Cocos2d Tips: Design Game Objects for your game


Keep your game classes organized and logically ordered. Create a GameObject class, a GameCharacter class and then subclass these.

Its important to keep your Game Objects ordered as well as your code.  The more you order your objects, the cleaner your code will be as well.  A GameObject is anything used in a game from labels to sprites to power ups.

Its important to create a base class for all objects because, as a simple example, you want to be able to constrain all game objects to your screen.  Its not unthinkable to believe that at some point you might inadvertently cause a game object to be pushed off screen.  Thus it would be nice to have a method, something like keepObjectInBounds, to make sure you can call it on any or all objects to make sure they are not halfway or all the way offscreen.

After you build you base class, GameObject, you will subclass it to build objects.  But you can also subclass GameObject to build functionally similar types of objects like a Powerups subclass or an Enemy subclass.  For example, you might want all Powerup class objects to have a setAutoDestroyTime method that gives any object of that class a specific time to live onscreen before it disappears.

So let’s look at init methods, designated initializers, self and super.

You can create a GameObject Class in 1 file set or in different file sets.  This is where self and super take hold in your brain.  Self means, that very same class you are currently in.  Super means its super or parent class.  For example, lets say we create this hierarchy:

  1. GameObject (Everything is a GameObject)
    1. GameCharacter (Distinguish between GameCharacters and GamePowerUps)
      1. GameEnemy (Some GameCharacters are bad and some are good)
        1. GameBigBossEnemy
      2. GamePlayer
        1. GamePlayerHero 
        2. GamePlayerFriend
    2. GamePowerup

GameCharacter is a subclass of GameObject.  Which is to say, GameObject is the parent of GameCharacter.  This makes sense because your GameObjects may be characters that have a personality and animation and perform actions.  But some GameObjects will just sit there, like trees and clouds.  Clouds may be moved by wind and that is about it!  But a GamePlayer, our hero, needs to have a lot more internal logic, his own Artificial Intelligence logic if you will.

By the same token, GamePlayer may include our hero instance but it may also include a friend instance, which may be like his sidekick or something.  The logic for a friend will not include any methods to follow or attach our hero, whereas enemies will have to have that sort of logic.

This is the reason we subclass objects, in order to use a base functionality for all (all game objects must remain onscreen at all times) but give specific functionality to others (followMainPlayer method).

Lets say you’re inside GamePlayerHero and you are coding away.  If you say

[self blowHisNose];

You are calling the blowHisNose method which should exist in our GamePlayerHero class.  If it doesn’t exist in self, that object will look to its super for a corresponding method.  You don’t need to call:

[super blowHisNose];

That is what object inheritance in OOP is all about.  So if the method doesn’t exist in GamePlayerHero, the hero object looks to its parent class or super class, GamePlayer.  If GamePlayer doesn’t have that method then it will keep going until it finds that method.  If it doesn’t, your game will crash :).

So lets take a look at a practical example, an initializer.  An initializer is just another name for a Class’ init method.  Every class must have an init method.  Here is a typical one:

-(id) init {

if((self=[super init]);


return self;


This means, if this class is equal to super’s init, then do this and return self.  Since a class will always have a super class, this basically says, if I have a super class (which is always) then return myself.

Now if a child class (lets call it GameCharacter class) of this previous class (GameObject class) wants to add functionality to its parent class, then it will NOT OVERRIDE its parent class methods, it will simply add methods.  So the new GameCharacter class could just have a method like this:

-(void)checkAndClampSpritePosition {

CGPoint currentSpritePosition = [self position];

CGSize levelSize = [[GameManager sharedGameManager]


float xOffset;

if (UI_USER_INTERFACE_IDIOM() == UIUserInterfaceIdiomPad) {

// Clamp for the iPad

xOffset = 30.0f;

} else {

// Clamp for iPhone, iPhone 4, or iPod touch

xOffset = 24.0f;


if (currentSpritePosition.x < xOffset) {

[self setPosition:ccp(xOffset, currentSpritePosition.y)];

} else if (currentSpritePosition.x > (levelSize.width – xOffset)) {

[self setPosition:ccp((levelSize.width – xOffset),




Which says, Im not initializing anything (because if it does, it will erase its parent’s init method), Im just adding this method to any instance of GameCharacter AND every instance of GameCharacter will still have to be initialized from GameObject’s init.

