I have a 2D tile matrix (of walls for now) and for each wall I check if my player's next Rectangle (next as in where it will be in next frame) is colliding with that wall, and if it is, I want my player to stand right next to it, without going inside of it.
Here is the formula I came up with (this is just for left and right collision):
//*move entity
if (KState.Down(Keys.W)) en.AddForce(new Vector2(0, -10));
if (KState.Down(Keys.S)) en.AddForce(new Vector2(0, 10));
if (KState.Down(Keys.A)) en.AddForce(new Vector2(-10, 0));
if (KState.Down(Keys.D)) en.AddForce(new Vector2(10, 0));
foreach(Item i in map.Items)
{
//resolve left-right collision
if (en.NextBody(elapsedTime).Intersects(i.Body))
{
//check distance between top left corners
int distance = (int)Math.Abs(i.Position.X - en.Position.X);
//stop player in his horisontal movement
en.AddForce(new Vector2(-en.Force.X, 0));
//place player next to the wall his about to collide with
en.Position = new Vector2(en.Position.X - distance + i.Body.Width, en.Position.Y);
}
}
//stop player in place
if (KState.Clicked(Keys.Space)) en.AddForce(en.Force*-1);
en.Update(elapsedTime);
**
Which gave me this result (note that fps is not stable and that currently rectangles are squares 32x32):
Top line of text is players Rectangle, and bottom one is Rectangle of the wall that player collided with last.
Now, when the rocks (player) is moving to the left, I'm holding left key for some time, and when going to the right, I'm holding right key for some time.
Problem: As you can see in gif, when player's going left, he's moving 1 pixel to and from left wall. And when he's moving right, he bounces once, and then stays right next to the right wall. I don't want any bouncing but I don't know how to fix this in logic that I'm using.
Also, if you know good logic for how to resolve Rectangle collision so that they are not intersecting from any sides, I'd appriciate sharing it with me.
Edit 1: I managed to resolve collision to the right problem. Instead of the code above I've put
int sign = Math.Sign(i.Position.X - en.Position.X);
en.AddForce(new Vector2(-en.Force.X, 0));
//wall is right and player is left
if (sign > 0) en.Position = new Vector2(i.Body.Left - en.Body.Width, en.Position.Y);
//wall is left and player is right
if (sign < 0) en.Position = new Vector2(i.Body.Right, en.Position.Y);
but the problem with the left side collision is the same. I tried
if (sign < 0) en.Position = new Vector2(i.Body.Right + 1, en.Position.Y); and if (sign < 0) en.Position = new Vector2(i.Body.Right - 1, en.Position.Y); just in case, but the problem is the same (on "...-1..." case player gets stuck in wall).
Edit 2: As #paste requested, here is the code for NextBody(float elapsedTime) from player class.
public Rectangle NextBody(float elapsedTime)
{ return new Rectangle((int)(force.X * elapsedTime) + body.X, (int)(force.Y * elapsedTime) + body.Y, body.Width, body.Height); }
This is what's happening:
The first time the objects collide while the rocks are moving left, the algorithm works as it should. It sees that the objects are overlapping, it sets the force to Vector2.Zero, and it moves the en so that it is touching the item's right side. Good. Your object is stopped and in the right place.
What happens in the next frame is where things go wrong. You set up your force vector, and then call NextBody(). Let's just look at the X-coordinate. It gets set to:
(int)(force.X * elapsedTime) + body.X
What does this evaluate to? force.X is non-zero, and elapsedTime is also non-zero, but they're less than 1, so when they get converted to int, the result is zero. Therefore, your body is in the same place as it was before, which means the Rectangles do not overlap. This means your collision response code gets skipped and your force vector is still not zero!
Then, further down, you call your player's Update() method. I'm guessing that in that code you update the player's Position and Body. But since force is not zero, and you are using a Vector2 for Position, the x-coordinate will end up being something like 31.841304... instead of 32. If you use that number to calculate the new Body and use it in your Draw method, you'll end up drawing it at x = 31. The next time you check, it'll detect a collision, and the player will bounce out again.
There are a number of ways to fix this. What I do in my collision code is to actually move the objects first and keep a record of where they were in the previous frame. That way, the last adjustment to their position is when they are moved out of the collision area.
Related
In short: How do I make a scrolling gameScene which is NOT infinite?
I'll try to explain what I want to achieve with an example: Hill Climb Race
In this game you drive a car (or any sort of crazy vehicle, actually ;)) up a hill.
Now there is one particular thing about the game that I can't get my head around:
It's pretty obvious that the tracks of each individual stage are NOT laid out randomly. i.e. the course of the track is always the same each time you play it.
What I want to learn is:
How do you create a scrolling game scene like that? Is it a huge background node which gets scrolled or is there some sort of fancy tiling involved?
My game needs to scroll both axis (x,y). The players node starts in the center of the game area and can be moved around. There are some obstacles spread around in the area, some of which are not visible initially, because they lie at the edges of the 'game world'.
I suppose the easiest solution would be to use a big background node, but how will that affect the memory consumption of the game?Thanks for your help!
