Collide2, how to use it ?

Game Mechanics
#1

I know it’s supposed to take two space IDs, but where do you get the space IDs from ? spaces has a getId() method
that returns a SWIGTYPE_p_dSpaceID thing !?

#2

Would still be nice to know but not so important anymore. Just tried “quickStep” and that makes the simulation about a million times faster.

#3

Does anyone really not know this ?

#4

[quote]I know it’s supposed to take two space IDs, but where do you get the space IDs from ? spaces has a getId() method
that returns a SWIGTYPE_p_dSpaceID thing !?
[/quote]
Yes, that’s the ID.

To improve the API, we could have a lookup table of ID’s to Space’s then add a method which takes the normal Space object instead.

Want this added to the todo list?

Will.

#5

I’m not sure I understand. Does this mean you can’t use this function now ? I cant call the collide2 with SWIG_ID’s.

Yes a collide2(space,space) should deff be on the todo list.
But does this mean there’s no solution to this in the current version ?

Thanks, Newbton

#6

WD, have you any update on the progress on quickStep() ? It looks as if this method is so much faster than step(), and allmost as accurate. But it doesnt work well with jointmotors, no matter how much force you apply to them they are very weak. Accuracy is good, but there must a bug in force parameters. My ragdoll can barely stand with a gravity of 0.4 using quickStep, even if I apply a force of 1 million to the jointmotors (I’m using jointballs with jointstops)

#7

I had the same problem… I ended up finding a good combination of dParamFudgeFactor, dParamCFM, dParamStopERP, and dParamStopCFM for the joints. I converted CFM and ERP into spring and damping parameters, which made tuning easier.

#8

if there is a bug in quickstep, maybe we should submit it to to the ODE mailing list.

Will.

#9

Are you using

public void collide2(int o1, int o2)

of JavaCollision?

Have you tried space.getId().getSwigCPtr() to get those integers? That is what collide(Space) does.

Will.