Let’s see a more practical example.  Let’s look at an Enemy Class which can generate many different kinds of instances, all from the same code:

@interface ShipTarget : CCSprite {

int _curHp;

int _minMoveDuration;

int _maxMoveDuration;

NSString *filename;


@property (nonatomic, assign) int hp;

@property (nonatomic, assign) int minMoveDuration;

@property (nonatomic, assign) int maxMoveDuration;

@property (nonatomic, retain) NSString *filename;


@interface Asteroid : ShipTarget {




This says, Im a ShipTarget class and my instances will all be based off of CCSprite.  This means all my instances will have any methods and properties of CCSprite unless I override them (create new ones).  At the same time, I wish to create an Asteroid class which derives from this ShipTarget class and it will return an object by calling its class method named target.  Then in its implementation you have:

@implementation ShipTarget

@synthesize hp = _curHp;

@synthesize minMoveDuration = _minMoveDuration;

@synthesize maxMoveDuration = _maxMoveDuration;

@synthesize filename;


@implementation Asteroid

+ (id)target {

Asteroid *target = nil;

if ((target = [[[super alloc] initWithFile:@”Comet.png”] autorelease])) {

target.hp = 1;

target.minMoveDuration = 2;

target.maxMoveDuration = 4;


return target;



Notice again, the use of self, or target.  If there is such a thing returned from calling target’s super class, (target is ShipTarget class and ShipTarget’s super class is CCSprite) initWithFile… then set those properties and return it.  There is such a thing as target’s super class initWithFile method.  CCSprite has an initWithFile method.  Don’t believe me?  Then don’t, right click over initWithFile and select Jump to Definition and watch Xcode take you to CCSprite:

-(id) initWithFile:(NSString*)filename

 So we set the properties hp, min and maxMoveDuration.  Even though those properties don’t exist in CCSprite, they exist in the new subclass, ShipTarget.  We added them ourselves in the interface and later synthesized them in the implementation file.  Since target is of Asteroid class (which is itself type of ShipTarget), it therefore must have ShipTarget properties hp, min and maxMoveDuration.  This also means that any ShipTarget target we create, will have those 3 properties and we can set it.
This is very handy when creating the different GameObjects we discussed because they can all be based on a set of properties common to all enemies, or to all players, or to all powerups etc.  This saves you time when creating subclasses because you don’t have to create repetitive code for each object type.  And more importantly, it keeps your code tidy because if you need to add a property to all enemies, you simply add it to ShipTarget and voila…all your enemy targets will now have that property for you to assign.

Cocos2d Tips: Scrolling, Parallax and wider levels

Wider Levels & Scrolling

When you create a game in Cocos2d, your screen measures 960 pixels wide on iPhone4+ and 2048 on iPad. If we want him to move farther to the right then we need to make the level bigger.  Normally we set:

levelSize = screenSize;

But now we basically wish to make:

levelSize = CGSizeMake(screenSize.width * 2.0f, screenSize.height); // levelSize code

What we are doing is changing our scene levelSize to 2x the screenSize.

