Compound collision and center of mass

[carli2]

I have a question:

When you have materials lighter than air, like the hot air of a bimp, this moves the center of mass out of the body. Am I right?
Or how would I simulate that the body of the blimp is lighter than air? (it does not have a negative mass, dosent it?)

[PJani]

it should not move com out of body, com is always contained witheen the body aabb.
hot air baloon...you need to use bouancy equation in force callback if you want to do such thing. tip: gravity force - bouancy force.
bouancy always points oposite of gravity force and is the same thing as is in the water -> only densities are different.

i am currently working on something like that but without properly working compound(density) my project is at halt(cant continue without working compund).

[PJani]

Correction: the last added sub collision sets mass and com of whole body.

[Julio Jerez]

the inertia of the whole compound is set whn ethe function AddRemove End called.

yes the com is always inside the aabb

to set the inerta of a shape of the diffrent densities yo ucna have and auxialry compound tha is a copy o fthe original compound.
the on the auaxilaty shape se the scale of the equavalen body usin the raltion densisty = mass / volume
this will give you a shape of dioffner siace but with same mass,

basically a lig air ballum you make the ballom shape,
then make a copy of teh origulan shape

the calculate the mass of each sub shape,
by using the density of te hvallume the new the scale will be he new density, which will make a smalle shape of the density is ligher tha air
now you have a shape of same mass, but dirrent volume, which will make a body with diffrent inertia
the use thi new shape top set thebody inertia

this is no totally acurate because a ballome is not a solid body it a thing body which mean it is a body wi a large inertia since is concetrate on the surafce, is thsi property tha make balloms so resistance to rotata
it is the same as a beach ball and balling ball. beach ball rotate very litle.

an better way to do this is to use two balloms one small tah the other, and sustrat the inerta of the first for the inertia of the secund.
and use the result as the inertia of the body.

[PJani]

ok i am a bit confused. Does this work right with multiple subcollisions or i am missing something?

NewtonBodySetMassProperties( m_body->getNewtonBody(), ci.density * ci.volume, NewtonCompoundCollisionGetCollisionFromNode(m_compound, ci.node));

EDIT:

Oh, i see 0_0 i thought this sets per sub collision mass...silly me

[Julio Jerez]

it shoud be working with compound, all of teh work is competed for inertia. I test with few shapes and it works.
The original bug was beacsue I was applying the central thoerem to inertias that waer not at the center of mass, that gnerated incorrect values, after I fixed that it is all working now.

i do no see what the problem is.

[PJani]

Sorry it was my mistake, i was sure that with NewtonBodySetMassProperties you set per sub collision mass not the mass for whole body!


So you are saying i should make another copy of compound, localy scale its sub collisions and use NewtonBodySetMassProperties for non homogeneus density?

[PJani]

Okey i was testing compound and sometimes it happens that some bodies get lodged in between two sub collisions of my compound. These two sub collisions are 0.0 units(they are touching each other) apart from each other.

[Julio Jerez]

These two sub collisions are 0.0 units

what this means?

[PJani]

With that i meant there is no space in between sub collisions...they are sharing same verticles at that face!...of course i mean 2 convex sub collisions

[Julio Jerez]

can you post an image?

[PJani]

I will try to make screenshot when it happens.

[PJani]

Actually is more worse than i thought. Bodies at certain area are just going through sub collision. I will make a video.