I know this is an ugly tack on, but maybe you could define a pctimer for the weapon and update it's position relative to the parent periodically.
Maybe you could also use iJoint (although I haven't had luck with this yet).

solved now... turns out due to the big enough timesteps of ode and fairly uniform cube... it was being treated like a sphere (hence it was rolling about on point and spinning). making one dimension quite small compared to others solves this problem.

a game does not use 'renders', it needs 3D geometry to render in real-time.

Most likely you can use cinema 4D if you can export to max or blender formats (or use a conversion program - usually freely available on teh internet).

why not have two different scenes, and a bit of mist inbetween?

also certain ipo blocks can be 'baked' into an animation block, which can be multiply combined to make your animation. (Look under Action Editor as well)

after a conversation with Jorrit, he managed to explain to me I should transform the local vector to world space using Other2ThisRelative(csVector), so now I'm using something like:

Code:

void AddForce(csRef<iPcMechanicsObject> pcmechobj , csVector3 move)
{
move = pcmechobj->GetBody()->GetTransform().Other2ThisRelative(move);
pcmechobj->AddForceDuration(move,false,csVector3(0,0,0),0.2);
}

bool Behaviour::SendMessageV(const char *message , iCelPropertyClass *pc , celData &ret ,
iCelParameterBlock *params , va_list args)
{
if(strncmp(message , "pccommandinput_" , 15) == 0)
{
csRef<iPcMechanicsObject> pcmechobj = CEL_QUERY_PROPCLASS_ENT(entity , iPcMechanicsObject);
if(!pcmechobj)
return false;

if(!strcmp(message+15 , "forward1"))
AddForce(pcmechobj , csVector3(0,0,-25));
else if(!strcmp(message+15 , "forward_"))
AddForce(pcmechobj , csVector3(0,0,-2.5));

else if(!strcmp(message+15 , "backward1"))
AddForce(pcmechobj , csVector3(0,0,25));
else if(!strcmp(message+15 , "backward_"))
AddForce(pcmechobj , csVector3(0,0,2.5));

else if(!strcmp(message+15 , "strafeleft1"))
AddForce(pcmechobj , csVector3(25,0,0));
else if(!strcmp(message+15 , "strafeleft_"))
AddForce(pcmechobj , csVector3(2.5,0,0));

else if(!strcmp(message+15 , "straferight1"))
AddForce(pcmechobj , csVector3(-25,0,0));
else if(!strcmp(message+15 , "straferight_"))
AddForce(pcmechobj , csVector3(-2.5,0,0));

else if(!strcmp(message+15 , "rotateleft1"))
pcmechobj->SetAngularVelocity(csVector3(0,-1,0));
else if(!strcmp(message+15 , "rotateleft0"))
pcmechobj->SetAngularVelocity(csVector3(0,0,0));

else if(!strcmp(message+15 , "rotateright1"))
pcmechobj->SetAngularVelocity(csVector3(0,1,0));
else if(!strcmp(message+15 , "rotateright0"))
pcmechobj->SetAngularVelocity(csVector3(0,0,0));

return true;
}
return false;
}

But this still behaves weirdly! Albeit less so, it will sometimes take a leap into the air (at first I thought this was from the rotation being dirtied by other objects, but it turns out it is not), and when I rotate round sometimes the axis directions seem to change randomly in relation to each other (although usually remaining on the xz plane), sometimes I have occasionally caught it looping back round as per the previous behaviour.

It seems doing this is no different to using

Code:

pcmechobj->AddForceDuration(csVector3(0,0,-25),true,csVector3(0,0,0),0.2);

Since using true when I've calculated the local->world converted vector has the same effect as when I use the above case but with false (i.e without mesh rotation affecting force's direction), except in the odd case when I was moving back and I caught it leaping in the air.

