Secrets of the Sages: Level Design

Author: by Marc Saltzman

The worldwide PC gaming community hasn't been the same since id Software's Wolfenstein 3D was unleashed in the spring of 1992. More than a quarter million people scrambled to download this racy, 700KB shareware game from their local bulletin board system (BBS), and thus, the first-person perspective 3D "shooter" was born.

The next seven years yielded many memorable shooters--Doom, Dark Forces, Duke Nukem 3D, Quake, Jedi Knight, GoldenEye 007, Unreal, Sin, and Half-Life--and in so doing, it launched a specialized and necessary art form, known as level design. Loosely speaking, level design pertains to creating the architecture of maps in the game, plus dealing with object placement, mission or goal of the map, and often mini-missions within the level as well.

So what makes for a well-laid-out and challenging level to complete? What are some of the more common mistakes found in amateur level design? Is there a science to it? You bet. This chapter features the world's top level designers and their invaluable opinions on what makes or breaks level architecture in 3D shooters. Keep in mind that many of these pointers can also apply to other 3D games, such as third-person perspective action/adventure hybrids (e.g. Tomb Raider).

Tim Willits, id Software

As level designer on Ultimate Doom and Quake and lead level designer on the legendary Quake II and the upcoming Quake III: Arena, Tim Willits has gathered quite a bit of knowledge (and respect in the industry) on how to create a successful map for 3D shooters. He shares with us his words of wisdom on this exciting facet of 3D game development. Before Willits sits down to create a map he asks himself, is this going to be a single-player map or Deathmatch map? He expands:

Before you can do anything, you need to figure out what kind of level you want. It's a lot easier if you stick to either Deathmatch (DM) or single player (SP). Though it can be done, making a map great for both DM and SP is a very difficult task. Usually if it's great for DM it'll be too circular for SP, and if it's a fun SP map it's usually too straight for DM.

Note: Deathmatch, or DM, refers to non-team-based multiplayer maps typically played over the Internet or on a LAN with other human players. It's a "kill or be killed" scenario where the goal is to stay alive and rack up as many "frags" (points) as possible. Kill a player, get a frag; die yourself, lose one. Hence the name "Deathmatch."

While many of the same principles can be applied to creating both SP and DM maps, Willits breaks up his advice into separate groups for SP and DM:

Single-Player Levels

According to Willits, there are a number of rules to adhere to when devising successful single-player levels. The following are the most significant rules to keep in mind.

Focus and Continuity
Of the utmost importance in Willits' opinion is focus and continuity of the task. And as he explains, it can be easy to lose both during a game's cycle:

Every game has one overall mission or goal. The game then is made up of many single levels. Every level in turn must also have one overall mission. And every map must have a reason why it exists. It's important that the designer doesn't forget this--it happens a lot. A designer will be working on a level with a goal in mind. And then something happens--sometimes a technology is introduced into the game or a technical problem arises and the focus of the map shifts. Sometimes the designer doesn't even realize that they've lost focus on their original goals of the level, but they have. When this happens, the designer must step back, look at where things are going, and focus more attention on the overall design and goals of the level, sometimes reworking areas or changing the goals to accommodate the map's mission. Basically it's crucial that designers stay focused on their ultimate goals in designing a level.

As a side note, Willits reminds level designers that there must be one person who focuses on the entire design process, to ensure that levels don't stray too far from their original goal. A fresh set of eyes from someone no too close to the level designing is ideal for all games.

Architecture Design and Gameplay Elements
According to Willits, SP maps require a pretty linear flow, and they need to guide the player through the level with both architecture design and gameplay elements. To clarify, architecture design is basically how the areas are constructed. There should be natural breaks in levels that separate the major components of that level, as well as the level's mini-missions (Willits gets to this in a moment). Gameplay elements, on the other hand, follow the events orchestrated by the story of the level within the game. Willits cites an example:

If the player's mission on the power station level is to destroy the nuclear reactor, then the level may be broken down into areas such as the control center, waste pumping station, core reactor, and coolant subsystems. Each one of these areas must look like they're supposed to look, as well as perform some function in the overall level. The player may need to enter the security codes in the control center to grant access to the coolant subsystems. Once in the coolant subsystems, the player could drain the core reactor's coolant, causing an unstable heat exchange within the core. Finally, the player could reverse the waste in the waste pumping station, creating a chain reaction that would destroy the entire nuclear reactor.

Risks and Rewards
Risks and rewards must be peppered throughout SP maps to challenge the gamer while plowing toward the end of a game. Willits believes it's essential that each new area contain these kinds of obstacles. Here's an example:

The player enters a new area of a map and there's a slime pool that's too far to jump across. On the other side of the slime pool is a button that extends the bridge, but it's guarded by a monster. The player's mini-mission is to extend the bridge. The obstacles to accomplishing that mission are the monster and the fact that the button is on the other side of the slime pool--too far away to push. To accomplish this mini-mission, all the player needs to do is shoot the button from his side and then avoid the monster while crossing the bridge; that's it. Simple. It may not seem like a mission, but it is. It's a challenge that the player must face and overcome in order to continue with the game. A single-player level is a collection of these mini-missions tied closely around unique areas in some cohesive manner.

