The Serious Art of Designing Fun

Torque Game Builder is definitely the way to go if you're building 2D games. I was able to produce a platformer within 2.5 days of effort (coding and art integration). By a platformer, I mean basic game flow and cameras, player control mechanics, and collection and scoring system. My objective was to produce a realtime multiplayer version of the game but I ran into a major issues which slowed my progress (more on this later).

I've included a video of the prototype for the more visually inclined. A big shout out goes to Aaron White and Roger Lee for providing the art. If you're wondering about the guinea pig theme, it's a product of Aaron's crazy imagination. You can find out more about him at his website designerwhite.ca.

Now, on with the actual evaluation of Torque Game Builder 1.7.4.

Follow up:

The good

The engine comes with (almost) all the tools you could possibly need for building a 2D game. Features include the GUI builder, tile map editing, particle creation/editing, scrolling texture support, built-in physics, collision shape editing, sprite animation editing, spline path editing, triggers, behaviour modules, and built-in online support.

These tools provide a lot of flexibility in how the game is crafted. In my case, I chose to use a static background which I then layered with (invisible) collision volumes to add collision support. For the collectibles, the sites were designated with triggers which would spawn an animated sprite whenever they were active. Upon collection, it destroys the animated sprite and spawns a visual effect.

I was especially pleased with their behaviour module support - a smart of way of increasing code re-usability. The behaviour module approach allows programmers to decompose an object's behaviour/functionality into separate independent module packages and to slap on a user interface for tweaking each module individually or on a per object basis. This makes it remarkably easy to share code while allowing developers to customize the functionality within a controlled set of parameters. Suffice to say, part of my progress was due in part to the ability to behaviour modules created by other developers. Big kudos to the GarageGames guys for thinking up this feature!

All in all, I was pleasantly surprised at the depth and breadth of the toolset. The main engine was surprisingly stable. Unfortunately, the same does not apply to the GUI builder. Which brings me to...

The bad

There was significant instability in the GUI builder. It crashed frequently and without warning (yes - I checked the log files) and slowed development. The same applied to the collision detection and response system - one of the most difficult things to get right. All collision response systems inherently have some limitations/quirks and it's up to the developers to work within the limits. In TGB, there are significant collision response issues surrounding convex hull collision resolutions and collision volume scale. What's worse is that the documentation doesn't cover the limitations of the system - I'm referring not only to the documentation that comes with the installation but also as documented at the TDN (Torque Developer Network) webpages. It was nigh impossible to figure out the source of the instability with the existing documentation and a lack of access to the source code. After trying to resolve the issues for more than half a day, I hacked in a collision detection and resolution system in script code to circumvent the issue.

The ugly

There are two major issues with TGB - the documentation and the build game option.

The documentation included with the installation is relatively sparse (as I've mentioned previousy). Worse, some of it is misleading (the resource pack tutorial is a clear cut example). The TDN documentation is a little better but it suffers from the same issue. This is probably due to the fact that posted articles do not have to indicate which version it's compatible with. GarageGames should really consider revamping their tutorials and pruning the TDN website.

The more serious issue is the 'build game' functionality - the ability to take a TGB project and to publish a standalone game. The only problem is that the publishing results was untrustworthy! Towards the end of my evaluation, the published game "dropped" out functionality. In my case, the published game failed to pick up the initialization settings assigned to objects. This led to missing collision volumes and non-functioning pickup objects. Fortunately, I was able to circumvent the issue by copying the game data folder from the Torque project into the publish game data folder.

Conclusion

No engine is perfect. In particular, no engine is perfect for all games. Torque Game Builder comes close with respect to the 2D game genre - and that's without the source code! I intend to purchase the engine despite the issues because, ultimately, the amount of effort it saves me far exceeds the USD$250 price tag.

For the record, it took closer to 2.5 days in order to get comfortable with the game engine itself.