The Art and Science of Level Design

Cliff Bleszinski, Epic Games
Session #4404 at GDC 2000


Disclaimer: This was for GDC 2000. Since then, Level Design has already changed QUITE A BIT. Someday, I'll write another version of this document that focuses on the latest evolution of LD responsibility.

It is becoming increasingly difficult to define the role of the team member known as the "Level Designer." Level design is as much an art as it is a science; it requires artistic skills and know-how as well as an extensive technical knowledge. A designer with tremendous traditional art or architectural experience will not succeed if he cannot grasp issues such as framerate, gameflow, and pacing. A designer who understands these elements yet has no architectural or art experience is doomed to fail as well. Art and Science are the Yin and the Yang of design and it takes the efforts of very talented and dedicated individuals to produce high quality levels.

I. Defining the Role

In the earlier years of the gaming industry, there was no such thing as a Level Designer. Programmers were the "one stop shop" of game creation; they were the ones responsible for designing, producing, and finishing products. With the evolving state of 3d technology, the need for these "digital architects" has appeared, and 3d environments are more gorgeous than ever.

Above and beyond everything that is outlined in this presentation, the role of the level designer on any given project will be defined by two key factors:

What technology will be used for this project?

A project administrator can cut down on training costs and time by hiring talent that is experienced with editing tools that are presently available to the community. For instance, if an Unreal Technology Licensee were to hire talent they’d benefit from acquiring someone who has previously created content with the editor and released it online, or has worked at another technology Licensee. A savvy recruiter will comb map design collection pages as well as closely examining the content produced by peers who are using the same technology for their titles.

What kind of project will we be building with this technology?

Taking a master deathmatch level designer and asking him to create sprawling landscapes for an Everquest style Massively Multiplayer Role-Playing Game would be a big mistake. Even if the designer were able to adapt and create great content then the time and overhead taken to train him in the new design and direction would not be worth the effort. It is possible for a designer who is "trying out" for a job to test his hand at another style in an effort to impress his potential employers, but by the time his content is presentable the job may have passed him by. Although many design elements are universal and will carry over from one style of game to another, it is key to reduce any extra time or risk that is taken in hiring new talent as budgets are constantly rising.

On Ownership

Until recently, at many development studios, there has been a notion of "ownership" in the realm of level design. A level was "owned" by a designer; no one touched his work and he was the one solely responsible for the content. Level Designers would become defensive, even hostile, if another LD suggested modifying his work.

The gaming industry is about evolution. The designers, programmers, and hardware manufacturers who do not evolve quickly fade out and die. The Level Designer is no different from these rules, much like his peers he must evolve. That said, it is no longer possible for one LD to maintain "ownership" of a level as computers and gaming machines are becoming more and more capable of rendering extremely detailed environments. The talent that is hired must be comfortable with the idea of others modifying and improving their work.

There is a direct correlation between the detail that a technology is capable of and the amount of ownership that one designer has over a particular level. With Moore’s law holding true (processor speed doubles every eighteen months) and 3d accelerators constantly raising the bar the detail that game engines are capable of is staggering. It is simply impossible for one driven person to build the necessary amount of detail into level locations in the allocated time, and the more detail technology can push the more people will be required to work on levels.

In addition to having dedicated world texture artists and environment concept designers the need will soon emerge for dedicated "prop" people; artists who create content that will fill up previously static and barren environments. Most architecture is relatively simple, much of the detail in the real world comes from the "clutter," the chairs, tables, and decorations that fill these places up.

Teams may soon see the addition of "scripting" people who are responsible for storyboarding in-game events as well as assisting in the design and direction of these events. A person of these abilities would need cinematic experience as well as excellent knowledge of the tools that are used to create "cinemas," such as a scripting language or editor.

It is very likely that the level designer will be like a chef, taking various "ingredients" from other talented people and mixing them into something special while following the "recipe" of a design document. Right now there are companies that have artists lighting levels, as well as doing custom texture work on a per-surface basis. The level designer will evolve to the role of the glue of a project, the hub at which everything comes together.

The Glue

Jay Wilbur once said, "Level Design is where the rubber hits the road."

This quote holds true today, and will continue to hold true in the future. Level designers are quickly becoming some of the most important members of a development team.

