Pong Collision Detection question

Embrace_It

New Member
Reaction score
9
Atm, I'm making a Pong Clone. While building the game up from strach, I'm reading articles about collision detection. The following is the collision detection from Aaron's Game Programming Tutorials:

Code:
// Check to see if the ball is going to hit a paddle
bool CheckBallCollisions(Entity& paddle)
{
    // Temporary values to keep things tidy
    int ball_x = g_Ball.screen_location.x;
    int ball_y = g_Ball.screen_location.y;
    int ball_width = g_Ball.screen_location.w;
    int ball_height = g_Ball.screen_location.h;
    int ball_speed = g_Ball.y_speed;

    int paddle_x = paddle.screen_location.x;
    int paddle_y = paddle.screen_location.y;
    int paddle_width = paddle.screen_location.w;
    int paddle_height = paddle.screen_location.h;

    // Get which paddle we're checking against
    if ( paddle.screen_location.y == PLAYER_Y)
    {
        // Check to see if ball is in Y range of the player's paddle.
        // We check its speed to see if it's even moving towards the player's paddle.
        if ( (ball_speed > 0) && (ball_y + ball_height >= paddle_y) &&
            (ball_y + ball_height <= paddle_y + paddle_height) ) // side hit
        {
            // If ball is in the X range of the paddle, return true.
            if ( (ball_x <= paddle_x + paddle_width) && (ball_x + ball_width
                 >= paddle_x) )
            {
                return true;
            }
        }
    }
    else
    {
        // Check to see if ball is in Y range of the computer's paddle.
        // We check its speed to see if it's even moving towards the computer's paddle.
        if ( (ball_speed < 0) && (ball_y >= paddle_y) && (ball_y <= paddle_y +
              paddle_height) )
        {
            // If ball is in the X range of the paddle, return true.
            if ( (ball_x <= paddle_x + paddle_width) && (ball_x + ball_width >=
                  paddle_x) )
            {
                return true;
            }
        }
    }

    return false;
}

Now assume that we move the ball, then check for collisions, and that the ball has a speed of 10 pixels/second towards the right. Also, imagine that the ball is currently 9 pixels away from a paddle (in the x-direction). The next time we move the ball, this function registers a collision, since the ball moves "into" the paddle. Upon returning from the function and seeing that it returned true, we inverse the xspeed to make it go in the opposite direction, so the next time we move the ball, it is moving towards the other end of the screen.

What I don't understand is that you would see the ball bounce off 9 pixels too early, right? Correct me if I'm wrong, but if this is the case, what would/could be a solution? Maybe set the ball to move the 9 pixels instead of 10? Then afterwards move it normally with the right speed?
 

Samael88

Evil always finds a way
Reaction score
181
Have you changed the ball widht in any way? Or the ball position compared to where the image is at?

Those are the most common misstakes I do when it comes to these things.
 

Slapshot136

Divide et impera
Reaction score
471
the way that you described it yes, the ball would bounce a bit early, but this check should run fairly fast (30+ times a second?) so that it's not so noticeable

a slightly better way would be to calculate that the ball would move 9 to the right, hit the wall, and then move 1 to the left, basically calculate where the ball should be as a result of the collision as well as what it's new trajectory should be
 

Embrace_It

New Member
Reaction score
9
@ Sameal88: I actually havent made the ball graphic yet :p but I'm taking into account the width of both the image and the ball itself.

@ Slapshot136: I was playing around with that idea, but I'll have to take into account that it could hit a wall as well, and suppose the ball is 5 pixels away from a paddle, the it would stay in the same spot (moving 5 to the left, then 5 to the right) and then start moving in the opposite direction.

Collision detection sure is hard :p
 

Samael88

Evil always finds a way
Reaction score
181
Pong collision could be made fairly easy you know:p

If you want the ball to travel in a constant value, all you need to do is to invert the values every time the ball hits something:p
Like, if it hits the upper wall you would want the Y speed value inverted and the, if it's hit's a paddle you would want the X speed value inverted. It is fairly easy to do when I come to think about it:)
Even tho that would give a rather predicable game:p

I can't seem to find anything in that code that could cause it. Have you checked that your not putting the code in the wrong order?

