The Design of Descent Levels

[sdfgeoff]

Work on DEEP Space is reaching an interesting stage, where coding is slowly giving way to modelling and level-creation.
As part of this I'm having to spend some time making levels. I have issues planning and developing these levels, so I thought I'd ask some of the experts over here.
So, a few questions:
1) How do you come up with an initial concept?
2) How do you decide placement of enemies, and key objectives (reactor, exit, hostages, keys etc)
3) How long does it take to develop a single level (roughly)

For those who only play the descents I also have some questions:
1) Which do you prefer, equipment and boxes everywhere (like D3) or stark caves/buildings (like D1 and D2)
2) Do you prefer maze type levels, or simple "blast in, kill as much as possible, escape" type?
3) Is there a place for puzzles or challenges in a level (like the pillars thing in D3) (as part of this, are objects like "teleporters" a good game-play item)
4) What makes some levels more-replayable than others?

In short, what makes a good 6DOF level?

[Isaac]

Lots of Counter-Strike, Team Fortress, SWAT, and other multi player games have developers design all of their levels with center area that's connected by two loops, meaning there's no dead end on the primary path. In other words, a real simple version of a level would be in the shape of a figure eight. A more complex version would be a series of figure eights that inter connect.




More figure eights for a more complex level:

[Sirius]

It's a complex question, and most rules can be bent.

Concept is possibly the tricky one. Back in the early days it was just "well, I need an ice level for this spot in the level set" or "I'd like to make a tunnel ratter/dogfighter/level that looks kinda like level 3 of Pandemonium except...". Which is perfectly valid, the level may or may not be memorable though.
More recently I usually have an idea for roughly what the level is supposed to be, look like, resemble... I do actually sometimes just come up with this stuff (Year of the Tiger) but often draw inspiration from other things, sometimes multiple sources. Enemy Within's "The Sunspire" was pretty much a Descent-ized spin on the Unreal level of the same name - initially I'd set it in an ice theme and planned to call it something like "The Frozen Citadel" but after the texture scheme change I couldn't think of a better name for it, so that's what I ended up with.
And Pinhead was just "LET'S PUT A LEVEL INSIDE A BALL!" - I have no idea how I came up with that though... unfortunately after a couple years some of that slips away.
Anthology was intended to echo the levels of D1 and D2 (which is why I named it that)... and later on I added D3 to the mix. This was basically a whole lot of concepts all mixed together - there were certain iconic (to my mind at least) elements of levels from the built-in missions from those games, so I basically did something "similar but different". I set the layout similarly. The D1 stage of Anthology I drew the layout for on paper, since D1 levels tended to be a bit more flat, and more like "rooms connected by tunnels". Thus I did a similar thing but maybe overdid the complexity a bit. The D2 stage I didn't, because the general differences between D1 and D2 layouts that I saw were that D2 was more vertical and more "3D" in general, a bit less tunnelly, and in some cases had a definite structure. So I tried to think of a "super-structure" to fit those terms, and still somewhat interesting and unique - and settled on a three-pronged "fork" in the rough shape of a sphere, where segmented (Turnabout Bore-style) arms came from a hub at the bottom, curved around vertically and joined up again near the top; and the inside of the structure could then be filled with a few layers of rooms. Very 3D, very complex, and it turned out, very effective at getting people lost in there...