Nine times out of ten one finds that programmers are the bottlenecks on a project. A game is not supposed to ship until it is clear of all "A" class bugs, and this requires much programming gusto to clean up and ship a game. On many projects this bottleneck will eventually slide into the realm of the level designer as they’re where the "rubber hits the road." The LD is the one who is taking everyone else’s hard work and tying it together into a cohesive package. The designer takes the textures created by the artists and places them on his level geometry, or asks an artist to create custom work for his level. He'll figure out where and when to place hostile AI that was created by programmers and 3d artists while all of it is being rendered by the work of the engine programmer.

A level designer is not just an architecture monkey or a guy who throws "cool stuff" into the pot of development. Above and beyond everything else they need the ability to judge what is fun, what gameplay elements work and what do not. He needs to judge what content works in any context while making sure his work is cohesive with the rest of the game.


II. Design Commandments

Now that the role of the level designer is defined the following are some tips for him to live by.


Designer, Evaluate Thyself

The best level designers are never afraid to step back and re-evaluate their content. Often this requires a period of respite from the work in question; distance can clear up a clouded mind. A great designer isn’t afraid to throw content out or re-work a concept that needs attention.

It is also extremely important for a level designer to recognize when he is becoming tired of his own work and when his work is not coming together. There is a huge difference between the two; in one instance a designer becomes weary of playing his own content over and over and is just sick of it. A great level might get scrapped or reworked because a development cycle is dragging on and a designer feels the work is not as fresh as it used to be. The designer must recognize that his view is tainted; he has been playing this content for months on end and by nature the work becomes stale to him. This does not mean that the work will have any less impact on the user, however! At this point, a designer should have his map tested repeatedly by new and experienced players and simply polish the work instead of reworking it.


Thou Shalt Seek Peer Criticism

Assembling and maintaining a great team of designers is a challenging task. It is important to hire easy-going talent that gets along well together. A great designer is never afraid to take criticism from his peers; in fact, a great designer is the sum of himself plus his peers. Many artists feel that they’re more talented than the next, this cockiness can be the weak link in a design team. The ideal designer seeks criticism even from those he may consider "less talented" than he, because even if he believes that the critic in question has no skills the commentary will be fresh and from a new perspective. The best way to go about doing this is to have periodic "peer evaluations" where a lead designer or lead level designer picks two designers and has them evaluate each others work while acting as a mediator.


Thou Shalt Value Rivalries

In addition to taking suggestions from one another it is key for level designers to feel a desire to "one up" each other. Healthy competition in any area of a development team means improved results. However, a positive, healthy competition can quickly turn ugly as one designer may accuse another of stealing his style or designs.

If a designer is emulating the style of another this benefits the project, as the environments will become more consistent. The Design Lead should encourage overlapping designs and work towards smoothing ruffled feathers and the Art Director should make sure the environments are consistent by leading the team’s aesthetic designs.


Do Thy Homework

As much of this work is moving into the realm of the Art Director and art team, the designers remain the Digital Architects and they will still be responsible for much of the look and feel of the levels. Therefore, if a project calls for an accurate Roman Empire then everyone had better be doing his or her homework. Having a shared directory of R+D images on a server as well as an art bible that is referred to all designers and artists will contribute to a more consistent look and feel.

At any given point in development a designer needs to be able to step back, look at his work and think, "Does this make sense?" More often than not he’ll discover little details that make no sense, such as structurally impossible architecture that seems out of place or ice beasts near lava pits. The users may not notice these details on the conscious level but will sure feel it on a subconscious level which will affect his overall game experience negatively.

Good research lends itself to good planning. Some designers simply sit down and build while others carefully plan every nook and cranny of the game. The best designs are the ones that are a combination of careful on-paper planning and improvisation.


Thy Framerate Shall Not Suck

If the designers are working with a technology that can push 100 million polys then they’re going to try to make the tech look like it can push 3 times that. Although much of the framerate issue falls upon the programmers, with optimizations and level of detail technology, it is extremely important that designers have hardcoded guidelines for framerates, detail levels, and RAM usage.

The Lead Level Designer should be the one responsible for enforcing hardcoded design limitations. Unreal Tournament had extremely strict limitations on how detailed a level’s geometry could be, as well as overall framerate time.

Framerate can be sacrificed somewhat if a title is slower-paced and does not require action-oriented reflexes. However, if the team is building an action game and levels are bloated and framerates are dying then the hardcore action users will reject the title and every review will read "looks nice, runs terribly."