Like this for example:
Move();
Collision();
Show();
That would be a horrible misstake to do:p
 

Embrace_It

New Member
Reaction score
9
Pong collision could be made fairly easy you know

Yeah :p

Even though that seems to be true, lets assume that the call order is:

Collide();
Move();
Update()/Show();

Moving at 10 (x- and y-speed) the ball is 9 pixels away from the top wall, and is moving towards it. Now because 9 - 10 < 0 this would make Collide() return true, and we would change the sign of the y-speed. Now should the update function render the ball just on the edge (max(y-yspeed, 0))? Or should it do nothing? The next time the ball moves it should be going in the right direction.

Am I making sense here or am I missing something?
 

Samael88

Evil always finds a way
Reaction score
181
If the ball is moving at 10 pixels, then I don't see why you have the edge so that the ball is able to get 9 pixels from it? I would just move the wall that 1 pixel if that could. Does that work?
I always try to find the easiest possible way to fix a problem just not to make it to complicated.
 

Slapshot136

Divide et impera
Reaction score
471
moving the wall.. sounds like a bad idea, sorry, the wall will not be in the way 99% of the time, this is only for when there is a collision, otherwise the ball will travel like normal

using the time and speed that the ball is going find the position, and if there is an obstacle in the way, then find the time it would take to reach the obstacle, after which you would have a new time and new speed, and you would re-calculate the position that the ball should be in (after bouncing)

i hope i worded that in a way so that it will be easy to code
 

Samael88

Evil always finds a way
Reaction score
181
@Slapshot: That could actually work, but that will be a bit more complicated than it sounds.

Then you would need to basically check it pixel by pixel.

Btw, I have found out what might be the error:)

Let's see if I get this right:
Do you by any chance have the collision and the movement together?

If you do you probably have this:
if (ball hits wall)
move ball this and that.

if (ball does not hit anything)
move ball this and that.

then it would move the ball twice if it hit the wall, thus making your problem.

Other than that it should not be any possible problems to it what I can think of right now:)

Usually it should be this btw:

1. balls is 9 pixels from the wall.
2. ball moves into the wall, and the function changes the direction of it.
3. Ball moves 9pixels away from the wall.

Either you do that, or you move the wall 1 pixel and the ball should be able to bounce away. This could cause a bug to appear that makes the ball go into the wall, but that can be worked around with raising the "move per second" and lowering the "pixel moved" ratios, if you know what I mean:)
 

Embrace_It

New Member
Reaction score
9
@ Slapshot136: I did think about doing it that way, but I figured that it would be a pain in the *** to code considered what you have to take into account when the ball hits just right or left of a corner. Imagine if the new trajectory and speed would cause the ball to bounce again.

@ Samael88: Sorry, but I haven't started coding that part yet, but I can assure you that it would not be a problem if I had :) I see what you mean by "moving the wall". I can set the wall's width and height to a number like 60 which, if the ball's speed is 10 would never create the situation that we are discussing. One problem though, is the fact that I intend to increase the ball's speed slowly over a period of time, which would mean that I have to increase the speed by at least 5 every time.

Well, I think we are getting there, so I'll try out a few things as soon as the game has been developed enough for me to do so, and I'll see how it goes. Ofc, feel free to feed me some more great ideas ;)
 

Samael88

Evil always finds a way
Reaction score
181
I just had a great idea:) I don't know if it has been brought up before, but here is goes:

Say you have a speed of 10.
First you need to check the distance from the wall, if it is less than ten then you could use a certain function for that scenario.

Say the distance is 5 from the wall, still a speed of 10.
first you calculate where the ball will hit the wall after moving a distance of 5, then you take the projection you get there and use it to calculate where the ball will be 5pixels later:)

I hope you get what I mean:)
 

Embrace_It

New Member
Reaction score
9
Hi Samael88,

Sorry for not answering for such a long time, but I've been busy :)

I think that Slapshot136 already proposed a similar idea, and it seems like a complicated idea. In calculating the future position of the ball after a collision with something, I have to bear in mind that the ball can almost instantly collide with something else. A wall and then a paddle for example.

I've thought long and hard about it, and I've decided to try and make the game scene so that a strange situation like this will never happen. That is, I'm gonna make the width and height of the playing area so that it will always be 10 pixels away.

As always, thanks for your great amounts of help and advice. I'll be sure to post the finished project here when it's done :thup:
 

GooS

Azrael
Reaction score
154
Havn't read all code nor all posts but a simple rect collision detection should suffice:

Code:
bool rectCollisionDetect(args)
{
bool within_x_range = (ball.x > paddle.x && ball.x < paddle.x + paddle.w);
bool within_y_range = (ball.y > paddle.y && ball.y < paddle.y + paddle.h);
if(within_x_range && within_y_range)
   return true;
return false;
}

using a rect struct

Code:
struct rect{
  int x;
  int y;
  int h;
  int w;
};

w = width
h = height

shouldn't it? If you don't have a rect struct then you should be able to get x,y,h,w values some other way, even if it's but as a magic number somewhere
within the code.

