb.de/index.php?dir=Windows/bf-blender/ErwinC/&file=blenderBullet2005july20_vc8.zip I disabled the deactivation (which means objects stay active forever, with visible jittering) This should stop the objects from deactivating too early. Tuning of deactivation will be pr...
by Erwin Coumans
Wed Jul 20, 2005 11:29 am
Forum: Physics authoring tools, serialization, standards and related topics
Topic: Linux bulletblender available
Replies: 0
Views: 5892

Linux bulletblender available

Thanks to bandoler, who did a lot of work, getting the scons files to work, tracking down a bug and creating a linux release build: http://blendertestbuilds.kidb.de/index.php?dir=Windows/bf-blender/ErwinC/&file=blender_bullet_linux.tar.bz2 demos http://blendertestbuilds.kidb.de/index.php?dir=Win...