Thou Shalt Deceive

Pay no attention to the man behind the curtain. If a designer can simulate a newer technology with some trickery then by all means allow and encourage this. If the programmers are exclaiming things such as "I don’t remember programming that!" or "How did you do that?" then something special is going on. If a scene can look more detailed with creative texturing then go for it. If bump mapping or specular highlighting can be faked even though the engine does not "truly" support it then why not? Only the hardest of the hardcore gamer will know the difference.

This mode of thinking can be carried over to nearly every aspect of development, not just level design. Programmers can find creative ways to "fake" new effects; a sneaky artist can make a character look more detailed than he actually is with good texture mapping or smart use of polygons.

A very basic example of this would be a designer who uses "masked" textures to create the illusion of much more detailed geometry. For example, making a grate on a wall can be done by only using one polygon that’s masked instead of constructing the actual holes of the grate out of individual level brushes. Masked and translucent surfaces were used in many areas of Unreal Tournament to simulate weather effects such as snow and rain.


III. On Design Techniques

Many of these techniques can be applied to either a single player oriented title or a multiplayer oriented title. Every technique is governed by an overall concept of "Gameflow." It is the mystical "life force" that makes a good game fun and it is very much a reward-response system that challenges the gamer and then provides a "treat" for completing tasks. Time and Time again this document will refer to the "Carrot on the End of the Stick." This is the incentive for the gamer to keep going; many of these "carrots" are built and planted by the level designer as these drive gameflow.

The Gameflow of a Single Player title is driven entirely by the level designer. He’s the one who is creating a task and then placing the carrot in front of the gamer, encouraging him to complete the task. It is very much a cause and effect design, create a problem and then encourage the player to solve it. This is why many titles use violent elements as their focus as it is the easiest and most basic form of conflict.

Multiplayer Gameflow varies quite a bit from Single Player Gameflow; it is more about rationing risk and reward in a social environment. A Level Designer who is building for a Multiplayer-oriented title is much like a playground architect. He’s building the space where real people will be driving the game and experiencing the action firsthand; the gamers themselves largely dictate the gameflow. Designer-placed elements such as AI or story can often prod this along but more often than not it is the gamers who are the catalysts that keep that carrot on the end of the stick for the gamer. A title like Ultima Online creates a world where designers carefully place resources around the player and the users who harvest these resources are proud to wear their spoils of war. This creates a desire for the "have-nots" to become rich and prosperous and drives the game.

Controlled Freedom

Let the player think he has a choice in where to go and what to do but gently guide him to his destination.

This is an avidly debated topic; if a player has the freedom to go anywhere and do anything (as many gamers claim they want) then he will quickly get lost and frustrated. By keeping level design somewhat linear and giving the illusion that there are multiple paths one has the freedom to choose then the player will have a more enjoyable play experience. This way, the player experiences the best of both worlds; the player gets to the carrot on the end of the stick, and feels like he made the right decisions on where to go.

This can take more time to design but ultimately adds up for a more enjoyable single player experience for the user. It is completely possible to build a title that revolves around the notion of "go anywhere, do anything" but a developer who does this must allocate plenty of time and funding to make this a reality. Often previous titles that have had very much "open-ended" designs have had users that have found themselves lost and asking "what do I do next?" Only the most hardcore of the hardcore gamer will stick with a title that is too open-ended. It can be done; it simply requires longer design times and a more focused and dedicated user.

Pacing

Constant scares dull the senses.

The scariest horror movies are the ones that lull the viewers into a false sense of security and then spring something scary upon them, and a great level is no different. An excellent recent example of this is System Shock 2. One minute the player is being chased down by pipe wielding maniac hybrids, the next he’s tucked away in a quiet bedroom aboard the Von Braun, reading log files from dead crewmembers while wondering what will be around the next corner. If the monsters were constantly in the player’s face the game would cease to be scary. However, the down time lets the player forget, for a moment, the peril that he is in… just long enough so that his guard drops and he’s scared (and killed) by the next baddie.

