[In this in-depth postmortem, Twisted Pixel, the Austin, Tex.-based independent developer behind XBLA and PC title The Maw, explains what went right and what went wrong with the fast-paced development of the XBLA action platform game Splosion Man.]

Introduction

Splosion Man is Twisted Pixel's second game based on an internally developed original IP. Our first game, The Maw, was well received, and allowed us to grow the company and begin making a name for Twisted Pixel. Having already pushed ourselves hard with the first game, we decided to make something a little smaller while still retaining the character and personality that our studio is becoming known for.

The character and personality stuff happened, but the smaller part didn't. For reasons described below, it was important for us to complete the game in a short time frame (six months) but to still improve the quality, presentation, and gameplay over The Maw. As a collective team, we feel we achieved these things with Splosion Man, but it wasn't an easy journey and we learned some valuable lessons along the way.

What Went Right

1. Creative Direction and Level Design

One of the reasons we were able to create Splosion Man as quickly as we did is that we had a strong art and design direction from the very beginning. Using games like N+ and Sonic the Hedgehog (the Sega Genesis versions) as reference early on helped everyone on the team understand what we were going for with the visual style and the feel and length of the levels.

Concept art for new character designs can take a very long time to nail down just right, especially when you are dealing with a main character that is basically made of fire and inner explosions. But instead of being more work than our previous game, The Maw, it actually was much simpler to visualize Splosion Man and the secondary characters in the game.

Whether that was because the team had already worked closely together before, or that the bizarre nature of the character design made it easier to throw out whatever ideas we wanted, I'm not sure, but it was a very quick turnaround time for characters that convey quite a bit of emotion with almost no dialog. The same for sound and music -- it all just came together and everyone understood very early on what this game was, and how it should look and feel.

Level design was originally a huge concern for us, because no matter how funny the game was or how cool the characters were, the levels had to be just the right length and just the right challenge or it would all fall apart.

Our first great success in reaching the high bar we had set for ourselves was accomplished with the additions that our programmers made to our internal tool set (which are described in the next bullet point). These changes gave the level designers a great amount of freedom to not only design cool levels, but do so very quickly.

And quickly made they were -- with our schedule, some of Splosion Man's levels had to be designed and implemented in a day. It was an insane challenge, especially with multiplayer levels where the thought process in designing is completely different than for the single player levels.

The second great success came from the ability that our level designers displayed in not only being able to reference games from the past that would help us as a starting point, but in also being able to create entirely new ideas from a small set of puzzle types that they could use over and over again and still keep new levels feeling fresh.

2. Tools and Building upon Existing Tech

During the development of The Maw, we also built our internal engine as well as our game editor, respectively named BEARD and RAZOR. As you might guess, this took up quite a bit of programming time just to get the engine to a usable state.

At the time, we weren't able to spend a whole lot of time adding more advanced features to the engine or fixing usability issues in the game editor. However, while developing Splosion Man, we were able to spend time making the engine and game editor more complete.

When we first started developing Splosion Man, we knew how short the schedule was going to be so it was important to make things as easy as possible on the designers so that they could create levels quickly and effectively.

The first issue we addressed was the need to run the game as a separate application from the game editor. Designers would open the editor, make some changes, export data, run the game, test their changes, and then repeat the process. This wasted a lot of time during the development of The Maw because level iteration took a lot of time which was exacerbated by the fact that our level load times during development were slow.

Our solution to this problem was to modify our game editor so that the designers could actually play the game within the editor. Users could easily toggle between layout mode and play mode with the click of an icon. This decreased iteration time immensely, allowing designers to quickly modify a level and see how it played and then repeat the process without ever having to leave the editor.

The next big task we tackled was adding a 2D layout mode to our editor. The Maw was a 3D game in which the player was able to wander around a 3D world. Splosion Man is based on 2D gameplay, although it uses 3D assets.

So we decided it would be best to add a new mode to the tool to allow designers to more easily place objects in a 2D plane. We added a layering system so they could easily place objects in the foreground or background of the playing area.

We also knew that since we were so shorthanded on artists that the designers would need to build the level geometry themselves in the editor. So we created a feature that allowed the designers to "draw" in the levels. Underneath the covers, the editor would automatically build all of the collision and visual geometry needed to represent those drawings in-game.

After we had the features that we needed in order to make the game more efficiently, we were then able to spend some time on some features that would make the game look a little bit better. We added morph target support, reflections, improvements to our shadows, distortion support, and generally upgraded our shading pipeline.