As an example of not rewarding the player enough, Willits recalls:

I once played a game where there was a tower in the middle of a courtyard with some monsters in it. It looked important and it was a centerpiece of that courtyard. I killed the two guys in the upper portion and navigated to the top. I was pretty disappointed when I finally reached the top and there was nothing there. Every time you have an area in the map that looks important and there's a fight to reach it, you need to reward the player with some "goodie."

Environmental Feel, Teasers, and Flow
An important consideration of an SP map, according to Willits, is the overall environmental feel. Aside from looking good and playing well, what does he mean by this?

A designer must make the level look the way the player expects it to. If the designer calls a map a warehouse, then there better be some crates lying around, because players will be looking for them. Also, a designer must try to make the level seem like it fits into the rest of the world. Don't mix time periods if you're not traveling in time, don't mix construction materials along similar time periods. For example, don't build your first map out of sheets of metal and have the follow-up map made mostly of brick and stucco. Players want consistency; they're comfortable with it because it surrounds their everyday lives.

Another nice touch in creating a good environmental feel is to build the map with the hint that there's more out there. "Create fake facades that can be viewed through windows but unreachable on foot. Have boxes come out of walls and vanish through other walls on the other side of the room. Create architecture that sweeps out past the playing area," says Willits. If these items are placed in the levels by the designers, players will feel like they're involved in something "bigger."

Along with these "teasers," a few outstanding visual scenes or landmarks will also help capture the environment that the designer wants to create, says Willits.

Spend some time developing a spectacular view. Maybe a grand entrance, a detailed outer building, or even a super advanced control center. Make players turn a corner for the first time and say to themselves, "Wow." It stays with the players, and they remember the level long after they completed it if they were impressed by something cool-looking. This isn't so important in DM maps, mainly because once you run past it no one cares what it looks like time after time. Don't spend too much time on something visually stunning in DM; spend more time on flow.

In terms of flow, the levels need to start out pretty easy and then advance in difficulty, maintains Willits. As a rule, he builds the first level as a training level.

If you want to build some cool objects that move or some sort of complex geometry to showcase the engine, put it out of the path of the player. For example, air vents with spinning blades look just as good horizontal behind grates. Or moving pumps along side walls is another good use of moving things that are non-threatening. I know you want to add a lot of interesting things in the first couple of levels, but just keep them as non-intrusive as possible.

Perhaps you're only interested in creating top-notch multiplayer maps? If that's the case, pull up a chair to Willits' DM 101 and a few pointers on Capture the Flag--style games as well.

Deathmatch Levels

There are basically five popular styles of Deathmatch levels: arena, circular, linear, location-based, and theme-centered. Many of these styles can be included in one map, and some have crossover traits, according to Willits.

Arena
In a nutshell, arena levels usually have one central area where most of the combat takes place. Most of the hallways and passages either lead from this central area or to it. Says Willits:

The map has very few other large rooms or areas of significance. The arena style of DM is very focused, very refined; the maps are quickly learned and easy to master. Players will always know where they are and should never get lost navigating the hallways around the arena area. Players will find these maps fast paced with high frag limits, which will be reached quickly. An example level is map07 from Doom II.

And a word of caution to designers:

Try not to make the arena areas too architecturally complex. This is the area where all the fighting occurs, so it has to run fast. Complex architecture may look good, but it only slows down the game. Try to build this area as simple as possible.

Circular
As the name suggests, these maps are circular in design, or as Willits says, "built in such a fashion that the player would never need to stop and turn around along its main path." He expands:

Build with as few dead ends as possible--they're best built with none. Use numerous entrances and exits around its central core, which would allow free-flowing movement without hitches. The map would also need good weapon distribution, where either side would not have an advantage. There would be as little holding ground as possible. (Holding ground is a place where a player can stock up on health and ammo in a room and camp.) An example level here is dm6 from Quake.

Linear
Linear maps are built with only a few alternate paths. Willits amplifies:

The architecture becomes a roadmap, where people instantly know which side of the map they're on. Nice open areas or wide hallways where players can enjoy jousting-type combat. Even weapon distribution to force players to move back and forth. Have the ammo for the weapons on the opposite side of the map, forcing players who want to stock to travel. An example level is e1m1 from Doom.

Location-Based
Location-based DM allows players to always know where they are. You may not be able to figure out how to get somewhere else fast, but you immediately know your location. These maps are not free-flowing as in circular or linear maps, but instead are made up of many unique identifiable areas. Each area should have some distinct combat areas or mini themes included in it. For example, in dm3 from Quake is a water area for swimming, a thin staircase for vertical fighting, and a computer room made for close fighting. Each area has a special weapon or power-up that fits the environment. These maps are great for team games.