The best multiplayer titles are driven by a system of good pacing through intelligent resource distribution. It requires a bit of effort in one’s skills or "character" to improve at the game and move up in the online world. For instance, in a Deathmatch or Teamplay game a player acquires his skills by learning battle arenas and how to aim. A designer’s pacing in a level will determine if the game works or not, if every character starts with a crazy arsenal or if there are areas that are impossible to breach or defend then the game suddenly fails to be entertaining. In titles that are more role-playing driven or strategy oriented a designer must be cautious with wealth and resources. If, for example, there isn’t enough ore to mine in an adventure game then players cannot build weapons and the game system crumbles. The designer is the key to making the entire game system work and often has to work and rework his ideas to make sure they balance the world well.

Risk Incentive

In single player design, there are oodles of ways a designer can utilize this time tested technique to let the gamer make his own decisions about how much trouble he’s going to get himself into for treasure.

That’s the beauty of risk incentive. The player weighs the risk; he assesses the challenge, and gets to make a decision. He feels like he’s in control, and the designer provides him with a choice.

For example, in a traditional shooter the designer might place ammunition or health below a pair of sentry turrets. The turrets can easily be avoided by crawling behind a pair of desks, however if the player wants to make a dash for the goodies it is his choice. Therefore, if the guns rip him to shreds and he screws up he blames himself, not the designers.

In a Deathmatch style game a player will have the choice of going for an ass-kicking weapon, only if he risks his neck by going into an extremely open and well guarded spot.

Never underestimate the usefulness of this technique.

On Revisiting

The concept of "Revisiting" or "Doubling Back" refers to the gamer seeing an inaccessible area of a level and wondering "How do I get there?" The gamer then proceeds to complete a series of tasks which move the game/story along (as well as his virtual self) and he then suddenly looks around and realizes "Oh! I’m up there now!"

Revisiting areas from a different angle is a good thing for designers to practice. It keeps the gamer motivated as he tears through your designs, as well as saving time and money. The same rooms are viewed from multiple angles as well as revisited, and this saves the designer from building more areas. This will be more and more of a blessing as levels become more detailed and expensive to produce in the near future.

Many Multiplayer titles are dependent upon revisiting areas of levels as Multiplayer design often focuses around character interaction instead of just exploring. A recursive design is extremely important in any kind of social title.

Supply And Demand

Leave the gamer always concerned about running out of ammunition and/or health, but not to the point where he's running around bullet-less, dying constantly, while cursing the designers and their product. This is yet another carrot on the end of the stick trick that makes for a satisfying gaming run. It teaches resource management, and makes it a better experience when the gamer finds health and ammo. Good supply and demand makes these goodies more valuable.

In a Multiplayer Title the designer has to account for players trying every available option to exploit the game and level design. It is key for the designer to balance the amount of resources that are available in any multiplayer game to prevent a tiny percent of the playing population from enjoying all of the virtual "wealth."

Scene Composition and Contrast

Relatively simple objects arranged in an interesting method can result in a far more eye-pleasing image. This is true with art, architecture and, of course, level design. It becomes especially relevant when working with low-polygon geometry and strict detail budgets.

Many art classes will spend time focusing on the idea of scene composition. This is another example where an art background will come in handy for a designer.

Work With The AI Guy

AI is tied directly into the structure and composition of a level. It is where the AI does its thing, it is the place where all that hard work on the part of the AI guy is supposed to be shown.

It is crucial for a level designer to construct areas that take advantage of the AI while working with the AI guy and figuring out what the AI is going to do. For instance, if there is an AI that is really good in firefights, ducking behind boxes and taking pot shots at the player, a designer should plan to build an environment with waist high crates all over the place. If the AI guru programs a great pack AI, make space that accommodates it. Often AI does not work perfectly, it is important to maintain patience and have faith in the AI talent as the designers manage to iron out kinks in the system.

Smart designers and programmers will work together to create memorable scenes where puzzles and areas are built around crafty artificial intelligence.

On Sound

Steven King, in Danse Macabre, said something along the lines of:

"When the lightning crashes and the door opens and you see a ten foot bug standing there, a part of you sighs and thinks "Whew, I thought it was going to be a TWENTY foot bug."

Designers must work closely with sound technicians to assure a compelling and exciting audio experience. A great designer never underestimates how much mileage he can get out of a good bump in the night. No matter how good the talent is, the monster that is in the gamer’s head is always scarier than what is seen onscreen. If the title calls for chills and thrills, let the sound do much of the work!

