next »

[encoder]

so i am playing around with box2d and think how would it be like to make some soft body physics models (there are quite a few out there). and found my self creating a set of circles in a circular pattern (the usual cos-sin equation).

i had 16 circles...
not enough

32..
still not enough
124,...
good framerate

at this point it was just: no more joint systems and i really had to find out what my system can do. and the next logical step was to create a another circle to run trough them.

so i left the function at 124 and duplicated it another 10x4 times.
and here is what 5k balls (+1 big ball) look like (5 iterations).



how accurate box2d is?
if those upper and lower tales resemble a soft curve, i say it might have some scientific applications.

now i am rerunning the test with 60 iterations. if i don't post it in a few moments then i accept donations for a new PC.

[encoder]

still running on one code and eating up about 1/2Mb ram.
my advice:

don't ever ever never ever ever put this kinda shit in a game. not until 2020.

[encoder]

k.
the tails are finer and there is a clear shock-wave visible. however there is a lot of clustering, maybe due to the imprecise nature of double floats. it was fun playing with my ballz (5+ k of them), but now i get back to my soft body. the next thing i will try to rig something like a plate with breakable random strength joints simulating crystal based materia like metal, or diamond (if the joints are not so random).