Support NeoGAF

[Ghost]

Probably a stupid question ahead (im sure i learnt this in school but have since forgotten):

Is there a way to find a point on a line given how far you want to move down it...

I need to move the agent towards the target position at say 4 pixels a turn, preferably in a straight line.

I had a ratio sum (using the difference between the current position and the target position to divide up the 4 pixels per axis) which works but the players run in curves not straight lines.


Any help greatly appriciated, mocking also welcome.

 

[Dilbert]

I'm really confused by your question.

If you want to write the equation of a line which contains (15, -15) and (500, -300), you can do that using the point-slope formula:

(y - y0) = m(x - x0)

where m = the slope of the line (the change in y divided by the change in x) and the fixed point is (x0, y0).

In this case, you get (y + 15) = -(285/485)(x - 15). You can do a little algebra to put that into a more standard-looking form.

What confuses me is your mention of "pixels." Pixels exist only in the x- and y-directions, not in an arbitrary direction. You can't move four "pixels" down a line. You can start at a point and move some unit of distance, but then you need to "round" that ending location to correspond to the nearest pixel for display.

 

[Ghost]

Sorry i was getting confused between math and programming when i was writing, that's why theres no minus on the 15's as well.


Thanks for the help, ill try this.

 

[Ghost]

Wait, how does this help me find a point on a line given a distance i wish to travel?

Never mind i got it, thanks.

 

[Ghost]

Ok thanks, thats actually what i had before but i see what the problem with it is now (cant see how to fix it, but thats another problem)

 

[Dilbert]

In general, if you start at a point and want to specify your next position by giving a distance and a direction, you want to be working in polar coordinates. (I'll be using the @ symbol for "theta," which is the angle expressed as a clockwise rotation from the polar axis.)

If you're at some point (x1, y1) and want to figure out where you'll end up after going a distance r in the direction @, you can use the Cartesian-polar transformations to help you:

x' = r cos @
y' = r sin @

(x2, y2) = (x1 + x', y1 + y')

The trouble, of course, is figuring out what @ is, but it's the best coordinate system for handling the kind of vectors you want to use.

 

[slayn]

you seem to be having problems with more than just the mathemtical formula... care to elaborate?

 

[Ghost]

Its cool, the point-slope algorithm works, but it creates curves over long distances, that wont be a problem for my simulation (hopefully) because the target position will be changing all the time so curves wont really show. It's syntax also changes depending on the coordinates, so i have to have a bunch of if statements to account for the possible different slopes, so looks a mess. Which is why i thought there must be a more straight forward way.

Thinking about how to implement finding the direction of my vectors made my head hurt (It was the first thing i thought of a few days ago) and i have to demo this working monday so ill stick to the messy solution and hope inspiration strikes next week sometime.