LD#15: Mineral Miner Post-mortem

Ludum Dare #15 is over, and I already wrote that the results are in. Aside from placing well in Community, which shows how much I love participating in LD48, I also saw my overall ranking come in at around the 40 percentile. I was ranked #63, which sounds good, but there were a number of ties for previous rankings. Out of 144 entries, Mineral Miner was 87th. It’s much better than coming in almost dead last in the previous Ludum Dare (and not completely last only by virtue of two other entries not getting rated at all), but I’ve done better.

Let’s look back on this project and see what happened. First, let’s go over a summary of the game. Mineral Miner turned out to be a puzzle game in which you drive around a cavern in a tank, getting out to collect minerals. You can only collect one mineral at a time, so you need to drop off collected minerals in your tank to collect more. If you are near a monster lair and not inside the safety of your tank, a monster will come out and chase you. If a monster catches you, you lose. If you collect all the minerals, you can leave the level and win.

What Went Right:

• Rapid Prototyping on Paper I took a free, online game design course at GameDesignConcepts.wordpress.com, and I was able to put use those skills to great effect. During the competition, I posted about my prototypes. With only 48 hours, it can feel painfully slow, but I iterated through the design, adding a new mechanic, trying it out, and deciding whether to keep or remove it, and then repeating until I had something I thought might be fun. Painfully slow? It took me almost no time at all. In previous LD48s, I’ve been known to add mechanics at the last minute in an attempt to make a game out of the code I was writing. This time, I knew exactly what mechanics I needed, and there were no real surprises here. The finished game ended up playing exactly how I hoped it would.
• Quick ‘n Dirty Graphics I’m not terribly familiar with art tools, such as The Gimp, and so every LD48 I find myself looking up how to use it to create what ends up being ugly art for my games anyway. I decided that this time, I wouldn’t try to make anything fancy. If I have any images that need text, I will use the basic text tool instead of the script-fu that creates cool looking logos if you tweak parameters just right. The tank? A square with a dot to let you know which way is the front. The driver? A yellow circle. Hey, it worked for Pac-man. I was able to focus more of my time on making the game because I wasn’t frustrating myself with trying to create halfway decent artwork.

  

  

• I Made Sound Effects This is my fifth Ludum Dare, and only the second time that a game I made had sound effects. Because I had a game that pretty much worked the way I expected it to, I had time for some polish. I made a list of sound effects I thought I would need, used sfxr to create the beeps and boops, and wrote the code to tie it all together. Adding sound really makes a big difference to a game, and I was glad that I could do so for this one.
• Faster Build Times and Lighter Distributables Because I had been doing some work on my Vampire Game, work that involves using TDD from the first line of code, I also did some work on my build scripts. Going from a 10 minute build time with a distributable that is already 10+MB due to including source libraries to a build that finishes in seconds and is less than 2MB is amazing for productivity, especially as it comes down to the final hour of the competition. Everything happened so much faster, keeping me focused on game development instead of getting distracted as I waited for a build to finish. Now, it isn’t as if my builds always took 10 minutes, but going from checked out source code to a complete build would. Once the libraries were built with my old system, compiling and linking would still take some time, much longer than the time it took with my new build scripts. Plus, one of the complaints I would get from previous competitions is that my game package was so large, so that’s one complaint I did not see this time around. B-)
• Simple AI Goes a Long Way I remember taking a few minutes to think about how I wanted the monsters to interact with the level. Should they obey the walls and other obstacles, like the player has to? If so, that would take a bit of AI programming. I don’t have much experience with AI, and I didn’t want to take the time to learn it for this LD48, so what did I do? I made the monster head towards the player every step, ignoring the environment. I could explain it away. It’s a monster. Maybe it climbs walls like crazy? The big surprise was how well it looked. Besides making it move towards the player, I also made the monster randomly move horizontally or vertically to do so. Combined with the sound effect when it comes out of its lair, the twitchy looking monster moving really fast at the player actually feels scary.

What Went Wrong:

• Distractions I have two cats, and both of them have been featured in previous LD48s, so I won’t focus on their antics too much. My home office wasn’t in a usable state, so I was out in the kitchen or living room. The cats love distracting me from productivity, and LD#15 was no exception. The one thing I did my hardest to control was external obligations. Anytime someone wanted to make plans with me for the weekend of LD48, I would politely tell them that I was busy. And it worked! I was able to focus almost entirely on eating, sleeping, and LD48ing…except for the Chicago Fire game.

  

  I won tickets to the Fire vs D.C. United game, which happened to be the same weekend. They were really good tickets, too, and so I made an exception. In practical terms, I lost a good chunk of Saturday. I was able to get the game finished, but having an extra hour or two would have been good for tightening up the graphics and audio. On top of knowing that, the Fire lost, so it wasn’t even as fun a game to watch from the 2nd row as it could have been.

