Project in Review, Part 3: 2D or not 2D?

[cross-posted from my blog here!]

This is a continuation of a multi-part blog series focusing on the development of Archimedean Assault. Go read part 1 and part 2. It’s good!

Once the scope of our game was reasonably well defined, we set forth like good little game jammers and started hammering out code, 2D art assets and music.

Not far into this process, we realized the multiplicative nature our scope had on our art assets. A robot. Alright, he has to walk around and shoot. But, also, he has to transform! A transforming robot. Okay, that doubles the number of art assets for our robot’s actions. We need assets for both versions of the robot. And, we also need a transition animation going from one form to another. Okay, fine. A free-roaming game world in which the player can move 8 directions.

So, we need at least number of actions * number of forms * number of directions (x * 32 for those keeping score at home!) frames for any given action, and that’s before we get into animation. That means, if we want the robot to fire a gun and have muzzle flash, we need to get the muzzle flash from each direction in each form. When we start talking about animation and number of frames per animation, hand-drawn art assets quickly balloon out of control.

That’s before we start talking about the environment assets in which the game will actually take place.

Being the newbs that we are, we don’t have a dedicated pixel artist. We’ve got nguyenn1, who is by my estimation, about 10% of a dedicated artist. It takes us a few weekends to discover that we wrote a check for art assets that our group couldn’t cash. At least, not within a reasonable timeframe, and not within a reasonable quality.

We mitigated our dependency on pixel art by borrowing heavily from anywhere we could. danc’s sprite sheets on Lost Garden quickly became an invaluable asset that we leaned on.

Of course, this restricts us to assets that fit into the kind of game we’re trying to build, and art decisions became something to the effect of “Well, what’ve we got that kind of works?” That strategy doesn’t work when you have very specific demands, like “floating citadel” or “a robot that transforms from 8 different angles.”

After realizing that there was no way we’d effectively deliver on our artistic demands, I started looking at alternatives. 3D models are readily accessible at comparably low costs from sites like TurboSquid. After several months, the art assets for the player weren’t yet complete, and environmental art assets were still quirky and buggy. But, we could buy a 3D model that matched up pretty well with our vision for $5.

If $5 can save months of artist time, it certainly seems like money well spent. However, working with 3D assets requires time elsewhere. It requires non-trivial developer time and ramp-up time on tools and techniques across the team. Once I have a $5 model for the player, then what? The landscape of 3D modeling software isn’t terribly friendly (importers for fixed content, exporters to proprietary formats that either didn’t support our content pipeline or munged content beyond recognition, rigging models and animating them in-game…), and animation seems to be a mess across the board.

So, $5 doesn’t seem to save us time, but does distribute the load across the team a bit more evenly. But, how can we be sure that we aren’t wasting our time chasing something because we think it’s fun?

Advertisements