Theme-Centered
And last on the list for DM maps are theme-based maps. As Willits put it, a theme-based map uses something unique to combat and over-exaggerates it all over the map. Perhaps this is better explained by an example:

An example of this is e1m4 from Quake, a.k.a. The Sewage System. This map is covered with water; most of the fighting is in or around water. Everywhere the player looks, he sees water or something related to water. In almost every area, the player can enter or exit the water. The water is the "theme" or the special combat characterization throughout the map. Theme maps are great for players who enjoy something totally unique. Theme-based maps are also more difficult to navigate through, and should only be used for medium or advanced play. Themes need to enhance gameplay, not detract from it. Note: Good id Software examples of theme-based maps include wind tunnels (Quake, e3m5); low gravity (Quake, e1m8); low light, such as the mine levels in Quake 2; hazardous materials such as lava, slime, or pits of death (Quake, dm4); torturous devices such as spike shooters or security lasers (Quake 2); wide and open areas (Quake 2); teleport craziness (Quake, dm1).

Capture the Flag

"Capture the Flag" is a popular team-based multiplayer game where the object is to steal the other team's flag and bring it safely back to your own base. There are now many new custom variations of Capture the Flag (CTF) games. In the following section, Willits offers advice on creating maps for CTF fanatics.

Symmetric Levels
With CTF games, it's important that levels be nearly mirrors of each other to make things even between the two teams. Willits maintains, "In theory it's possible to have two bases look different, but even in practice this has rarely worked." He cites a bad example of a CTF map from Quake 2, and why: Strike is a fairly big failure in that regard due to BFG and teleporter placement (putting red team at a large disadvantage). Also, there are more methods of entering the red base than the blue base, making blue base easily defensible. This map also has uneven ammo and weapon placement; the blue base has far better resources within. All this is solved very easily by making both sides identical.

Asymmetrical Levels
Willits says if the level is not symmetrical there should be a balanced strategy that needs to be employed by each individual team. For example, if one side is largely covered by water, the team should be given rebreathers. Similarly, protective environment suits should be accessible on the slime side.

Random Tips for CTF Maps
Willits grants us an assorted medley of tips on creating CTF maps:

There should be a good supply of weapons and ammo near a base, but don't overdo it. This makes the base too easy to defend and difficult to attack. If a designer is using power-ups, they should never start off within the base. While still making it defensible, there should be multiple entry points and exits to a base. Centralized placement of major power-ups is a good idea. The power-ups still need to be located far enough from each other to prevent players from using a single power-up and crushing everyone on the map. Create some good sniper locations but, if players are going to snipe, they should be vulnerable in some way, too. There should be obvious color coding of areas, but don't rely on colored lighting, since colored lighting tends to neutralize player colors and you can't see what team they're on. Use colored textures instead. Focus on good weapon placement and think it through. Weapon placement may be more important in CTF than normal DM because it can greatly shift the balance of power from one side to the other.

We'll return later to the masters at id Software. But first, we've got some divine intervention. Bow your heads because next up is The Levelord.

The Levelord, Ritual Entertainment

Much of the excitement and anxiety experienced during bouts of Duke Nukem 3D, Quake: Scourge of Armagon, and Sin can be directly linked to The Levelord's miraculous touch. So what's his secret for keeping you awake until 4 a.m.?

The Levelord's three commandments may seem obvious, but according to him they're quite often neglected or overlooked: the "fun factor," game action/player interaction, and authenticity of setting.

The Levelord has requested that we keep his written material the way it was submitted, so this is the Lord speaking throughout. Hear ye! Hear ye!

The Fun Factor

The first and foremost question to be answered about any game, whether it's a shooter or whatever, is "Is it fun?" This applies to the game itself, of course, but it also has great bearing on each and every component of the game, including level designing. As a game is developed, this question should be answered many, many times. The Fun Factor is often forsaken for cutting-edge flash, and much effort is taken away from the game itself due to the ever-increasing computer performance on which games are played. Nonetheless, it seems paradoxical that many games simply are not fun; they look great, but they aren't worth playing. Cutting edge is indeed fun when it adds cool weapon effects and faster game performance, but it's not cool when a developer spends too much time with research and development, only to forget the main purpose of a game: fun! The Fun Factor is also frequently back-seated to realism. This is not to be confused with a game being realistic or authentic, but is rather a seeming side effect of the "reality" portion of our games' virtual reality. Too often fun ideas and features are shelved because developers say things like "Hey, you can't change momentum in mid-air in real life!" or "A real bullet doesn't do that!" when the more important statement at hand was "That was so much fun!" There are no defined rules for fun and the only way to ensure the Fun Factor is to play test. The easy part about adding the Fun Factor is that most all of us have the same concept of fun; that is, if you the game developer think it's fun, then the game audience is likely to think so, too. The Fun Factor is not transient or ephemeral, either. It should survive countless trials and tests and still be entertaining in the end. This is the only way to ensure that a game is fun--to play test it over and over.