We built something like this into our SKATiledMap. The trick is to add the object you want to follow to the background you want to scroll. This will also keep the background on screen.
-(void)update
{
if (self.autoFollowNode)
{
self.position = CGPointMake(-self.autoFollowNode.position.x+self.scene.size.width/2, -self.autoFollowNode.position.y+self.scene.size.height/2);
//keep map from going off screen
CGPoint position = self.position;
if (position.x > 0)
position.x = 0;
if (position.y > 0)
position.y = 0;
//self.mapHeight*self.tileWidth gives you the size of the map in points
if (position.y < -self.mapHeight*self.tileWidth+self.scene.size.height)
position.y = -self.mapHeight*self.tileWidth+self.scene.size.height;
if (position.x < -self.mapWidth*self.tileWidth+self.scene.size.width)
position.x = -self.mapWidth*self.tileWidth+self.scene.size.width;
self.position = CGPointMake((int)(position.x), (int)(position.y));
}
}
self in this case is the background and autoFollowNode is the players. You could just use self.size.width instead of self.mapHeight*self.tileWidth Hopefully that makes sense and his helpful.
I have a BIG problem!! I want to make something like old space shooter games like “Asteroids”, where the ship, when going out of the screen, is reappearing at the other side. For example, when the ship go out at the top of the screen, it come back at the bottom. But in my game, there’s a camera following the player, showing only a quarter of the world, and I want to simulate an infinite world this way! Here’s a picture showing what I mean :
What I thought doing was simulate the scroll by only moving game objects, stored in an array, but not the player, and calculating at every frame if the objects are out of the world boundary and re-adding them at the other end of the world (i.e going out at the left would add it back at the right).
But I don’t really like that way of doing… I’d like something more… intuitive..?
Do you guys have any idea of how doing it? Like, any tutorial on the web or just the right words to explain what I mean so I could do a bright research on google (I’m french, so I had a really hard time writing that question)!
Thank you in advance!
Knowing the players position, x,y check to see if the player is within the bounds of the area. If the player has left those bounds, set the player's position to the opposite bounds, possibly adding width/height. I'll do it here for the Y coordinate only:
const float minWarpAreaY( 0 );
const float maxWarpAreaY( 400 );
//if (player.y < minWarpAreaY) { player.y = maxWarpAreaY - player.height; }
//if (player.y > maxWarpAreaY) { player.y = minWarpAreaY + player.height; }
if (player.y < minWarpAreaY) { WarpPlayer(0.0f, (maxWarpAreaY - minWarpAreaY)); }
if (player.y > maxWarpAreaY) { WarpPlayer(0.0f, -(maxWarpAreaY - minWarpAreaY)); }
void WarpPlayer(float amountX, float amountY)
{
player.x += amountX;
player.y += amountY;
for (eachObject in World)
{
eachObject.x += amountX;
eachObject.y += amountY;
}
}
Something along those lines should help.
I've been reading about AABB collision. I have a pretty rough implementation of it at the moment. I am now able to detect when two rectangles intersect on both the X and Y axis.
To calculate the depth of intersection on X, I basically do:
overlapX = (Game1.p[0].boundingBox.halfWidth + Game1.p[1].boundingBox.halfWidth)
- (Math.Abs(Game1.p[0].boundingBox.center.X - Game1.p[1].boundingBox.center.X));
Right now I am just testing the depth on X as I want to see where the problem is. I am not taking the direction, left or right, into account. I am just assuming the p[0] object is colliding on the left side of p[1].
Then for the actual movement and collision detection:
public void Move(float x, float y)
{
Position.X += x;
Position.Y += y;
overlapX = (Game1.p[0].boundingBox.halfWidth
+ Game1.p[1].boundingBox.halfWidth)
- (Math.Abs(Game1.p[0].boundingBox.center.X
- Game1.p[1].boundingBox.center.X));
overlapY = (Game1.p[0].boundingBox.halfHeight
+ Game1.p[1].boundingBox.halfHeight)
- (Game1.p[1].boundingBox.center.Y
- Game1.p[0].boundingBox.center.Y);
if (Collision.testAABBAABBX(
Game1.p[0].boundingBox, Game1.p[1].boundingBox) &&
Collision.testAABBAABBY(
Game1.p[0].boundingBox, Game1.p[1].boundingBox))
{
if (overlapX < overlapY) Position.X -= overlapX;
}
}
Now when I collide with p[1], overlapX reads 1. So position.X should -= 1. However, here is where the problem begins. Here is an image during the overlap. You can see that there is indeed an overlap shown by a vertical yellow line:
So my problem is that when I instruct the position.X to move back by the overlap value, it does not, at least not until I move the rectangle either up or down. If I do that, it magically corrects itself:
Any clues on what's going on? Maybe it's right in front of me and I'm not seeing it, but the response only seems to trigger after the next movement.
Let me know if more info is needed.
This happens because your collision resolution code is placed within your motion code, which doesn't seem to run unless you make a move.
The solution is to move your collision resolution code outside your Move method into a separate method that is called by Update every frame.
