Gravity

[jellysnake]

So I made a little prototype to test out some Gravity related code, along with a few other things, and then it kinda got a little carried away with unneeded options and stuff. So yeah, what the hell may as well share it.
Have fun, all instructions are shown when you run the file.

Gravity.love

Gravity, "Now with bug fixes!"

(3.58 KiB) Downloaded 153 times

p.s. Any and all feedback as to code conventions, tips or anything else is highly welcome.

[undef]

Backspace causes a crash, sim.lua line 84: attempt to get lenght of 'varPlanets' (nil).

Do you use Euler integration to calculate the orbits?
If so, and you're still interested in playing around with gravity I can recommend having a look at fourth order Runge-Kutta integration:
http://en.wikipedia.org/wiki/RK4

I just found a nice explaination (far better than I could ever write it probably) on why it is superior here:
http://gamedev.stackexchange.com/questi ... nt-gravity

I implemented it a while back, this might be easier to read than the pure mathematical approach:

Code: Select all

local function rk4( object, gameSpeed )
	local k1, l1, k2, l2, k3, l3, k4, l4
	local _mass, _position, _speed, _toOrigin = object.mass, object.position, object.speed, object.toOrigin

	k1 = gameSpeed * _speed
	l1 = gameSpeed * accelerationToOrigin( _toOrigin        , _mass)

	k2 = gameSpeed * (_speed + l1/2)
	l2 = gameSpeed * accelerationToOrigin( _toOrigin + k1/2 , _mass)

	k3 = gameSpeed * (_speed + l2/2)
	l3 = gameSpeed * accelerationToOrigin( _toOrigin + k2/2 , _mass)

	k4 = gameSpeed * (_speed + l3)
	l4 = gameSpeed * accelerationToOrigin( _toOrigin + k3   , _mass)

	local position = _position + k1/6 + k2/3 + k3/3 + k4/6
	local speed    = _speed    + l1/6 + l2/3 + l3/3 + l4/6

	return position, speed
end

The local variables k1, l1, and so on only serve readability. You could also do it all in one big ugly term.

[jellysnake]

1. Bug is fixed. Thanks, thought I had fixed that.

2. I implemented the gravity using this line of code:

Code: Select all

varBalls[w][1]:applyForce((varPlanets[q][1]:getX()-varBalls[w][1]:getX())*(varPlanets[q][2]:getRadius()/10), (varPlanets[q][1]:getY()-varBalls[w][1]:getY())*(varPlanets[q][2]:getRadius()/10))

or simply:
Force to apply on the x axis is equal to the distance between the ball (on the x)and the planet multiplied by the strength of the gravity.
Same for the Y axis but obviously y instead of x.
Or even simpler using Newton's law of universal gravitation;
http://en.wikipedia.org/wiki/Newton's_l ... ravitation

I'm not using the other implementations I have seen that are using (I think) Euler integration like you talked about. However that function will come in very hand for something along the same lines that I am making in a different language.