attaching dynamic parts to each other

Anybody know how to attach 2 separate, dynamic objects to each other? Whether I try parenting them to each other, or to separate bones in a common armature, they fly apart when I start the game engine (2.33a).

Why do I want to do this? I’m building a flight sim, and I want to spin the propeller by applying torque to it, not just setting its angular velocity. I want the propeller and fuselage to detect collisions with everything except each other. Can it be done? :-?

U can not do it.

It is possible with python, but it’s not perfect. I don’t really know exactly how to do it though.

Pooba

Just use polyheader on 2.33a for complex things.

Random

What do you mean?

What do you mean?[/quote]
you can set the bounds of an object beneath the actor buttons

polyheader uses your mesh

you can set things up so that two meshes collide like they have a pivot or ball joint using polyheader bounds, but it is very very likely they will come apart [particularly if you try to set up something with a large number of links, like a chain]

True, but not here!

Actually, they used programming, and not Blender.

Still, dismount is really cool!

Random

What do you mean?[/quote]
you can set the bounds of an object beneath the actor buttons

polyheader uses your mesh

[/quote]

So what cone, box, sphere, cylinder do?

The format still the same in the 3D view, a sphere.

of those box is the only useful one

it a box of the size of your mesh aligned with your mesh’s axes

the cylinder and cone aren’t any good because their proportions can’t be changed [at least from my experiments]

Cylinder & Cone take their radius from the ‘size’ field (like the sphere) but their height from mesh (like box)

Cylinder should be useful for stuff like wheels.

What do you mean?[/quote]
you can set the bounds of an object beneath the actor buttons

polyheader uses your mesh

you can set things up so that two meshes collide like they have a pivot or ball joint using polyheader bounds, but it is very very likely they will come apart [particularly if you try to set up something with a large number of links, like a chain][/quote]

Actually, setting the bounds to polyheader on both objects was one of the first things I tried, but the objects still repelled each other. I also minimized the ‘size’ parameter, but that didn’t help either.

When you suggest setting things up like they have a “pivot or ball joint”, do you mean partially enclosing one mesh within the other? What I did was make a shaft (i.e. cylinder mesh), then wrap the hub of the propeller around it. Effectively, it’s a cylinder inside a donut . Are you saying I can prevent the donut sliding off the pencil, e.g. by adding a couple of ‘collars’ at the ends of the pencil?

Actually, setting the bounds to polyheader on both objects was one of the first things I tried, but the objects still repelled each other. I also minimized the ‘size’ parameter, but that didn’t help either.[/quote]
if they aren’t parent and child they shouldn’t come apart instantly unless they intersect

yes

but again, I would not reccomend doing this [or even depending on any multi-object system in a game] simply because of how likely it is for it to come apart.

it is fun to play with, but unless you get like 300fps [200 is nice, but not good enough] constantly it will not stay together for more than a few minutes

does that mean u can do ragdoll physics?

yes

but it is very likely it will fall apart, particularly with low framerate

[it also is very difficult to control the amount a given joint moves, and the direction it moves in]

Yeah, just e-mail the Dismount people for ragdoll physics.

Random

what?

dismount was done using ODE, which has significant multi-body capabilites

ODE was being integrated into blender [it is still in tuhopuu] but joints weren’t really added. A point joint [I think] was added after 2.33 came out with the game engine, but I wouldn’t expect much to happen for the other joint types very quickly.

[also, until recentlyish ode in blender was slower than it should have been]

hhmmm well i put together a few test’s in blender 2.33a and it runs down the engine like hell my tests never do that so maby i experiment with the new blender when it comes out hopefuilly that would be soon. would it?

Actually, setting the bounds to polyheader on both objects was one of the first things I tried, but the objects still repelled each other. I also minimized the ‘size’ parameter, but that didn’t help either.[/quote]
if they aren’t parent and child they shouldn’t come apart instantly unless they intersect

yes

but again, I would not reccomend doing this [or even depending on any multi-object system in a game] simply because of how likely it is for it to come apart.

it is fun to play with, but unless you get like 300fps [200 is nice, but not good enough] constantly it will not stay together for more than a few minutes[/quote]

OK, you talked me out of it, I’ll simulate the desired mechanics some other way. I am, however, curious to learn why framerate has anything to do with it. I expect this is a rank newbie question in the game engine field, but how does a slow frame rate mess up collision detection?

because objects move further on a given frame when the framerate is lower they are more likely to pass through each other

[particularly if you have a small tolerance for error, a 1 meter box can stick through the floor by probably 2cm without people noticing much; a pivot for an elbow joint would probably break if it moved 2cm]

high velocities and low framerates are to be avoided for reasonable collision detection

google for pyode if you want to do joint physics through python scripts.

basically you would have your object not use blender physics but instead run a script to continuously retrieve the position/orientation from pyode’s world that you created.