Game Action and Player Interaction

First-person shooters are no longer the simple "shoot, find the key, and shoot some more" games that they were a few years ago. These games are now fully interactive environments, and what used to be considered randomly placed and sparse Easter Eggs are now the standard norm. If a level has a phone or computer, they had better be functional, and the player had better be able to blow them up. The player must be able to destroy just about everything! As a level designer, I spend a lot of my time making things destructible, but it's always worth the while. Take the time to become a good demolitions expert, because destroying things is not only good action, it is also never-ending fun. Another important aspect of the action in a first-person shooter is puzzle solving. A good level should be a series of challenges and rewards. The challenge can come before the reward, or after, but don't just haphazardly strew goodies and bad guys through your level. The players should feel as though they are being run through a gauntlet of contests and prizes.

All of these forms should be as animated as possible to improve feel of action. Make the player work for the rewards. Do whatever you can to make the player say "Ah ha!" Make as many secret areas as possible, too, as discovering secrets is one of the most fun puzzles in a game.

Authenticity

Related directly to level designing, the third question to be answered is "Are you there?" First-person shooters in particular rely on the sense of immersion. The most important duty you have as a level designer is to ring the player "into" your level. It's not until after this submersion that issues like action, game play, and even the Fun Factor enter the game. The more of your level to which the player can attach himself via familiarity, the stronger the player's sense of "being there" will be. Real-world situations usually make good levels because it's easier to capture the reality.

The closer that the player can relate to your level, the deeper the player will be submerged into the level. Continuity is also an important factor of a level's immersion.

Continuity is related to the level's main theme, and this theme must be maintained throughout the entire level. Too often I see levels that are patchworks of various themes. Your level must seem like a continuous place.

The almighty Levelord has spoken, and deems these as the true foundations of any killer level. "These seem apparent, but they can be the biggest burdens a game designer, especially a level designer, can face...without them, most every other aspect of a level will be missed or forgotten, and the game as a whole will suffer."

Paul Jaquays, id Software

Paul Jaquays is a "jack of all trades," but at id Software he provides level design for Quake II and game and level design for Quake III: Arena. Now who wouldn't want to be in his shoes, eh? When Jaquays heard about this Secrets of the Sages project, he wanted to offer a large collection of do's and don'ts on level design, as well as more general advice on the art of map creation. Without further ado, let's first jump into his collection of handy design tidbits.

Jaquays' 26 Level Design Tips

As many other programmers, artists, animators, musicians, and level designers have stated in the past, this last point on finishing a project instead of starting 10 new ones is essential, and not easy for beginners.

Paul's Advice on Game Design

Throughout his career in the gaming industry, Jaquays has accumulated quite a bit of knowledge on the art of game design. While he covered many individual pointers in the preceding section on map creation, the following details serve as more broad advice on game design, drawing from his own personal experiences as well.

Stop Imitating Yourself

I started in the game business as a designer of game adventures for the new (at the time) game called Dungeons & Dragons and later for the game called Runequest. There was a time when I was considered one of the best adventure writers in the field. One of the reasons that I quit designing pencil-and-paper-type role-playing games was I found that I had started to imitate myself, rehashing the same storyline over and over. I was no longer fresh. Thankfully, I had other career options within the game business that I could pursue. But the problem still remains: how to keep your ideas alive and new.

Choose the Unconventional Solution

There's a tendency in game design to use familiar or tried-and-true solutions to design. In the latter part of the Golden Age of Video Games (the classic 8-bit years), the solution for nearly every game based on a character or movie license was to create a side-scrolling game. During my tenure at Coleco, we were given very few opportunities to create new games. Most of our work was to analyze and translate arcade titles. The War Games movie license gave us the opportunity to create a game that broke the mold. The conventional solution would have been to make a side-scrolling "solve the puzzles, find the hidden goodies, and avoid the bad guys" game until at last you confronted the computer in the last scene. At that point, the game would start you back at the beginning and ratchet up the level of difficulty a notch. After seeing a special preview screening of the movie War Games, I was inspired by the sequence near the end of the movie in which the computer runs simulated scenario after scenario in which the outcome was always the same: nuclear war and complete world devastation. I was taken by the graphics that plotted the arcs of missiles as they approached their targets. If I could convince the powers-that-be, that short sequence of the movie would be our game. The actual gameplay derived from several unrelated concepts. The goal of the player would be to stop bombers, subs, and missiles (which were drawn on the fly as lines--no simple trick in the 8-bit pattern tile game systems of the day), from reaching their targets on a map of the United States. In a way, it had similarities to the popular arcade game Missile Command, where the player fires anti-ballistic missiles at incoming missiles dropping down from the top of the screen. Unlike the arcade, the play took place on six separate maps simultaneously. Like the juggler who keeps numerous plates spinning at once atop thin sticks, the player had to rapidly switch his attentions between a radar map showing the whole USA and six sub-maps that contained closer views of target cities and military installations. The player had to rapidly commit resources (missiles, interceptor planes, and attack subs) to deal with enemy attacks, then shift to the next map and do the same. If the player had the right stuff, he or she could defeat the game.