Key objectives I basically build into the map. Even from the concept stage of Anthology I knew where the keys were going to be; the hostages were pre-set by where their rooms were going to be. Reactor and exit similarly. I've pretty much always decided the objectives' positions early on, as opposed to just throwing them in somewhere that looks good.
Placement of enemies is a tough one, because it's the second factor (apart from the robots themselves) that sets how tough the level will be. I kind of just go by feel for how many enemies a room should have, but there are certain ways to place them within the room to get the most use out of them. You can hide them behind secret doors for traps, and also put them behind blind corners for ... well, also traps. A robot that is average at best with space to see it (say, medium hulk) may be brutally deadly when you run into it at point-blank and get surprised (e.g. the blind corner thing). However, doing that increases the difficulty of the level so I find it pays not to try to be as nasty as you can. One thing I did when balancing levels was to play through them on a high skill level and see what rooms felt too tough or (rarely) too easy. If there was something I couldn't get through, it would probably suck for everyone else as well so I'd tone it down. I had to take special care with robots like drillers though, since they aren't even a threat if you know where they are. If you have to, you could always fly "dumb" - beat every room by running right in and shooting stuff, and just dealing with incoming fire by dodging it.

How long it takes is basically a function of size and complexity. I don't expect levels for a new game to get finished in the four hours I'd sometimes build D1/2 anarchy levels in. On the other hand, my newer ones (especially D2X-XL) haven't been that quick either; I think most of my single-player levels for missions like Apocalyptic Factor and Enemy Within took a few weeks, but Year of the Tiger was going on for ... roughly a year, with most of the work concentrated between November and January/February of the following year. So, at least a few months, largely due to repeated revisions though. Anthology took over two years, but it's about the size of half a dozen normal levels put together, so take that with a grain of salt... and I also didn't work on it continuously.

On more of a "what makes a good level" perspective: I think if your engine allows for it, it's much better if the levels look like something rather than just being the abstract mazes of D1/D2 levels these days. In D1's day detail came at a huge premium, and making something seem to have a purpose required a lot of creativity within those limitations. Puzzles are fine, but good ones are better. If it's excessively arbitrary, it's possibly better just to not have it. I wouldn't say no to teleporters if the level concept called for them, or in general actually; there are probably ways to make them uncool (teleporter mazes between rooms that look much the same as each other come to mind) but you practically have to be trying.

Replayability is in my mind more of a multiplayer construct, and that basically requires that the level is a fun one to play in. There's a lot that goes into that - weapon balance, level flow and pace, appearance to some degree (not being able to see other ships clearly doesn't usually wash over well) - but it's hard to guarantee popularity. A lot of times, people will just do what they want to do.

[Enzo-03]

1) How do you come up with an initial concept?
For me, I seem to be able to plan my levels best when I play many popular levels, try to see what makes them so good and/or (re)playable, and then imagine a basic layout that incorporates these things. Basically, I just look at other levels and do the rest in my head, which probably isn't very helpful. I seem to subconsciously try to do the thing that Issac pointed out in most levels.

Sometimes I entertain myself with the questions "How would this work out in the real world? What would it be? What would it do?" etc. I don't know exactly why, but this seems to help me out most of the time. But when I speedmap (i.e. half an hour to an hour to make a complete level), I don't do this much.

2) How do you decide placement of enemies, and key objectives (reactor, exit, hostages, keys etc)
Whenever possible, I try to actually build the level with such functions in mind. I usually will build areas where I intend to put robots, and then put robots in them later on. On the other hand, I will also throw in surprises, so you're not gonna have to deal with relentless closet monsters. The reactor room is always a planned area, and usually very far from the starting position so as to allow the player to have to fight through the level to it. Again, mixing things up is cool, too. One level concept I have is simply blowing up the reactor at the start and then finding the keys afterward. Another is to blow up a boss robot to obtain a key and then have to go through untraversed hell to get to the exit.

So this kind of stuff comes right after the initial concept, where I decide how the level would best accommodate these things, and then make sure the level does so.

3) How long does it take to develop a single level (roughly)
Well I tried speedmapping a few times and managed a playable level in a single hour. I have a short attention span though, so when I try to make a quality level, it can take a couple of days. What helps me is to listen to atmospheric music, such as Future Sound of London . When I do this, I can make a good multiplayer level in two to three hours, completely textured and full of all weapon pickups. A single player level can take a couple of days even when I do try things like listening to music while levelbuilding. To this day though I have only managed two single player levels, but one of them does not count and the other is going to be part of a pack.

