Part Two! A personal best!
So I've poked around the Unreal II editor (perversely called UnrealEd 3) for a week. They say that you buy your next car based on what you disliked about the previous one and so far UnrealEd 3 is shaping up. I'm going probably going to get techy every now and again so you should skip those bits if they bore you to tears. However, everything I know about beating up game engines I learned from all the far cleverer people who have figured it out before me and graciously decided to share their knowledge on the internet (helpful link list at some point in the future), so if you're skipping bits because you don't understand them but would secretly like to then stop, find a copy of the game and figure out what the hell I'm prattling on about because that's the best way to learn.
Anyway, techy stuff (kind of): The modified Unreal engine that Deus Ex was carved from was very unforgiving about map geometry. Cassandra's Map God, Col, will 'pish' and 'taw' at that, being the Grandmaster that he is, but I spent many, many nights weeping uncontrollably into my shorting-out keyboard at BSP holes that multiplied with every fresh attempt to make them go away and I've no desire to repeat the experience. UnrealEd 3 uses a technique called 'Static Meshes' which are created in 'proper' 3D modelling packages, not in the arcane and baroque map editor, and provide complex and mostly error-free map geometry as easy as creating mesh models. Which is to say, not very easily unless you know how make 3D models. Which I occasionally need to do in my day-job so winner.
Despite this nice revelation I don't intend to spend a lot of time creating new map geometry. This experiment is about the characters and all I need is somewhere for them to mope around. Somewhere basic and fast to implement.
During the four-day epiphany that was my trip to Melbourne (yeah, I go on about it a lot, it was THAT good) I was incredibly lucky to find myself in the company of some distinguished members of the games-making industry. There was the Artist, the Veteran developer and the Designer of Modern Distinction sat around a table in a bar. The conversation was like some alchemy of the Elements of Games Making, each arising topic of conversation bombarded with ideas from four entirely different and diverse directions. Of everything that was discussed (at which I screwed my eyes up tight to try and absorb every word) it was something the Artist said that stuck with me the most. As we bemoaned the incredible difficulty of bringing good ideas into reality he wondered that, with all the talk of the craft of games making, why game developers big and small always set their sights so high? He said that, from his perspective, art didn't have to be a towering edifice to be worthwhile and why don't games makers limit their expression, making smaller projects with a greater chance of completion but with the same sense of adventure and expression? I've been holding on to that idea for six months.
Small then, geographically confined. The Unreal II engine will do that easily and quickly, indeed I suspect that was the premise behind the Atlantis scenes.
The problem with trying to make a geographically confined (thinking maybe two rooms, that small) first-person shooter is that it goes entirely against the grain of what an FPS is supposed to be. Good. We like that.
As far as I can tell, with my myopic mind's eye, the reason FPSs go on and on down endless corridors as far as they do is a question of pace and progress. In the beginning, when map editing programs were young and precocious children, first-person shooters told their narratives via the medium of architecture. Because it was far too complicated/processor intensive/whatever to offer much in the way of interaction with the environment they settled on two main methods: 'shooting things' and 'moving somewhere else to shoot things'. It's sad, that as the technology has moved on the thinking really hasn't. Most FPSs these days are still based around leading you through architecture that has taken a team of mappers months and months to create. It will take you ten minutes to shoot all bad guys in an area that represents weeks of full-time work. A bad investment, I think. For one guy trying to do it all himself it's a recipe for waiting a long time to see an end result. Or worse.
The question is, without the headlong rush through masses of geometry, what's going to push our story along? How are we going to know when we've finished if there's no eventual destination? No finishing line?
One answer to that question might lie in something I've been thinking about since Cassandra and it might be a games design device that goes back to the videogaming primordial ooze: the timed level.
Generally speaking, I've always found timed levels to be an enormous pain the the bum. I hate them, despise them, loathe them with intense, burning abhorrence. Watching the little clock count down to inevitable failure and equally inevitable restart and the same just-at-the-last-second abject failure. But that's the old abstract concept of the timed level, do this thing before the clock runs out.
Sitting on a bench in the park for a while and a number of things happen. A man walks by talking on his phone, a while later a woman walks past in the opposite direction and she's been crying. The wind picks up blowing rubbish along the pathway, a car parks on the double-yellows and a little while later a traffic warden comes by and gives it a ticket. My lift turns up and I get in and leave.
This is a linear sequence of events with a conclusion. In gaming terms it's just as full of opportunites for interaction as walking up through town instead, the difference being that in one instance I have to model a park bench and in the other I have to model the whole town. These events that occured and formed an ad hoc narrative of sorts could just as well have been running on an invisible timer, dispatching pawns at preappointed times until my lift arrived at the end of the sequence and carried me away to another game. Kinda reminds me of the way some of the Hitman levels work.
Unreal II will do that too, with Triggers and Dispatchers and who knows what else that's hiding in the class list.
So put these few concepts together and what can we come up with? What'll form the soft, jellified bones of this project that we can firm up and start planning for (Make a Plan! Stick to the Plan!)?
There's no way I'm getting into new character models (unless it's some light reskinning) so we'll have to make do with what comes with Unreal II, which is disappointingly unvaried. Deus Ex has a marvellously generic and flexible system that allows allow you to put together pretty much any kind of humanoid character you care to dream up pretty easily (as long as you don't want them to follow orders that are too specific), but Unreal II has a rigid library of precisely what appears in the game and at a much higher level detail. The simple answer to this is to base our project on the background of the game itself, dragging our project further away from the Total Conversion end of the scale and more towards to the user-created mission.
Out of necessity I'm going to set us in the Unreal universe and strangely thank the loose nature of it's backstory for giving us the freedom to get creative with it. I'm pretty uninterested by the alien lifeforms (for the time being) which leaves the human models. Luckily there are quite a few variations on that theme, unluckily they are all military variations, two flavours of futuristic mercenaries, your Space Marine pals (no relation to the Games Workshop brand, honest) and a couple of non-combatant techs. Looks like we're writing a war story.
That's no bad thing really, there are plenty of futuristic war stories in the wider media that are worthy of being entertained by and plenty of conventional war stories we could be inspired by. The Marines then, we've seen them work in the game so there should be plenty of example scripts to learn from.
Some Space Marines hanging around somewhere small waiting for an event to happen, maybe even waiting for a lift, although probably not in a park. One of the advantages of using the timed approach is that we know EXACTLY how long it's going to take to play our little experiment. One of the very early Cassandra objectives was to shorten play time to about the length of a TV episode. Although that objective was set long before I joined the team and had probably been lost by that time, it was one of the ideas that I thought was most interesting. Let's sketch out a rough thumbnail:
There's this small group of Marines that have been cut off from their main unit, or maybe they've been on a mission. They're holed up in a foxhole or bunker waiting for a dropship to come in and pick them up and all they have to do is survive until it arrives in about an hour.
This gives us a handful of NPCs to play with, contained to a small building. A series of timed events can occur culminating in the arrival of the dropship and the end of the level. Playing the drama out over an hour gives us the opportunity to change the pacing as we see fit, from lulls filled with tasks like setting up defenses and sitting around getting to know our colleagues to full-on assaults of the fortified position by the bad guys. One of the things that occurs to me is that the bots fight each other very well and quite autonomously and there will probably no actual need to script the dramatic death of your best friends. Given enough time to engage them in conversation during the quiet periods and maybe the bot's demise through the natural course of play will be enough. That's assuming you can't intervene to save their necks, of course. Hmm, maybe we can seamlessly mix in disguised linear narrative with narrative emerging from play all without leaving the confines of the bunker. Time to think about writing that plan.
Comments: on the _blackbored