Blending Flavors

Some of the most popular foods are those that blend unlike or even opposite flavors together in one tasty package. Sweet-and-sour Chinese dishes and Chicago-style hotdogs are just two examples. What does this have to do with game design? One of the products of which I am most proud is a book series called Central Casting. The purpose of the products was to create vivid back-stories or histories for characters in role-playing games. I created separate books that covered three distinct genre groups of games: fantasy, science fiction (or futuristic), and 20th century games. Players rolled dice and compared the results against a series of tables and lists. Roll by roll, they selected events and personality traits that they could use to define their game characters. Quite often, the results of dice rolls would seem unlikely to be combined together, but with a little creative thought the widely disparate events would blend together, like the myriad of flavors in a Chicago-style hotdog, into a uniquely original result.

Make It Real

Even if you plan on making your game setting wildly fantastic--that is, nothing you would ever see in the real world--take care to make it seem real. This is something I learned as a fantasy illustrator, painting covers for games and books. The way to make the fantastic elements of a painting believable is to realistically paint the mundane things in the picture. This establishes a setting that appears as if it could actually exist somewhere. You then paint the fantastic elements in the painting with an equal amount of care so that they partake of the reality of the rest of the painting. The same holds true when making 3D game levels. Give the player one or more familiar elements that he can relate to. By comparison with the real elements, the unreal things in the game should seem more real, or perhaps a better explanation is that they seem more plausible. And by contrast with the mundane, they will seem that much more fantastic.

John Romero, ION Storm

"A level designer has a very responsible position, because maps are where the game takes place," says John Romero, game designer, chairman, and cofounder of ION Storm. Having worked on such 3D shooter classics such as Wolfenstein 3D, Doom, Doom II, Heretic, Hexen, and Quake, Romero has plenty of opinions to share regarding level design.

Romero maintains that the inevitable breaking up of responsibilities at a development house can take away from the overall vibe and consistency of the level. "Some companies do their maps in stages, with many different people handling different tasks--architecture, texture placement, object placement... and this [specialization] can take away from the 'cool' factor."

Although most of his suggestions on what makes a good level are fairly universal (such as providing the gamer with landmarks throughout the map to give it personality, making the player totally immersed in this fictitious world, and so on), Romero also proposes an interesting rule that's broken more than it's obeyed, in his opinion. "The gamer must be in constant fear," believes Romero. "It's extremely important to keep gamers on their toes when playing shooters, with plenty of traps lying around so they're constantly in fear of dying..."

Tip: Landmarks can serve a dual purpose in well-designed maps: They can add to the immersion factor while offering navigation cues to players, helping them know their direction and location in the level.

Romero cites specific examples such as Quake's E2M5 ("Wizard's Manse"), where players get locked into a cage and it slowly descends into water with no way out. Another Quake map, E2M6 ("Dismal Oubliette"), also has a slow vertical sink, inspired by Disney World's Haunted House "elevator" ride. "The player panics because he's stuck in this room and after it stops completely, a Shambler is teleported into the room, spikes shoot out, then there's a Fiend to fight, and then Zombies. It's just one thing after another," says Romero with a smile. "This is key in good level design."

Romero will be able to show off these helpful tips with Daikatana when it ships in mid-1999.

Cliff Bleszinski, Epic MegaGames

At a mere 24 years old, Cliff has made quite a name for himself as a game designer and level creator on the award-winning PC titles Jazz Jackrabbit and Unreal. Although he agrees with John Romero that fear is an important element of level design, he believes that it's only a fraction of what makes a solid level. "Pacing is far more important. If the player is constantly in fear, then he'll become numb; if he's constantly surprised, then it will wear off and not be effective," explains Bleszinski. He continues:

The key to scaring the player in a level is knowing when, where, and how often to spring a surprise. If there are, say, five minutes of idle time exploring and chatting with peaceful aliens, then, when the door bursts in and a lava monster stomps in screaming bloody hell, you can bet the player will be shaking in his boots. This is why good horror movies don't spring surprise after surprise on the audience, because it loses its impact. Good pacing is a skill that applies to every element of level design. Pace your flow of monsters, and have areas where the player feels like he is being engulfed by less intelligent "cannon fodder" foes, as well as areas that have just a few devious baddies that are hard as nails to take out. Know how often to reward the player with goodies or health. Don't cover the level with items; rather, give him the prizes after monsters are killed, doors are opened, or a ledge is reached.

