This past weekend I participated in the 30th Ludum Dare 48-hour competition and created Fusebox, an energy management simulation game. What follows is a summary of my experiences creating it, and what I learned from doing so. I had a lot stacked against me, and while I missed some milestones that would have taken the game from mediocre to great, I think that I did really well considering the situation. Before we can assess that though, we have to start from the beginning.
Sure-footed as a Mountain Goat
One week before the Ludum Dare competition started, I was at the local rock gym with a friend of mine. They had more than just rock walls there, and in my first (and most likely last) attempt at this whole slack-lining thing, I fell and landed sideways on my foot. It instantly swelled up to the size of a potato, and I haven’t been able to walk properly since. I have made a pretty decent recovery so far, but one thing I can’t do is sit up. If my foot isn’t elevated above my head, it swells back up like a balloon and becomes incredibly painful. Since working on a computer with any amount of comfort necessitates putting your feet beneath the desk, I wasn’t sure how it would turn out.
This is what you get for hanging out with extroverted adrenaline junkies.
The downside to this was that I couldn’t get into a position that was ideal for either my foot or for writing code. I was at least slightly uncomfortable the entire time, and several times throughout the weekend I had to stop and move to the couch to give my foot a proper rest. This had two side effects. The first was that I lost a lot of development time to laying on the couch with my foot on the back. The second was that in order to try and take advantage of this time I brought my notepad and did as much design and planning as I could while I was away from the computer. This is probably the main reason the game is so complicated and over-engineered.
This is the first time I’ve taken hand-written notes in years. I had to use one of those weird scratchy tube things to scrape my thoughts in stone.
What is it Even Uniting?
One of the main reasons I do these competitions is to force myself to try something new. I’ve used new engines, tools, or frameworks every time, and I’ve never made a game in a genre that I’ve done before. It’s a great way to learn a lot in a very short amount of time, and when you’ve been programming for a length of time measured in decades, it’s not a stretch to try and figure something out in that time frame. Since my current game project is in Unity, and I’ve been struggling with it since day, I decided that I would force myself to figure out and use Unity for this competition. In retrospect, I’m glad that I did, but it definitely slowed me down quite a bit.
I also chose to do a very UI-intensive project this time around, for a couple of a reasons. I felt that my foot might get in the way, and I wanted something I could work on from the couch if the need arose. I also know that I hate writing interface code, mostly due to the fact that I’m not very good at designing interfaces, and I find the whole thing very tedious. I may have been setting myself up for failure here, but the goal was never to win the competition. In all of the work I’ve done on Project Dunwich, I have not even touched the interface yet. At one point I actually had to look up how to make a button. I was starting from scratch here.
So… Much… UI Code…
Despite using tools and techniques I was unfamiliar with, and dealing with Quato growing on my foot, I felt pretty good going into the competition. I had read through the list of themes, and I focused my thoughts on the highest rated candidates from the first four days of voting. This is by no means a fool-proof method of predicting the winner, and I wasn’t writing any code or committing anything to paper, just idly thinking about the design possibilities. I ran through some ideas while I went about my day, and initially I wanted to make something with more action, since my last two attempts sort of fell flat in that regard. Most of the ideas I thought of with any amount of action seemed either too obvious, or not connected enough to the theme, and UI was another focus area, so I settled for a management sim game.
Once the theme was actually announced, I was a bit relieved that the top theme won out, since it meant I already at least knew what genre of game I would make for it. The idea was simple, connect worlds together through some interface, but give those worlds multiple, intricate layers of connection. I like to make my games hit the theme in multiple ways, and that satisfied that requirement. Connecting worlds to an energy source was the obvious take on the theme, but having them be linked to each other as well added a nice extra layer of depth to the interpretation. I’m not sure any ever notices these little touches, but it makes me feel better about my interpretations.
I’m really happy with how the planet rendering turned out. It’s a shame I never got a chance to make that interface actually useful though.
I immediately started with graphics, since I didn’t think there would be very many, and I wanted to get it out of the way. I drew up a mock interface, some icons for the planet stats, and some graphics for the planets themselves, and actually had something passable by the end of the first night. I have used Inkscape quite a bit over the past few months, but I had never done clouds or noise in it, so that was a fun little challenge to overcome. In the end, I spent less than four hours total on graphics, and I’m glad I didn’t have to fight with them at all once I got into the interface code.