Edit:

Read the real question now, sry. Well if it's the left paddle you should take its (x+width) - ball.x on collision and move the ball to that x, however that would alter the speed and it might look abit odd, depending on how fast the ball actually is moving, it might not even be noticable.

//==GooS
 
General chit-chat
Help Users
  • No one is chatting at the moment.
  • Ghan Ghan:
    Howdy
  • Ghan Ghan:
    Still lurking
    +3
  • The Helper The Helper:
    I am great and it is fantastic to see you my friend!
    +1
  • The Helper The Helper:
    If you are new to the site please check out the Recipe and Food Forum https://www.thehelper.net/forums/recipes-and-food.220/
  • Monovertex Monovertex:
    How come you're so into recipes lately? Never saw this much interest in this topic in the old days of TH.net
  • Monovertex Monovertex:
    Hmm, how do I change my signature?
  • tom_mai78101 tom_mai78101:
    Signatures can be edit in your account profile. As for the old stuffs, I'm thinking it's because Blizzard is now under Microsoft, and because of Microsoft Xbox going the way it is, it's dreadful.
  • The Helper The Helper:
    I am not big on the recipes I am just promoting them - I use the site as a practice place promoting stuff
    +2
  • Monovertex Monovertex:
    @tom_mai78101 I must be blind. If I go on my profile I don't see any area to edit the signature; If I go to account details (settings) I don't see any signature area either.
  • The Helper The Helper:
    You can get there if you click the bell icon (alerts) and choose preferences from the bottom, signature will be in the menu on the left there https://www.thehelper.net/account/preferences
  • The Helper The Helper:
    I think I need to split the Sci/Tech news forum into 2 one for Science and one for Tech but I am hating all the moving of posts I would have to do
  • The Helper The Helper:
    What is up Old Mountain Shadow?
  • The Helper The Helper:
    Happy Thursday!
    +1
  • Varine Varine:
    Crazy how much 3d printing has come in the last few years. Sad that it's not as easily modifiable though
  • Varine Varine:
    I bought an Ender 3 during the pandemic and tinkered with it all the time. Just bought a Sovol, not as easy. I'm trying to make it use a different nozzle because I have a fuck ton of Volcanos, and they use what is basically a modified volcano that is just a smidge longer, and almost every part on this thing needs to be redone to make it work
  • Varine Varine:
    Luckily I have a 3d printer for that, I guess. But it's ridiculous. The regular volcanos are 21mm, these Sovol versions are about 23.5mm
  • Varine Varine:
    So, 2.5mm longer. But the thing that measures the bed is about 1.5mm above the nozzle, so if I swap it with a volcano then I'm 1mm behind it. So cool, new bracket to swap that, but THEN the fan shroud to direct air at the part is ALSO going to be .5mm to low, and so I need to redo that, but by doing that it is a little bit off where it should be blowing and it's throwing it at the heating block instead of the part, and fuck man
  • Varine Varine:
    I didn't realize they designed this entire thing to NOT be modded. I would have just got a fucking Bambu if I knew that, the whole point was I could fuck with this. And no one else makes shit for Sovol so I have to go through them, and they have... interesting pricing models. So I have a new extruder altogether that I'm taking apart and going to just design a whole new one to use my nozzles. Dumb design.
  • Varine Varine:
    Can't just buy a new heatblock, you need to get a whole hotend - so block, heater cartridge, thermistor, heatbreak, and nozzle. And they put this fucking paste in there so I can't take the thermistor or cartridge out with any ease, that's 30 dollars. Or you can get the whole extrudor with the direct driver AND that heatblock for like 50, but you still can't get any of it to come apart
  • Varine Varine:
    Partsbuilt has individual parts I found but they're expensive. I think I can get bits swapped around and make this work with generic shit though
  • Ghan Ghan:
    Heard Houston got hit pretty bad by storms last night. Hope all is well with TH.
  • The Helper The Helper:
    Power back on finally - all is good here no damage
    +2
  • V-SNES V-SNES:
    Happy Friday!
    +1

      The Helper Discord

      Members online

      No members online now.

      Affiliates

      Hive Workshop NUON Dome World Editor Tutorials

      Network Sponsors

      Apex Steel Pipe - Buys and sells Steel Pipe.
      Top