Bleszinski recognizes that there's a lot more to level design than the "fear factor" and "good pacing," so he also provides his very own five-step crash course in good level design. Pencils in hand? Here we go...

Geometry Building and World Texturing
Naturally, you need to construct your environment first. Ideally, the texture artists should have a head start on the level designers so that the level designers have content with which to texture their areas while building. Then the level designers can tell the artists, "I need a 32´128 girder with three bolts on it and no directional rust, and please put it into the FACTORY texture set!"

Lighting the Environment
The right kind of lighting can make or break a beautiful scene. Low lights tend to illuminate monsters more dramatically, while bright rooms reduce fumbling around in the dark. A level designer needs to be creative with his lighting; if the player is going deeper into a volcano, then the lighting should get "hotter" the further you go by getting brighter and more orange and red, or if you're sending him through a swampy area, use drab, depressing colors, such as green and gray.

Using the Unreal or Quake engine gives a level designer amazing control over realistic shadows. Building architecture that allows for shadows is essential; try putting support beams beneath a skylight to encourage sharp, moody shadows on the floor, or put a flame behind a polygonal grate to cast harsh shadows on the opposite wall.

Tricks, Traps, and Puzzles

Never force the player to learn by dying. Always give him a chance to figure out a puzzle without slapping his wrists. Remember, the person playing your game is playing it for fun, not for work. If you want to have slicing blades pop up from the floor of your Incan temple, make sure that you put some blood splotches and body parts around the exact spot that the blades spring forth, so that the attentive player will not be killed. Even if the player is killed, he will think "Oh, I should have seen those warnings, how stupid of me!" instead of "This game cheats! How was I supposed to know there was a trap there?"
If the primary objective of your game is to kill and kill fast, then don't slow the player down with boring, cumbersome puzzles. It's one thing to have three switches that need to be turned on in order to pass; it's another to have twelve switches that have a combination code three levels away that the player must physically write on a piece of paper to remember.

Monster, Ammo, and Health Placement ("Gameflow")

During the course of Unreal's development, I've been harassing the level designers to always make their monsters patrol a local area, or to have them spring out of the dark, or even crash through glass at the player. If the player walks into a room and the monsters are just standing there waiting for him, he's not going to feel that this is a very believable world. However, if he walks into a room and his foe is just walking past him to go work on a computer terminal, he'll appreciate the extra effort that has been taken to further the believability. Ammunition is always tricky to get right in a level. Too much ammo and the gamer breezes through the level without a sweat; not enough ammo and your gamer is running around the level hacking at your foes with his default weapon while pondering looking for cheats online. Right when the player is thinking, "Boy, I'm going to be needing some ammunition soon," there should be a box of bullets waiting for him. The same rules apply for health.

Drama!

In Unreal, we're trying to create a sense of being in a hostile alien world. This believability is greatly helped by what I refer to as drama. Hearing a scene occur behind a locked door. Watching an evil alien punish a friendly alien who assists the player. Witnessing the murder of a comrade before your eyes. Real events that occur real time in the game, many of which the player can interfere with. Well-done drama will stick in the player's head for years to come.

So how important is level design, anyway? Why is this such an integral part of the game design? "Level design is where the rubber hits the road," are the first words Jay Wilbur, former "biz guy" and CEO of id Software, told Cliff Bleszinski, during their initial telephone conversation a couple of years ago. This has stuck with Cliff as an "absolute truth." He expands on the analogy:

Game development can be compared to building a car. You have all these different parts that are created by talented people--programming, modeling, sound, and artwork--and at some point, everyone's hard work on a car comes together, and the tires hit the road. With a game, everyone's work is held together by the levels that use all of that, and they'd better be exceptional or the game falters.

Marc Laidlaw, Valve Software

Valve's first release, Half-Life, was a major evolutionary step in the history of the first-person perspective action genre. The game excelled in all key areas: storyline, art, animation, sound, music, pacing, and level design; combined, it made players feel like they were part of a living, breathing, populated world. The level design was instrumental in pulling off this effect, and even during the last third of the game, while on another planet with alien architecture, maps maintained their focus with lush outdoor and indoor environments and challenging obstacles to overcome.

Marc Laidlaw, the level team coordinator at Valve, was responsible for overseeing the six level designers who worked day and night to create Half-Life. Did you know there were a whopping 96 BSPs in the game? Laidlaw takes some time out of his busy schedule to discuss his own personal approach to level design:

Note: BSP is an editing tool that converts map files into Quake-compatible level files. Half-Life is based on the Quake/Quake II engine, so they call the mini-levels in the game BSPs.

The Goal
The first on the list is the goal of the level. That is, the point of it all in the grand scheme of things. Says Laidlaw, "Everything in the level should somehow contribute to that goal--even detours, diversions, and distractions should figure into the overall scheme, thematically."