Intelligent Backtracking

If a designer is forcing a gamer to backtrack he must make sure that it is done in a logical and non-frustrating manner. This is a dangerous time in design, as the "carrot on the stick" of seeing a new area is gone. A designer is re-using a previously seen area and it is important to make the area seem fresh or interesting as the player navigates it. This often requires subtle scene changes, or the addition of new hostiles to prevent the area from seeming "dead" and "used." Much like a used-car dealer will polish up an older model, a designer who is re-using an area must put more effort into it to make sure that it seems new and fresh.

It is also key to make sure that the gamer does not get lost as he is backtracking. If, for example, a gamer must activate a pump so he may drain an area with waste and cross then the route back to the previously hazardous area had better be pretty easy to navigate in reverse. Using controlled freedom here will ensure that the gamer knows where he’s going; perhaps by blocking off a redundant area or placing highly visible signs that direct him on where to go he’ll have more fun.


IV. The future

Gaming continues to evolve and is heading in a variety of directions, and level designers will be at the forefront of this revolution. Programmers will be responsible for entire tool sets that the designers use and it will be important to have a good synergy between designer and coder.

On Editors

Level Editors will become closer to high end modeling packages such as Lightwave or 3dstudio max, as real time scenes are approaching pre-rendered ones. Many developers have forgone traditional in-house level editors for packages such as these, so it can’t hurt for a designer to learn Max, Maya, or any of these programs. Chances are, in-game editing tools will be at a similar level of complexity in the future.

There are basic 3d editing concepts that these programs are built upon that everyone should know, a designer should understand how to manipulate low polygon geometry as well as high polygon geometry.

On Texturing

In the past, geometry was extremely simple and nearly all of the world detail was done in the textures. Many current titles feature approximately a 50/50 ratio of texture detail to world geometry detail, levels feature many custom textures, a simple polygonal arch will be framed by a custom texture that makes it look that much more detailed and planned. The real Next-Generation titles will feature a more detailed "material" system where simple maps are mixed to create realistic surfaces. For instance, a designer will be able to specify the shininess, depth, and color of any material that will then be placed on world geometry.

On AI

In the future Designers will have to work even more closely with the in house AI programmers. On one hand, design will become easier as AI will become better at tasks such as navigation and conflict, while on the other hand the job will become trickier as users demand more and more cinematic experiences. Smart designers will build many custom AI "scenes," such as exciting stand-offs between hostiles and teammates while building in a failsafe "backup" AI that keeps the scene convincing if the user "breaks" the action. By "breaking" the action the user may, say, blow up a character that is supposed to jump through a window or trigger some sort of action.

On Skills

Every level designer at Epic Games has primary and secondary duties. Some designers, besides working on levels, are competent texture artists. Others are good at modeling characters or decorations. As the level designer evolves it will become more and more important for him to be familiar with many of the tools that artists and 3d modelers use.

This reduces any potential "middle man" time risk. For example, if a texture artist has created a great brick pattern for a designer before leaving for the day and the texture does not tile horizontally on a surface correctly the designer can open the art in PhotoShop and make it tile himself.

A designer who can "do it all" is both a blessing and a curse. If he can create his own textures, architecture, lighting, and decorations then he’s an all-in-one package, a one-man design machine. However, designers who are this talented often have their own notions about how they want "their" work to look and can be very difficult to work with when the time arrives for "shared" design. Another problem with a "do it all" designer is that his time is divided between texture creation, world creation and decoration creation and often finds that he’s bitten off more than he can chew! These designers often require the least amount of management at the start of a project and the most at the end when they’re struggling to finish all they’ve started.

Conclusion

The last quarter of a game’s development cycle is the most crucial for the entire team, especially the level designers. Features that are often broken will finally be working and the game can be solidified and polished. Truly talented designers will shine in these moments.

It is important to remember that, much like many sports, creating a game is not a one-man show. As important as level designers are for the team, they are nothing without quality programmers and talented artists to back them up, and vice-versa.

The gaming industry is constantly evolving; a short while ago there was no such thing as a "level designer" and now they’re key team members on a project. The level designer needs to understand where he fits in amongst the other talented people he works with, and needs to have an open mind and a good artistic sense if he’s going to help put everyone’s hard work together into a fantastic product.

© 2000 Cliff Bleszinski. Not to be reproduced without prior written consent.



© 2002 Cliff Bleszinski