Calculating how much to move to some position

For Beginners

Started by toony June 30, 2011 03:50 PM

17 comments, last by ApochPiQ 12 years, 9 months ago

Author

June 30, 2011 03:50 PM

I wrote some lines of code to calculate the exact number of pixels needed to increment the position of moving objects (each time around the game loop) in order to get the object to some "x, y" position like "224, 329".

Here is my aproach:



// float orig[2], dest[2]; // [0] = x, [1] = y positions



// path calculation

if (dest[0] >= dest[1]) {

	orig[0] += dest[0] / (dest[1] + (dest[0]-dest[1]));

	orig[1] += dest[1] / dest[0];

} else {

	orig[0] += dest[0] / dest[1];

	orig[1] += dest[1] / (dest[0] + (dest[1] - dest[0]));

}

Is this approach clear enough and efficient in your opinion? The code isn't 100%, it's just a glimpse of the full code but this part is the juicy part !!

This code would be used to move a player by clicking on the ground where you want to move, like you see in games like baldurs gate for example.Try this program to see what the code does.

ApochPiQ

23,138

June 30, 2011 04:55 PM

I really don't follow your math here.

The typical formula is this:

new_position = old_position + (speed * (target_position - old_position))

Repeat on X, Y, Z axes.

Your math seems to exhibit weirdness, and mixes axes, which I honestly can't unravel...

Wielder of the Sacred Wands
[Work - ArenaNet] [Epoch Language] [Scribblings]

Nanoha

2,682

June 30, 2011 05:00 PM

Its a bit hard to get whats going on, somehting liek this is easier to follow:

float dir[2];

dir[0] = dest[0] - orig[0]

dir[1] = dest[1] - orig[1];



// You could clamp them to limit speed or something

dir[0] = clamp(dir[0], -1, 1); // max or 1 pixel in x direction

dir[1] = clamp(dir[1], -1, 1); // max or 1 pixel in y direction

// obviously you can now go faster diagonally



orig[0] += dir[0];

orig[1] += dir[1];

I'd suggest implementing a basic Vector class with a few functions, subtraction, addition, scaling and normalising. Then you can do:

float maxSpeed = speedPerSecond * time; // time since last frame

Vector dir = dest - orig;

dir.normalize();

dir *= maxSpeed;

orig += dir;

Interested in Fractals? Check out my App, Fractal Scout, free on the Google Play store.

toony

Author

June 30, 2011 05:30 PM

i find it strange that to me your code looks hard to understand. In my code i'm just trying to perfectly move something to another position. And speed/velocity variables contain values that will normaly move ahead of the destination i need. I'll take a closer look at your codes and see where that gets me. And my code might seem weird, but it works and for some reason my code looks easy to figure out for me.

ApochPiQ

23,138

June 30, 2011 05:33 PM

If all you want is to arrive at a destination, why not just do:

new_position = destination;

And be done? ;-)

Wielder of the Sacred Wands
[Work - ArenaNet] [Epoch Language] [Scribblings]

toony

Author

June 30, 2011 05:34 PM

I dont want to teleport there!

haegarr

7,408

June 30, 2011 05:39 PM

Err ...

...
// float orig[2], dest[2]; // [0] = x, [1] = y positions // path calculation if (dest[0] >= dest[1]) { orig[0] += dest[0] / (dest[1] + (dest[0]-dest[1])); orig[1] += dest[1] / dest[0]; } else { orig[0] += dest[0] / dest[1]; orig[1] += dest[1] / (dest[0] + (dest[1] - dest[0])); }

orig[0] + dest[0] / (dest[1] + (dest[0]-dest[1]))
= orig[0] + dest[0] / (dest[1] + dest[0] - dest[1])
= orig[0] + dest[0] / (dest[0])
= orig[0] + 1
and similarly
orig[1] + dest[1] / (dest[0] + (dest[1] - dest[0]))
= orig[1] + 1

Hence this is not "clear code". Perhaps it is efficient, but only if the compiler is smart enough to find the shorter version.

toony

Author

June 30, 2011 05:48 PM

I really don't follow your math here.

The typical formula is this:

new_position = old_position + (speed * (target_position - old_position))

Repeat on X, Y, Z axes.

Your math seems to exhibit weirdness, and mixes axes, which I honestly can't unravel...

new_position = old_position + (speed * (target_position - old_position))
step1: new_position = 100 + (1.0f * (338-100))
step2: = new_position = 338
Now im flying!

So much useful info in this forum! mostly from the mod team...

Nanoha

2,682

June 30, 2011 06:25 PM

new_position = old_position + (speed * (target_position - old_position))
step1: new_position = 100 + (1.0f * (338-100))
step2: = new_position = 338
Now im flying![/quote]

target_position - old_position needs to be normalized before scaling for speed, it'll end up being eithe -1, 1 or 0 which will work out ok with speed. It makes more sense if these are vectors rather than just numbers.

new_position = old_position + (speed * normalize(target_position - old_position))
step1: new_position = 100 + (1.0f * normalize(338-100))
step2: new_position = 100 + (1.0f * normalize(238))
step3: new_position = 100 + (1.0f * 1))
step4: = new_position = 101

if speed was 10, you'd be at 110 now which makes sense.

Interested in Fractals? Check out my App, Fractal Scout, free on the Google Play store.

toony

Author

June 30, 2011 06:36 PM

Ok the code will cap the number to the nearest -1 to 1, but the code will cause the object to move in zigzag form. Thats why my method is important to me. Well i have to have some sort of a divider variable to make sure the object moves in a strait line. I must have something that holds the number of increments it will take to get to the desired destination.I already have it working just wanted to see other approaches, not silliness.

Calculating how much to move to some position

This topic is closed to new replies.

Popular Topics

Recommended Tutorials

Calculating how much to move to some position

This topic is closed to new replies.

Popular Topics

Recommended Tutorials

Reticulating splines