The Gameplay
Consideration of the gameplay is also important. What will players be doing as they go through the level? "Is this something that will engage, immerse, and interest them to the point that they absolutely cannot stop playing until they've accomplished their goal?" asks Laidlaw.

The Atmosphere
Laidlaw asks, "What kind of mood are you trying to create in the player? One of horror? One of pure action? Are you trying to create a mood that contrasts with preceding or subsequent levels?" The goal is to select and invent details that suit the mood you want to get across.

The following is a look at how Laidlaw's emphasis on the goal, gameplay, and atmosphere can be applied to some of the key levels in Half-Life.

"Office Complex"
The "Office Complex" level of Half-Life is a deliberately drab environment composed of offices, laboratories, corridors, and stairways. In other words, fairly mundane locales about as far removed as imaginable from the usual fantastic trappings of a science fiction action game. With that in mind, Laidlaw applies the level to each of his three approaches to map design:

Goal: The goals in "Office Complex" are fairly obvious ones, such as exit doors, in plain sight. The layout of the maps are simple and, in essence, linear. However, by imposing obstacles, we force the player to take a devious route to the destination. For instance, you can peer through a pane of glass in a fire door and see your exit at the far end of a corridor; but since the fire door is locked, you'll need to explore other avenues. Clever use of obstacles can turn a completely obvious route into a complex non-obvious route. This technique was used numerous times in Half-Life. What's important is to make the goal obvious from the first, in order to give players a start on solving the problem of how to get there. And if they forget their goal somewhere along the way, they will experience a jolt of recognition when they finally gain the far-off door.

Gameplay: "Office Complex" occurs fairly early in Half-Life, and therefore gameplay was slanted toward the player who is still mastering basic skills, while preparing him for more dangerous encounters ahead. We force the player to crawl, dodge, and fight in tight quarters against ambushing headcrabs. Slow monsters, such as mawmen, give a steady sense of horror and menace, and allow the player plenty of opportunities to learn to team up with allies. He can talk to non-player characters (NPCs) and solve various small-scale room-based puzzles (such as shutting off switches to deactivate deadly turrets and electrical threats). In addition, we knew that the subsequent level would drop him into combat with extremely tough human soldiers, and in order to give him some experience fighting against squads of creatures with long-range attacks, we set up encounters with squads of vortigaunts, which are fun to fight and fairly easy to kill, thus allowing the player to develop some of the skills he'll need to survive encounters with human soldiers.

Atmosphere: The banal office atmosphere provides a perfect background for scenes of carnage and horror--the contrast with gruesome images is all the more striking because of the familiar setting. Most of the details were selected to evoke feelings of dread in the player. Scientists are dragged to their doom in airshafts; mawmen feed on corpses in the cafeteria. The one crucial atmospheric element that can't be conveyed in screenshots is that of sound. The first time the player steps into the walk-in freezer, he triggers music that evokes cold, echoing emptiness. And the claustrophobic echoing sounds not only reinforce the realism of the environment, but add to the sense of menace: Echoes are creepy.

Laidlaw then used another level in the game for contrast. This section of Half-Life takes place in a research and development lab, where the player catches a glimpse of some of the experiments that were in progress when disaster struck the Black Mesa Research Facility.

"Questionable Ethics"
The labs served numerous purposes in the game: To deepen the sense of conspiracy and cause the player to question what was actually going on at Black Mesa; to give the player access to a high-powered weapon that's crucial for battling powerful enemies in subsequent levels; and to force a style of puzzle-solving focused on protecting and working with human characters (scientists) rather than simply throwing switches.

Goal: This section of the labs is quite non-linear in layout, although judicious use of locked doors again gives it a linear flow the first time through. Non-linear areas tend to work against the usual dramatic virtues of pacing and rhythm and timing, and give rise to plentiful opportunities for boredom and confusion. They also make it easy for the player to miss or forget his goal. When designing the labs on paper, we decided that the player should see his goal (the exit) immediately upon entering the lobby. There's a scanner next to the locked exit, and by this time the player has been taught that only scientists or other NPCs can operate a scanner. Presumably the player will carry this knowledge with him as he moves away from the exit and fights his way through the other areas of the map, and by the time he encounters a group of hiding scientists, he'll understand that he's supposed to escort one of them through the labyrinthine labs back to the scanner. (To further aid the player in understanding the goal, we added a security guard to give a short briefing on the way into the area.)

Gameplay: Puzzles were built around a variety of enemy encounters, traps, and scientific devices, and assembled from elements that have some logical place in the environment: sterilizers, laser equipment, caged monsters. Getting certain weapons is tricky in that it requires exposing oneself to monsters, and then quickly working with the environment to neutralize the threat. The player must open a cage full of sharks, then retreat into a sterilizer control room in order to vaporize them before he is overwhelmed. In addition, the presence of enemy soldiers adds a constant threat and gives contrast to the alien menace. In some circumstances, the canny player may choose to wait out battles between aliens and other humans, and then deal with the weakened victor of those battles. In such cases, the player's best strategy is patience. This puts a twist on the usual action game tactic, where players usually can expect to solve all problems by direct, aggressive attacks.