Since I do not do this for a living, I sometimes find massive breaks between the times I develop levels, following wherever my interests meander to. This includes making Atomic Moth Invasion levels in Jane's Fighters Anthology


I do not only play the Descent games, but I will answer these anyway:
1) Which do you prefer, equipment and boxes everywhere (like D3) or stark caves/buildings (like D1 and D2)
I don't know what kind of game this is really supposed to be. In single player levels, clutter is fine. But in multiplayer levels I don't want that stuff getting in my way. I learned this the hard way by playing some of my more detailed levels (such as Nocturne). It wasn't (as) fun because the pretty parts of the level just got in the way. So my newer levels (Boxwerk, upcoming level called Seven) do away with this kind of nonsense, for the most part.

However, in these days you'd better not just make all your levels a bunch of boxes and expect people to like it. It's fine in Descent, but not in newer games.

2) Do you prefer maze type levels, or simple "blast in, kill as much as possible, escape" type?
A mix of both is great in single player. You should have much less elements of maze-levels in multiplayer though. Corridors are great in both, but multiplayer levels should be straightforward in the "How Can I Get To Killing As Many Players As Possible As Soon As Possible" department. I do not have as much experience with single player levels but you can have a lot more maze elements to these. That's what makes a lot of Descent single player so great. \o/

3) Is there a place for puzzles or challenges in a level (like the pillars thing in D3) (as part of this, are objects like "teleporters" a good game-play item)
Lots of old school video games have such puzzles. I didn't care for the pillars thing though. I also dislike when puzzles seem to be.. out of place, redundant, needless, and excessively tedious and/or difficult. For example, there's a level in Marathon which requires you to set the height of a bunch of platforms *JUST RIGHT* so that you can get to the next part of the level, and even the most seasoned Vidmaster hated it. So in the free release version of the game, the platforms automatically stop to the right height. Puzzles shouldn't break the flow of gameplay by any means.
I am not against the use of teleporters. Halo has teleporters and they're used all the time.

4) What makes some levels more-replayable than others?
Being memorable and/or fun. That's it. Replayable multiplayer levels are generally the ones which flow well and have something for every or nearly every playing style. This is why Blood Gulch in Halo is so popular: no matter how you like to play the game, you can do so in your own way and not only do great things for your team, but also have fun doing it. It accommodates everyone. At the same time it is straightforward enough that it does not take a genius to figure out how the level works. As for memorable single player levels.. they're just the ones that are fun.

[Descer]

As designer:

1. If I don't make a mission, there isn't one actually.
2. Imagination and feel. I see parts o' levels in my mind durin' and between the makin'.
3. From 3 weeks to 2 months.

As player:

1. Caves (and tunnels).
2. Kill, definitely. I, however, accept mazes, if they don't make the level too long (20+ min) or hard to navigate even with the map.
3. I hate those. Especially 'where does it open' ones.

[sdfgeoff]

@Everyone
Well that's provided me with a good half-hours reading right there. Thanks guys.
Let me summarise what you're saying:
1) You can plan things, and it is probably a good idea, but sometimes it's OK just to build from scratch.
2) Put enemy placements in logical areas, but don't make it monotonous. Probably a good idea to plan these
3) It takes however long you put into it.

1) Both
2) Both
3) Either/both
4) Balance in multiplayer, and not so important in single.
(And it seems memorability = re-playability, but I'd assumed that, so maybe I should have asked "What makes it memorable?" and don't even think of saying being re-playability...)

@Isaac
I like that 8 thing there. I'd heard it mentioned before, but the images really clarified it up

@Enzo-03
I agree about the ambient music thing, at this stage I'm going on some of the more active mixes from tekiu.

[Gooberman]

The levels/level designers I always thought were brililant no one played.

The D3 levels that always got the most play, it seemed to me, were quite simple. No chance of "getting lost" after 10 minutes of play.