Learning to build a game from scratch
A few days back, I got a chance to be a part of a team to make something at the Mumbai Hackathon. Amidst all the more worldly projects, we had decided on making a game. I knew nothing about game-building, and all that was apparent was that you had to take care of more than the usual number of conditions; there’s simply a lot more states that even a simple game can be in. To me, it was intimidating to begin. A little web-based game with nothing else but JS; we would see what we could come up with.
Fortunately, our teammate had already done some experiments, and we were able to grasp the basics. With the canvas element, everything was going to be about manipulating graphics. Being inspired by RPGs, we settled for a 2D map based game, grabbed some sprites off of opengameart.com and got to work breaking them apart:
Churning out a generic sprite loader to actually use them with the canvas element was one of the first things that we could do. There would be a single SpriteLoader object, that would know to do the ground level job of loading all image files and drawing them (or parts of them) on the canvas element. The sprites are exhaustive, for every game element that you can think of. The ones for the maps can be rendered in a variety of ways in space and time. Using them to their full potential was a distant possibility, but for the time being, we could just stick with what we needed.
The idea was this, aside from the utility objects, every game element would also be a separate object, with its own state that had things like its sprite, current position and action. The main game loop would draw a frame with all the objects repeatedly according to their current state. The Warrior object for example, with a constant position and action yeilded something like this.
And now for the exciting part: interaction. For moving him with different keys, we could define keypress handlers that changed it’s position and sprite, accordingly.
With that simple idea, it went from cool to mind blowing in like, 5 minutes.
The map, rendered over a larger area, had a 2D array to define it’s sprites. We populated the array with larger objects at random first, and filled in the the remaining space with single unit grass sprites.
With those in place, we could use them to add more of the environment elements:
Probably the best part of the project was the arhitecture. Overall, it helped to follow the ‘what an object knows and does’ thinking pattern. Also, it made it incredibly easy to strip off the boilerplate and focus on the mechanics. That, along with the hackathon spirit of decision making on the go; made us cut corners and get things working quickly.
Pretty impressive for something built from the ground up. And all that’s just the beginning. For starters, socket.io lets us get up and running with real-time multiplayer mode to get a step closer to BrowserQuest with just a few lines:
More interactions, combat, quests and realistic maps are all yet to come, and we’re no longer lost about how we can make them happen.