Box2D Forums

G'day!

Firstly, let me thank 'MayoButter' (Jesse Sternberg) for his selfless efforts and community helping/sharing spirit.
Thanks to Mr. Erin Catto, too.

I am still a n00b at WCK and my knowledge of physics is pretty limited.

Alright, let me now move on to a problem that I am facing with collisions.

1) Supposing I draw a 'box' on the screen and let it drop on a stationary 'box' lying on the ground, sometimes the falling box just goes 'inside' the stationary box and then bounces back. Instead of bouncing at the very first touch (with the stationary box) it does at times bounces a wee bit late after it has gone inside of the stationary box. I am not sure if I am doing something wrong.

2) Also, sometimes, if the box that is drawn on the screen is really small or thin, and if it falls on a box that is rotating on a 'joint' positioned at the center of the box, the falling box doesn't register any contact whatsoever with the rotating box (this does happen intermittently though)

I have managed to get a printscreen of both the problems and have attached it with this post. I haven't modified anything in these pictures except adding the text.


Thank you,
Macoy

Hi i might be able to help with this one. The box2d physics are calculated per timestep, so for example i have drawn a diagram, the top circle shows what happens with low iterations, the bottom circle what would happen with high iterations (using isbullet for example).



Play about with positioniterations, velocityiterations and the timestep, adjusting the scale could also help too if your only using small shapes. You can find these in the components inspector on a world instance.

Put the following,



on your critical objects. I'm sure there's a higher cost for this sort of prioritization, so it's best to figure out which objects you should put this on (maybe just your spinning objects instead of on each of 40 balls that could be hitting those 4 or 5 spinning objects).

@engiebenjy:
Thank you Sir, for your reply.

I did try to mess around with those 4 odd inspectable variables but I couldn't seem to find the right balance. It did seem like it worked okay to an extent but after doing a more thorough testing, those issues still lingered.

@burtonposey:
putting 'bullet = true' worked like a charm!
Thank you.