Weekend Updates - Indexing and Graphics

A hodge-podge of generally undirected game updates took place over the weekend which I wanted to write about.  Now that the in-game menu functionality is working I'm getting close to having nothing but a handful of random details to cleanup before the engine and game are actually playable entities.

Indexing

One thing I found that I overlooked in my originally programming of the engine was the fact that, since the player is movable, the player may at some points be in front of an object while at other points might be behind an object.  The problem came in when I realized I was not re-ordering the array of on screen objects, so when a player walked behind an object, the entire player sprite was still drawn on top of the sprite for the object which they were supposed to be behind.

On the surface this isn't a major problem and I was able to write up a sorting function to pass to Array.sort which ... sorted things out.  Writing up the function however brought a larger issue to light.  In order to get objects which were literally on top of other objects (ie, a coffee cup on top of a table), I had to introduce a gameHeight property to each game object.  This property I used to indicate how far off the ground something was assumed to be and my sorting function could take this into account when calculating an objects relative position.

This may end up being sufficient for this game, but I feel like the engine in the end needs to more robustly handle the fact that I'm essentially working in a 3d environment but drawing in a 2d context.  For instance, I may end up reworking the object positioning mechanisms so that their position is defined in 3d space and the engine does the work of calculating where to map those objects to in 2d space for presentation purposes.

Graphic Work

I've been putting off drawing more sprites, largely, because I got so sick of it after making all the sprites for the room itself.  This weekend however I decided to hunker down and start work on the panda walk cycles again.  It was as much of a pain as I expected it to be, but I got the forward walk cycle done and am now moving onto the left walk cycle (which will be repeated as the right walk cycle).

All of this graphic work has led me to the conclusion that, for my next game, I need to just pay someone to do this for me.

I still have some cleaning to do, but here's the general idea of what the final walk cycle will look like

So, what is left to be done?  I'll try to get a list together for a future post, if for no other reason than to provide myself with a checklist of sorts.

DnL8Tar

Game Play Decisions - In Game Menu

I've spent time over the last couple months deliberating on (and soliciting opinions concerning) alternatives to the Panda-Town engines current game play scheme.  Right now, when a player clicks on an object in the game, the engine assumes that the player wants to interact with the object.  The engine looks up the appropriate interaction based on the objects current state and then executes that interaction.

While simple, this scheme has a couple drawbacks which have become quite obtrusive in my game design.  First, it limits me to defining one interaction per object per state.  This means, for example, that if I want an object such as a coffee cup to be something the player can pickup, I can not also have it be something the player is able to look at since I can only define one interaction.  A second drawback is that a player clicking on an object may not be in fact attempting to interact with it, they may simply be trying to walk towards it.

I've decided that the play scheme needs to be updated to allow for multiple types of interactions and to force a player to indicate that they are actually attempting to interact with an object.  The direction I am going to take in implementing the first component of the change is to cause a small in game menu to open when a user chooses to interact with an object which exposes more than one type of interaction.  For instance, if a player can either "look" at or "pick up" an object, when the player indicates that they want to interact with the object, they will see a menu with these two choices.

The general idea

Deciding on the best scheme for the second component of the change has proved to be somewhat more cumbersome.  This is due in large part to a restriction I have placed on myself, limiting input to left mouse clicks.  I have placed this restriction on myself in order to allow games made with the engine to be played on devices such as an iPad.  Currently I'm planning on having a player indicate that they wish to interact with an object by double clicking on the object, but I have not settled on that yet.

Side Note: An interesting consequence of the first component of the change being made is that it makes it easier to code interactions from a distance (ie, allowing a player to look at an object from across a room without having to walk up to it).

Side Note 2: I'm readily accepting feedback from anyone who is a point and click game enthusiast (or who has at least played one in their life) concerning the control scheme I've indicated

Canvas 2d Context and Frame Swapping

While it may be a consequence of the programming languages of which I was reading, every game programming tutorial I have ever read introduces the concept of a buffer frame and frame swapping fairly early on.  The idea is conceptually simple:

You establish two frames, the first you are currently showing to the player, the second you are actively drawing on.  Once you are finished drawing the second frame, you swap the first frame with the second such that the second frame is now the one being shown to the user and the first is free to be used for the drawing of the next frame.

When I first started down the path of learning the 2d drawing context on the canvas,  I spent a good couple hours searching for the HTML5 canvas equivalent of frame swapping and buffering.  I came up empty handed.  That being the case, as is my way, I decided to just start coding, leaving the buffering problem as one to be solved later.  What I found out, much to my happy surprise, is that there is no need for this practice within the 2d context.  

As you draw to the canvas using the 2d drawing context, what you draw does not appear on the canvas immediately.  This is opposed to drawing in, let's say, Java.  Consider the following bit of pseudo code:

function drawStuff() {
  c = get Canvas;
  c.drawImage( something-great.png, 20, 250 );
  c.drawImage( something-else-amazing.png, 52, 17 );
  c.drawImage( a-so-so-image.png, 189, 442 );
}

If you were to code this in Java and watch it render, you would see something-great.png drawn to the screen first, followed by something-else-amazing.png, and then a-so-so-image.png.  There would be a delay between the appearance of each of these.  And even thought said delay might be barely perceptible, it would be there, and if you are drawing a hundred images to the screen each frame, you would start to notice it and it would become quite irritating.  This is why some form of frame swapping is needed in this case.  You don't want to show the user your frame until all the things you are going to place on it are rendered.

Contrast this to the 2d drawing context.  The way Javascript interacts with the 2d drawing context is such that nothings is actually visibly drawn to the canvas until the scrip doing the drawing has finished executing.  That is to say, looking at the example above, nothing new will be drawn to the canvas until the Javascript interpreter has reached the end of the function (assuming there are not other functions to run after that which are going to do more drawing).  You could argue that this in and of itself is an implementation of frame swapping however the implementation is baked into the browser itself so even if you are making a game engine you don't need to think about it. 

Additional Note of Potential Ignorance

I admit there may very well be a way to achieve the same result in Java without frame swapping.  I haven't done a lot of game programming in Java so I don't know all the magic secrets.  And certainly, many Java game engines hide this detail, however if you are MAKING the game engine you need to worry about it.

Game Engine Design - A Free-For-All

If you Google "game engine design best practices" or "game engine design techniques," you get a real melange of results, few of which provide much immediate value outside the satisfaction of personal curiosity, on the off chance that you are curious about the design of a particular game engine.  Since the engine I'm building / have built is my first one (well, probably third or fourth, but first to progress to the point of actually being able to run a game) I'm no expert on the subject, but I've learned enough to find that a game engine is a very single purpose beast.

This singularity of purpose leads to the lack of "best practice" in the domain.  While I'm sure if I took a series of classes on game design and game engine design I would be fed some "best practices," or at least, design patterns, the creation of a cookie cutter game engine, while theoretically possible, would be such a boondoggle it would rival if not trump the practice of low-level-design in traditional software programming.

Simply put, what works for Final Fantasy is not going to work for Angry Birds, and the effort of trying to make a one size fits all game engine, and coding to the engine, would be considerably greater than coding two game engines. Since I write all this as a personal opinion from the perspective of someone who has minimal experience in the field of game development,  I will be providing no hard evidence to support my rambling, so take my statements for what you will, however the direct correlation between the complexity of usage of a framework / engine and the level of flexibility afforded by said framework / engine has been demonstrated time and again (and I'm sure if you enter a magic series of keywords into Google you'll find some evidence to that effect).

And now for something actually interesting, another demo sprite.  Enjoy

a Panda, standing, in a v-neck