With graphics in hand, I set to create the game objects and renderers that would use them to actually put the images on screen. Unity actually made this really easy, though I have no idea if the setup I used is proper for an entity-component system. Since most of the game objects were just data containers, that didn’t take very long, and well before the half-way mark, all I had left was to write the code to process the interactions on the game objects, and then do a whole lot of UI work. After a brief stint on the couch to rest my foot, I started in on the UI.
It is a damn good thing I don’t have to actually draw interfaces. I drew it and I can’t even tell what’s going on in some spots.
As I mentioned earlier, I had no idea how to approach the interface. I created things using GUIText and GUITextures, I switched to converting screen coordinates to world coordinates and driving the UI with game objects, and eventually discovered the OnGUI method and settled on using that. Throughout the course of the day on Saturday I created as many interfaces as I could, to enable interaction with the game objects. I could just as easily have started by coding that all into the setup and working on the simulation, but that seemed like it would be harder to iterate on. Once I learned how to make the interfaces, it was a pretty smooth ride of create, test for usability, modify, repeat. I didn’t do things in a very efficient way, there’s a lot of copy/pasted code for UI stuff, but I just kind of zoned out and started writing.
By the end of Saturday, I had about half of the interface done, and none of the game logic. That seemed like a bad situation to be in, so I set out to right that first thing on Sunday. Since I had spent a fair amount of couch time writing out notes on how I wanted it to work, that actually went pretty fast. The logic is pretty complicated, there are a lot of moving parts that determine how the hardware will react, and how the planets will respond to their situations. The biggest problem with all of that is that I couldn’t get the interfaces done in time to actually explain all of that to the player.
My half-baked attempt at a tutorial. The best part is that I didn’t even get to implement some of the stuff I explained in here. Super useful.
The final interfaces were the ones that told you what was going to happen when you advanced the day, and the one that you manage your circuit board. You know, where you actually play the game. I knew what needed to be done, but by Sunday my foot was in open revolt against me. I spent a lot of that final day on the couch resting, and with nothing to plan, I just sat there mentally writing interface code to draw out how I wanted it to look. The funny thing about mentally writing out code, is that it’s a completely useless activity. When I felt good enough to try and implement it, everything fell apart on me. I had planned on whipping up those last two screens and then playtesting the game for bugs and balance. What ended up happening was a mad dash to get the interfaces working that ended about 15 minutes before the deadline.
At Least I Finished… Sort Of
I decided that I was done, and 15 minutes wasn’t enough time to get any of the last things I needed from where they were to where they needed to be to even be passable. I set to build my project and upload, and then Unity decided to remind me why I hate it so much. Apparently the method I was using to color the planets with HSV colors was only available when you ran the game through the editor. It wouldn’t even compile. Fortunately, HSV to RGB implementations are a easy to code, so I started throwing one together. In the past I’ve worked on them with hue being an angle from 0 to 360. Unity’s method had it as a float from 0 to 1. No sweat I thought, I’ll just multiply it by 255. If that doesn’t make sense to you, it’s because it shouldn’t. All of the planets turned green because I mixed up angles with rgb values, but I didn’t have time to figure out why. Up it went, and just like that it was all over.
I’m glad that I made the choices I did. I learned more about unity and its UI features in this past weekend than I have in the past four months. Sure, I could have scrapped the planet interface and focused more on good UI, and I probably would have been better off spending time on tutorials rather than tweaking button positions. Given the constraints I was working under, I’m really happy with how things turned out. I do have some ideas for how to improve next time though.
- Simple design with good balance – I had so much time away from the computer that my end product was as overly complicated mess, and there’s really no way for the average player to figure out what’s going on. Having a more simple design with a better balance would have been a better approach. Instead of having four types of compatibility and a two-hundred line calculation for fuse load, cut it down to two stats and spend more time on making sure the numbers work out over the course of the game.
- Interaction before eye candy – My planets look way better than they need to for a game that is mainly driven by button clicks. If I had put that off until the end, I would have been able to see that before wasting time on them, and I might have had time to implement things like a proper tutorial or a win condition.
- Playtest as early as possible – I put the core logic off for so long, that by the time I had it finished, I was already in crunch mode. This left me no time to make sure the numbers worked out, or that the game was even fun. With a game like this there’s really no excuse, I could have had unit tests written to test out the formulas and algorithms through all 100 days by Saturday morning if I had prioritized it. Good balance is going to be my main goal for next time.
That about covers it. I had a good time, and in the end I have another game to throw on the website and say “look what I can do in a weekend.” No matter how bad I do, or how stressful it is, that sense of accomplishment will always be worth it.