I’ve been doing stress tests. Features i’ve noticed:
Great stability, heavy on light works pretty well, for cubes of equel densities, a 64:1 mass ratio is barely stable resting with the large top cube on the bottom small cube. With the top cube falling on the bottom cube, 8:1 is stable, 15.625:1 is really pushing it.
Also, small face collision (a pinball on a static wire ramp, for example) has greatly improved. I did a stress test with a pinball two wire ramp with a dip in the middle as the mesh with quads sized about 0.03*0.18 with two colliding “pinballs” (ramp inelestic, pinballs 0.8 elestic), and it worked beautifully except there was some shaking when the balls were resting against eachother before they deactivated (fell “asleep”). There are less stabiltiy issues with inelastic block towers on a bouncy floor (stress test: 24 standard (2by2by2) cubes is about the stacking limit on a 0.9 elesticity floor (meaning that the stack de-activates instead of falling), a cube stack falling 0.5 units to the floor stays intact if the stack is about 10 or less high).
2.Compound objects: they are simulated stabily, even with linked objects lying on top of eachother. (my tests used 7 228 boxes in the shape of a digital 8).
Notes on compound objects: The parent’s center is always the center of mass of the compound object, the mass of the object is the mass of the parent, ignoring the children; the same goes for materials. When making a compound object, avoid nested children (children of children).
The speed is a bit lower than 2.42’s physics, but that’s a small price to pay compared to the UBER improvements.
I only ran through the test blends as they are for interest sake since I am not into making games.
I was impressed with the way it all works and I noticed it was much better than the last time I tried it out. Well done Erwin.
Trying to get the man over the high jump had me RAOFLAO
One thing I did notice in the car demo was that the suspension settings are too soft - kind of disconcerting to see the wheels some up through the body around corners etc- also I note that although it appeared that the wheels are UV mapped with a proper pic it doesn’t seem to be used which is a pity cos the model is quite good stuff though
My “stress tests” didn’t test the limets in terms of cpu, memory, max number of objects, etc. They tested certain things that don’t work well in 2.42, and tested the realism of physics.
Damn, I manged to get my hopes up, but there’s no Win32 build or i386 for my linux installation either. I guess I’ll have to wait or try to get one of the CVS builds.
RCRuiz did the userinterface for rigid body constraints, which makes it easier to author ragdolls, water mills, gears, hinged doors, without requiring python programming. Also the new ‘Show Physics Visualization’ in the Game Menu should help solving issues.
I have a question about Blender physics, and I am at loss finding a better place to post than this thread. If you have any idea who to ask, please direct me in the right direction.
Can Blender be used to design real world kites? How accurate is the Blender physics model? Is it modeled after real-world physical laws?
OR Is it just generalizations to get realistic-looking effects, or can Blender be used to design, for example kites?
I want to design a aerofoil-kite, but dont want the hassle of having to build dozens of prototypes. I was thinking designing it in a 3D-RAP/CAD would remove the need for building prototypes, and the ready design with suitable phsical characteristics could simply be sent to a sailmaker.
Would RCRuiz know answers to these q’s? I know he is an important figure, hehe should I say his time is valuable, and I dont want to bother him for simple stuff like this.
sorry but what you want is not what you could use Blender for.
you will NOT get around making prototypes. I cannot say that there are not complex
engineering application which could simulate it because all use is the typical Alias, Rhino,Cobalt combo but I highly doupt that there is software with that feature because so much needs to be simulated.
the air pressure which will push up the kite and the rope which stretches, material weight, aerodynamics, and and and.
most CAD/CAM have a material stress simulation but not and interactive animation.
i do not know how the physics in blender are integrated.
but that does not matter in this case here.
for what you want you need an engine which can calculate the wind lift forcing the kite to lift. for that the engine even needs to know where the gravity center is. how the wind will move the kite. how the kite’s mass will affects its rotation in space.
i am not sure if a simple wind tunnel analyses would help you. rhino has a nice stress analysis tool but it is totally inferior to stress calculations in real engineering applications.
it isn’t that simple as you think.
blenders 3d engine is tweak for fast performance. they use a heavily simplified approach to calculate physics. a ball has a simplified force boundary because otherwise the forces would need to be calculated along a very fine mesh to secure even calculations.
i rather say get a good book about building kites and just do it.
stuff though 
I guess I’ll have to wait or try to get one of the CVS builds.