Code:

if(!strcmp(message+15 , "forward1")) {
csVector3 move(pcmechobj->GetBody()->GetTransform().Other2ThisRelative(csVector3(0,0,-25)));
pcmechobj->AddForceDuration(move,false,csVector3(0,0,0),0.2);
}

// same as

if(!strcmp(message+15 , "forward1"))
pcmechobj->AddForceDuration(csVector3(0,0,-25),true,csVector3(0,0,0),0.2);

yeah I also had this problem in the linux build.

Hi,

I am trying to code a behaviour for the main controllable object using CEL

Code:

if(strncmp(message , "pccommandinput_" , 15) == 0)
{
csRef<iPcMechanicsObject> pcmechobj = CEL_QUERY_PROPCLASS_ENT(entity , iPcMechanicsObject);
if(!pcmechobj)
return false;

// trying to compute actual force forwards based on transformation of body
if(!strcmp(message+15 , "forward1"))
pcmechobj->AddForceDuration(movement,false,csVector3(0,0,0),0.2);
else if(!strcmp(message+15 , "forward_"))
pcmechobj->AddForceDuration(csVector3(0,0,-2.5),false,csVector3(0,0,0),0.2);

else if(!strcmp(message+15 , "backward1"))
pcmechobj->AddForceDuration(csVector3(0,0,25),false,csVector3(0,0,0),0.2);
else if(!strcmp(message+15 , "backward_"))
pcmechobj->AddForceDuration(csVector3(0,0,2.5),false,csVector3(0,0,0),0.2);

else if(!strcmp(message+15 , "strafeleft1"))
pcmechobj->AddForceDuration(csVector3(25,0,0),false,csVector3(0,0,0),0.2);
else if(!strcmp(message+15 , "strafeleft_"))
pcmechobj->AddForceDuration(csVector3(2.5,0,0),false,csVector3(0,0,0),0.2);

else if(!strcmp(message+15 , "straferight1"))
pcmechobj->AddForceDuration(csVector3(-25,0,0),false,csVector3(0,0,0),0.2);
else if(!strcmp(message+15 , "straferight_"))
pcmechobj->AddForceDuration(csVector3(-2.5,0,0),false,csVector3(0,0,0),0.2);

else if(!strcmp(message+15 , "rotateleft1"))
pcmechobj->SetAngularVelocity(csVector3(0,-1,0));
else if(!strcmp(message+15 , "rotateleft0"))
pcmechobj->SetAngularVelocity(csVector3(0,0,0));

else if(!strcmp(message+15 , "rotateright1"))
pcmechobj->SetAngularVelocity(csVector3(0,1,0));
else if(!strcmp(message+15 , "rotateright0"))
pcmechobj->SetAngularVelocity(csVector3(0,0,0));

return true;
}
return false;

As you can see, I am using iPcMechanicsObject::SetAngularVelocity() to turn the object (since I am using the physics engine this is the only way of rotating the object I know of).

Code:

if(!strcmp(message+15 , "forward1"))
pcmechobj->AddForceDuration(movement,false,csVector3(0,0,0),0.2);

Like this the object moves fine, only rotating has no affect on the movement axis of the object (I can see the orientation of the mesh changing but it still always moves forwards relative to the world).
If I change the false to true (so it applies the force to the objects axis), the behaviour when I turn and go forwards is extremely strange... it will arc and spiral of just move in small circles on the spot... sometimes you can see if moving up and down in a sin wave as it moves forwards.

So, then I though something strange must be going on here so I tried this instead

Code:

if(!strcmp(message+15 , "forward1")) {
csVector3 movement(csVector3(0,0,-25) * pcmechobj->GetBody()->GetTransform());
pcmechobj->AddForceDuration(movement,false,csVector3(0,0,0),0.2);
                }

but now the behaviours just seem to have been reversed, so obviously Im missing something fundamentally important.

http://the-grid.sourceforge.net/rec.tar.gz   hopefully all you need to do is type 'make'

What can I do to make the object move in the direction I rotate it?

Thanks for your time.