• The Sound Effects Were Very Rough By far the biggest complaint from people playing my game is that the audio hurts. I was able to get audio in within the last hour of the competition, but I didn’t have time to adjust it. I knew that some of the sound effects were loud and obnoxious, especially the one that plays when you bump into walls, but I couldn’t dedicate the time to tweaking it. The deadline was looming, and I still had a few more programming tasks to complete.
• There’s Only One Level Right before the end, I realized that I did not have a victory condition. I had programmed a way to lose if a monster caught you and also if you tried to leave the level without collecting all of the minerals, but someone will eventually collect all of them. What then? Ideally, I would have added code to load the next level, created that level, and kept going. In fact, Level 2 is in the distributed game, although it is a copy of Level 1 and there is no code that knows about it. I was thinking about taking Level 1 and breaking it up into at least three levels, with each level introducing new puzzles and getting progressively more difficult. Three doesn’t sound much better than one, but it would have made a big difference. The player would have felt that progress was being made, and the later levels could introduce the trickier ways to deal with monster lairs.
• Level Loading Bug I could not figure it out in time for the deadline, and I still haven’t looked at it since, but every so often, the level loading code would choke on the data, bringing the game to a halt. Sometimes shutting down the game and rerunning it would work. The data came from a text file, and my code is supposed to load a single character at a time, mapping the value to a tile. Sometimes, however, it would choke because a single character variable would have a value that is two characters long. For a time, I was dedicated to fixing it, but with only 48 hours, a good chunk of which I couldn’t make use of, I decided that since it wasn’t a show-stopper, I would keep going. I really wish I could have figured out why that bug was there. Besides ruining the perceived integrity of the game, I know at least one person didn’t review it due to this crash.
• Making a Puzzle Game I didn’t set out to make a puzzle game. I didn’t want to worry about creating a lot of content. One level might not be such a problem if the level was varied and fresh each time you played. I could have created a procedural level generator, but I never built one before. I didn’t want to spend time learning how to do so, nor did I want to spend the time tweaking the algorithm to make nicer levels even if I did end up accomplishing it. Out of all of the ideas I came up with, the game I liked the most ended up being a puzzle game, which unfortunately meant I was either going to spend a lot of time making clever levels or finish a game with hardly any levels. It ended up being the latter.

What I Learned:

• Rapid Paper Prototypes Work My game design skills are sorely lacking, but I’ve been able to practice what I learned in the game design concepts course, and it really paid off with Mineral Miner. I’m not claiming that it’s a fantastic game, but it did rank #45 in the Fun category, putting it in the top 50%, and #27 in the Innovation category, which puts it in the top quarter! It feels good to know that the game design I prototyped early on before writing a single line of code came together, and the comments for my entry showed that people saw a lot of potential in my game. Everything I wanted to put into the game, I learned from minutes of drawing on paper and messing around with tokens. I didn’t need to have a game engine coded up to explore, discard, and introduce mechanics, which means I saved a lot of time that would otherwise have been wasted on code that would get thrown away and changed needlessly.
• Quick ‘n Dirty Graphics and Audio Can’t Be Permanent My art and audio work was minimal and saved me a lot of time, allowing me to work quickly at getting the game play up and running. Unfortunately, my overall rating got hurt here. I was near the bottom in the Graphics category at #104 and surprisingly a little better in the Audio category at #77. I was hoping for time near the end to replace crude art and sounds with better ones, but it didn’t happen. On that note, even if it did happen, it wouldn’t be more than marginally better since I don’t have the practice and skill with my art tools. One suggestion was to use images of my prototype work, and I agree, the drawings look much nicer.
• My Pacing Still Needs Work I felt much more confident about my entry this time around, but I still found myself finishing the game at the last minute. There’s very little time for polish when the complete game forms only an hour before the deadline! It’s especially a concern since I decided to go with quick and low-quality art in order to get the game running as quickly as possible. I probably could have set mini-deadlines for myself. 48 hours sounds like an incredibly compressed period of time to make a game, and it is, but it’s still enough time to flounder. Early on, I have two whole days to worry about everything. In the last 5 hours, I’m in crunch mode. I could stand to manage my time and prioritize my tasks much better.

If I could do LD#15 over again, I would try to manage my time better. I could have had the prototype work done much earlier on, leaving me with more time to do the actual programming and arting. I might have been able to get more levels and variety in if I didn’t waste 5 or 10 or 20 minutes at a time wondering what to do. Still, even though Mineral Miner wasn’t a winner of Ludum Dare, I felt it was a success. I designed early on paper instead of designing with hard-to-change code, and I was able to focus on making the game I felt had a chance of being fun. People said they enjoyed the game and wished there were more levels. It was a complete game, meaning that aside from the level loading bug I mentioned above, everything that happens in the game happens because I designed it that way. In 48 hours, a complete game that provided some entertainment for others is a good accomplishment.