11-15-04 - 2


"Oddworld : Stranger's Wrath" stories - we have great big seamless levels. The state of the whole world is persistent, so if you drop a coin somewhere in the game, then play for hours and go back to it, the coin is still there. In the shipping game, the levels are pretty linear. We imagined them as being more like an RPG, with exploration, and big wild spaces, spawning persistent enemies you could kill for money and resources. We have towns, but we imagined big towns like hubs for the levels. You could stock up in town, talk to people, then go out on missions, go back to town and stock up, finish the mission, etc. We still have some of that, but not much.

We have various different character types and races. In the shipping game, they basically never mix. We originally planned on mixing them, and wrote most of the code for that. Part of the original vision was a sort of play of the races and their interaction - like, you have the oppressed races and their oppressors, you have rednecks and city folk, and they sort of interact and mix and talk to each other and you can interact with them and manipulate their reactions, etc. Most of the races are in the shipping game, and they have their character and whatnot, but they don't interact.

With a lot of features we had a sort of chicken/egg problem. We'd sort of do a feature, like supporting moving collision and animated objects. I always imagined levels where there were giant gears and pistons and you could run around on them - avoid getting smashed yourself, and knock your enemies into getting smashed. Nobody ever implemented anything like this and there was no call from design to improve the "moving collision" code, so we didn't work on it. The result was the code is rather sketchy and unpolished, because it just wasn't a priority. The design guys tend to try to use what works and exists, so they never used moving collision much. This is why coherent direction and cross-department vision is so important, you need someone in design who knows what's useful and what's not, and you need someone in code who knows what can be done and what can't, and they need to get together and agree; then the game needs to be designed based on the features that will exist, not the features that do exist.

Everything takes us 5X longer than most game devs, because every feature has to be extremely polished. A game like GTA:SA could never be done here - their controls suck, it glitches like crazy, but they have tons of features. We would take one of their features and spend years tweaking and polishing it. Obviously GTA:SA is very successful, but I find it intolerable to play. I think the optimal game dev model is somewhere in between, but much closer to GTA. Basically polish doesn't matter much except on the player controls. The game should be responsive and solid so far as controls are concerned, but if it looks like crap, glitches, pops, the animations are terrible, no one cares (well, the average consumer doesn't care), and it's a waste of dev time. In a way, that makes me sad, because I have a very high quality standard and like to work on solid product, but it just doesn't pay in the marketplace.

No comments:

old rants