Atmosphere: We selected details that seemed appropriate to the research environment, and that added to the underlying story. Certain areas were tailored specifically for the study of alien creatures, which tends to raise questions about how long the researchers at Black Mesa have known about the aliens, and what exactly they were doing with them. As much as possible, the details also provided opportunities for gameplay; for instance, sterilizers that were used by researchers to cleanse rooms of biohazards turn out to be just as effective at vaporizing pursuing soldiers. Since we wanted a mood of high-tech horror to pervade Half-Life, we avoided a lot of opportunities for wacky comedy, and instead tried to set up situations for suitably dark and ironic humor. The "Tau Cannon," for instance, is given to the player at the end of a macabre sequence. And as a bonus, in addition to the weapon itself, the player is "rewarded" with a unique piece of music that punctuates the action and puts a weird spin on the otherwise relentless mood of horror.

On Limitations of Tools
All of these aforementioned objectives must be accomplished with the tools at your disposal, and within the limits of your game engine. Laidlaw explains:

Working within your limitations can seem restricting at first, but it also allows you to plan in advance and really get creative. It would seem as if having a team of programmers available to constantly implement new features would be utterly liberating, but sometimes this makes it hard to just get down to work and use what you have. It's very hard to anticipate what you might have one day. Level designers who have built a lot of Quake maps are used to working with known quantities; they know the limits of the Quake engine, they know the entities they have to work with, they know how to create spectacular effects that take both these factors into consideration. When the rules suddenly change, and a lot of new entities and features are in the works, and the engine itself is altered, a kind of paralysis can creep in. This is a predictable stumbling block for people who are moving from mapping as a hobby to level design as a profession; but being aware of it in advance, you should try to stay aware of your strengths and the known quantities, and master the new quantities a bit at a time.

Have a Plan
Laidlaw says if you can keep it all in your head, fine. But putting your ideas on paper is good.

Sketches are helpful. You can evaluate your overall scheme more easily when it's clearly stated. If you can't sum it up in a few words, or sketch it out so that someone else will understand it, you might be painting yourself into a corner when you've put a lot of work into actual level construction.

Don't Work in a Vacuum
Laidlaw urges budding level designers to bounce ideas off a few like-minded people:

A few brains working on the same problem will often come up with more interesting and varied ideas for gameplay than one person working alone. Each person tends to favor a particular style of gameplay, a particular design approach; in the course of a long game, you'll want to represent a variety of styles, to keep things fresh.

Get Creative
Finally, Laidlaw recognizes that it's important to study and learn from your favorites, but says to use them only as a springboard to invent something new.

Everyone starts out trying to re-create the great game experiences they loved, and this is a good way to learn the basics. But you won't get anywhere rehashing old ideas. Try to think in terms of the game you would love to play, if only someone out there would finally make it. Then make it yourself.

Eric Biessman, Raven Software

As project coordinator, lead designer, and level designer at Raven Software, Eric Biessman has received many e-mail messages from those who want to know how to create breathtaking levels, as seen in Heretic II. Here he sets the record straight:

Paul O'Connor, Oddworld Inhabitants

As lead game designer on Oddworld: Abe's Oddysee and Oddworld: Abe's Exoddus, Paul O'Connor has learned quite a bit about game design (not to mention that he's been designing games in some form since 1981). O'Connor provides a different take on this chapter, since the first two Abe's games are "side-scroller" platform games, in 2D.

O'Connor asks, "Will the player understand the level, puzzle, or situation? Is it clear to the player what he must do to solve the puzzle? Is it fun?"

As each Oddworld designer works on his own levels, and as the order of those levels sometimes isn't determined until late in production, how does this particular level fit into the overall game flow in terms of difficulty and what you're demanding of the player? If this level requires mastery of a specific [mechanism], has that [mechanism] already been introduced earlier in the game? In other words, is the level design and difficulty level appropriate for the anticipated placement in the final game? How does this level advance the story of the game? What vital information does the player gain by completing this level? How does it connect with the preceding and following levels?

O'Connor comments on sketching out levels:

I do, occasionally [sketch out levels] if the situation is novel. The level editor we used on the two Abe games was flexible enough for use as a composition tool. Usually, I'd just sit down in front of the editor with a rough idea of how many screens the budget would afford for this portion of the game, and a notion about the type of play I wanted to accomplish in this area, and then go from there.

Marc Saltzman is a freelance journalist for over 40 game-related and consumer publications, including USA Today, Playboy magazine, PC Games, Gamecenter, Yahoo!, Internet Life, Next Generation, Happy Puppy, Family PC, and many more. He has also written two books on Internet gaming for Macmillan Publishing and two in-game manuals: Quake II and Sin.