Our level is now twice as big and we have a GameScene, which contains a GameplayLayer & a BackgroundLayer with some background image.  We want to add a layer that will scroll in the opposite direction as our player moves.  To do this we will need to add a method a method called adjustLayer:

-(void)adjustLayer {

Player *player = (Player*)[sceneSpriteBatchNode getChildByTag:kPlayerSpriteTagValue];

float playerXPosition = player.position.x;

CGSize screenSize = [[CCDirector sharedDirector] winSize];

float halfOfTheScreen = screenSize.width/2.0f;

CGSize levelSize = [[GameManager sharedGameManager] getDimensionsOfCurrentScene];//1

if ((playerXPosition > halfOfTheScreen) && (playerXPosition < (levelSize.width – halfOfTheScreen))) {

// then scroll background

float newXPosition = halfOfTheScreen – playerXPosition;

[self setPosition:ccp(newXPosition,self.position.y)];      // 2 self is the game layer



Here comment //1 is a call to a method which returns the desired levelSize.  Or you can simply hard code it in this case by replacing this line with the //levelSize code line from above.  Remember we are running this method in the GameplayLayer, which is self in this code.  This needs to be called constantly so you must add this line to your update layer.  What we are doing here is keeping the screen centered on the player.  Now in the update method add this code:

//1. Get the list of objects on the layer & loop through them

CCArray *listOfGameObjects = [sceneSpriteBatchNode children];

for (GameCharacter *tempChar in listOfGameObjects) {

[tempChar updateStateWithDeltaTime:dt andListOfGameObjects:listOfGameObjects];


//2. Call the adjustLayer method

[self adjustLayer];

Perfect!  Now let’s add the image that is actually going to scroll.  Add this method to your GameplayLayer.m:

-(void)addScrollingBackground {

CGSize screenSize = [[CCDirector sharedDirector] winSize];

CGSize levelSize = [[GameManager sharedGameManager] getDimensionsOfCurrentScene];

CCSprite *scrollingBackground;

if (UI_USER_INTERFACE_IDIOM() == UIUserInterfaceIdiomPad) {

// Indicates game is running on iPad

scrollingBackground = [CCSprite spriteWithFile:@”mars_landscape2x.png”];

} else {

scrollingBackground = [CCSprite spriteWithFile:@”mars_landscape2x.png”];


[scrollingBackground setPosition:ccp(levelSize.width/2.0f,screenSize.height/2.0f)];

[self addChild:scrollingBackground z:1];


In order for this to appear, we need to call it from the init method by adding this line:

[self addScrollingBackground];

Let’s review.  addScrollingBackground adds a new image to the GameplayLayer.  (This image is positioned over the true background, which is in the BackgroundLayer.)  Therefore, the new image you add here should be a sort of rocky ground floor that will slide sideways as the player moves.

Make sure the image is twice as big as the screen so there will be enough of it to scroll through as the player moves.

Now Build & Run and watch your ground scroll differently as your background as your player moves left or right.


Parallax makes one layer scroll faster or slower than the other.  It’s quite easy to achieve this effect.  First add a new instance variable to your GameplayLayer.m file like so:

@implementation GameplayLayer {

CCParallaxNode *parallaxNode;


NOTE: We are extending the @implementation line to include an ivar.  No need to declare it in the @interface file.  Remember .h files (header files) only need to include what properties or methods you want your object to expose to other calling objects.

And now replace your addScrollingBackground method with this one:

// Scrolling 3 Parallax backgrounds

-(void)addScrollingBackgroundWithParallax {

CGSize screenSize = [[CCDirector sharedDirector] winSize];

CGSize levelSize = [[GameManager sharedGameManager]


CCSprite *BGLayer1;

CCSprite *BGLayer2;

CCSprite *BGLayer3;

if (UI_USER_INTERFACE_IDIOM() == UIUserInterfaceIdiomPad) {

// Indicates game is running on iPad

BGLayer1 = [CCSprite spriteWithFile:@”mars_landscape2x.png”];

BGLayer2 = [CCSprite spriteWithFile:@”some_rocks.png”];

BGLayer3 = [CCSprite spriteWithFile:@”some_pillars.png”];

} else {

BGLayer1 = [CCSprite spriteWithFile:@” mars_landscape2x.png”];

BGLayer2 = [CCSprite spriteWithFile:@” some_rocks.png”];

BGLayer3 = [CCSprite spriteWithFile:@” some_pillars.png”];


parallaxNode = [CCParallaxNode node];

[parallaxNode setPosition:ccp(levelSize.width/2.0f,screenSize.height/2.0f)];

float xOffset = 0;