From what I have read, in Box2d, you get the angle of a revolute joint with the GetJointAngle function, but when trying to set the angle the member m_referenceAngle is protected. Can the angle not be programmatically set?
I found that I can apply the angle from one joint to another body as:
float FirstAngle = firstArmJoint->GetJointAngle();
secondArmBody->SetTransform(b2Vec2((750.0/PTM_RATIO),(520.0f+100)/PTM_RATIO),hourAngle);
I put this in ccTouchesMoved so that when the user drags the first object (from which FirstAngle is retrieved) the second object (secondArmBody) is also moved.
What happens is that the second body rotates at the top of the image and not at the anchor point.
Any ideas?
SetTransform() can be used to set the position and rotation of a body. This happens completely independently of any joints on the body. For example, if you want to make sure a body is perfectly upright at a given moment, you can call
body->SetTransForm(body->GetPosition(), 0);
passing 0 as the angle value (upright). I've never tried this for a body with a joint on it, but I doubt it would work properly.
When I ran into the problem of having to make a revolutejoint point at a certain angle, I solved it by enabling motor on the joint and adjusting the motor speed until the angle matched the one I wanted. This simulates realistic motion of the joint. Example:
Creating the joint
b2RevoluteJointDef armJointDef;
armJointDef.Initialize(body1, body2,
b2Vec2(body1->GetPosition().x,
((body1->GetPosition().y/PTM_RATIO));
armJointDef.enableMotor = true;
armJointDef.enableLimit = true;
armJointDef.motorSpeed = 0.0f;
armJointDef.maxMotorTorque = 10000.0f;
armJointDef.lowerAngle = CC_DEGREES_TO_RADIANS(lowerArmAngle);
armJointDef.upperAngle = CC_DEGREES_TO_RADIANS(upperArmAngle);
_world->CreateJoint(&armJointDef);
Pointing the joint
float targetAngle = SOME_ANGLE;
b2JointEdge *j = body->GetJointList();
b2RevoluteJoint *r = (b2RevoluteJoint *)j->joint;
if(r->GetAngle() > targetAngle){
r->SetMotorSpeed(-1);
} else { r->SetMotorSpeed(1); }
Basically, you see what side of the current angle the target angle is on, and then set the motor speed to move the joint in the correct direction. Hope that helps!
http://www.box2d.org/manual.html
I'm currently trying to make a basic platformer with XNA and I'm wondering how to create a "jumping effect." I currently have basic keyboard input which allows for sideways movement, but I would like my sprite to slowly progress into a jump rather than instantly teleporting there (right now I have something like Rectangle.Y += 40 every time I jump, making the sprite instantly appear there). Does anyone have any insight?
I'm not totally across how to implement this in XNA/C#, but in Flash games I've made I just added a vertical velocity property. I'll try write everything as C# as I can..
Example; create the velocity property:
float verticalVelocity = 0;
Vertical velocity should be constantly reduced (by gravity). Set up a gravity property somewhere accessible from your player:
float Gravity = 2.5;
And in your update() method for the player, increment the verticalVelocity by Gravity. Also increment the Y position of your player by the verticalVelocity. This will simulate falling:
verticalVelocity += Gravity;
Position.Y += verticalVelocity; // this may be -= in XNA, not sure where the y axis beings
When you hit a surface, the velocity should be reset to 0.
And finally, to jump, simply subtract a given value from verticalVelocity:
public void Jump(float height)
{
// Only jump if standing on a surface.
if(verticalVelocity == 0)
verticalVelocity -= height;
}
You'll eventually want to add gravity and possibly other forces to your game, so I highly recommend you save yourself a lot of pain and implement some kind of basic force system. This can be done using Vector2s, as you can just add them to the speed of your character. Then just apply an instantaneous force to your character to push it up.
If you really don't want to use a physics engine, you can make a Vector2 with the high point of the jump for the Y and the characters X, and then use the Vector2.Lerp method to interpolate between the characters position and the end point of the jump.
This is generally a very bad system to use, and I highly recommend you either use an existing physics engine, or make your own simple one.
use a sinusoidcode should look something like this:
float ground = 0.0f;
float angle = 330.0f;
jump(){
if(ground == 0.0f)ground = Rectangle.Y;
if(Rectangle.Y <= ground)
{
Rectangle.Y+=Math.Sin(angle/(Math.Pi*180));
angle++;
}
}
You can accurately create a gravity effect if you modify the ySpeed dynamically, as opposed to just adding 40.
You want to declare a ySpeed
ySpeed = 0;
Then you want to use an acceleration variable
acceleration = 0.25;
Okay, now that we've done that, let's add gravity, provided that our player isn't touching the floor.
if(playerLocationY + playerHeight > floorLocationY)
{
gravity = false;
}
else
{
gravity = true;
}
if(gravity)
{
ySpeed += acceleration;
}
Now that we've got that down, we want to include something that allows us to jump.
if(KeyPressed == UP)
{
ySpeed -= acceleration;
}
This will move our player in the upward direction
We now want to make sure we actually move, so let's add one last line and we're done.
playerLocationY += ySpeed;
Congratulations, you made it.