// Ground moves at ratio 1,1

[parallaxNode addChild:BGLayer1 z:40 parallaxRatio:ccp(1.0f,1.0f) positionOffset:ccp(0.0f,0.0f)];

xOffset = (levelSize.width/2) * 0.3f;

[parallaxNode addChild:BGLayer2 z:20 parallaxRatio:ccp(0.2f,1.0f) positionOffset:ccp(xOffset, 0)];

xOffset = (levelSize.width/2) * 0.8f;

[parallaxNode addChild:BGLayer3 z:30 parallaxRatio:ccp(0.7f,1.0f) positionOffset:ccp(xOffset, 0)];

[self addChild:parallaxNode z:10];


Don’t forget to change the GameplayLayer’s init method call to:

[self addScrollingBackgroundWithParallax];

What we are doing now is adding more layers and offsetting their center positions.  Thus when they move, they will move at different rates.  Build & run the app.

You can use scrolling to add clouds or stars or alien birds to make some eyecandy for your users.  If you choose to scroll many objects on the screen though, you might want to look into reusing them.  This is a very popular technique used to improve performance of a game.  You create one instance of the object at a time and add it to an array for tracking.  When that instance moves offscreen, you remove it from that array.  This saves you a lot of memory.

Cocos2d Tips: Transitioning between game screens

Let’s learn how to create menu for a cocos2d game and how to move between a menu, the game scene, a level complete scene etc.

Eventually your game will grow and you will have to create a clear structure to keep everything organized.


As you can see, as with any app, the AppDelegate is the first object called. As such, it is the entry point to your game and will call the very first scene of your game. This means the first scene should probably the main menu which welcomes the user to the game and presents him with options.

As you can see from the Cocos2d tips “Creating a Menu”, menus can be quite complex: ‎

When a cocos2d game is created, a Singleton instance of the GameDirector is instantiated and charged with managing scenes. After all, your Main Menu scene is just another scene/layer called by the game director. Thus it is quite common to create another Singleton class used to tell the director which scene to call.

Let’s call it the GameManager and his sole job is to tell the director which scene to call. Before we take a look at some code, keep in mind a game may grow to have quite a few scenes and while text is more human readable, it is easier to manage variables as numbers. The main reason is that switch statements use numbers, not text.

So it is common practice to use code such as this in a Constants.h file that you import into your GameManager.m Class:

typedef enum {





} SceneTypes;

Now you can easily use a method such as this:

-(void)runSceneWithID:(SceneTypes)sceneID {

SceneTypes oldScene = currentScene;

currentScene = sceneID;

id sceneToRun = nil;

switch (sceneID) {

case kMainMenuScene:

sceneToRun = [MainMenuScene node];


case kFirstScene:

sceneToRun = [Scene1 node];


case kSecondScene:

sceneToRun = [PlaneScene node];//[Escena2 node];//


case kThirdScene:

sceneToRun = [Scene3 node];



CCLOG(@”Unknown ID, cannot switch scenes”);




if (sceneToRun == nil) {

// Revert back, since no new scene was found

currentScene = oldScene;




if ([[CCDirector sharedDirector] runningScene] == nil) {

NSLog(@”runningScene is nil…its ok because we are in process of setting the new scene???”);

[[CCDirector sharedDirector] runWithScene:sceneToRun];

} else {

[[CCDirector sharedDirector] replaceScene:[CCTransitionFlipAngular transitionWithDuration:0.5f scene:sceneToRun]];


currentScene = sceneID;



Now this simple switch can be used to change scenes.  All you have to do at the end of the AppDelegate is call GameManager’s sceneToRun method and pass it the MainMenuScene.


At the Main Menu scene, depending on the level selected from the menu you can pass a particular scene constant to GameManager in order to determine the level scene to call.


From within your game level layer, after defining if your player won or lost, you can call GameManager to run a custom EndOfLevel scene/layer.  Then from that transitional scene you can call the MainMenuScene again.