Stefan Doetschel: The Principles of Level Design

Stefan Doetschel (s-dot.de) is a well-known level designer, who worked on a number of very interesting projects throughout his career. Having a strong architectural background he had the pleasure of designing maps for BioShock Infinite, XCOM: The Bureau, Borderlands: The Pre-Sequel, as well as good old UT 2004. In this exclusive interview Stefan talks about his work, shares advice on good level design and talks about the difference between environmental artist and the level designer.

About Stefan Doetschel

I’m born in Germany. After finishing high school and having had a few jobs and a completed apprenticeship as a joiner I studied architecture and worked for about 14 years.

When the first Unreal Tournament in 1999 came out it had a working Unreal Editor included and I started making death-match and capture-the-flag levels for fun and to play with friends. We had a group of people playing UT regularly on LAN parties or against other teams. I also started getting involved into the mod scene. I made levels for Tactical Ops (which eventually got published by Atari) and a mod called Domain 2049 which made it onto one of the special editions DVD’s released for UT 2004.

In 2004 I applied for a job in Sydney (Australia) not expecting to get it but I got hired. So I quit my job as an architect and moved to Australia. The game was never released and the company closed down which isn’t unusual in the games industry unfortunately.

Then I moved on to work at Krome in Brisbane for a few years mostly on work for hire projects like Star wars: The Clone wars or Hellboy. In 2010 I arrived at 2K Australia. After struggling with an earlier version of “The Bureau” we moved on to help Irrational Boston with BioShock Infinite. Then we worked on Borderlands: The Pre-Sequel and the Claptastic Voyage DLC with help from Gearbox Software.

Borderlands: Pre Sequel © 2K, 2015

BioShock Infinite was definitely the largest project I worked on. At 2K Australia we spent a long time on the unreleased multiplayer part of the game before we helped creating the last chapter of the game.

Personally I rank The Pre-Sequel higher than Infinite because our studio had a bit more control over the final product. Development was a cut short but I’m proud at most of my level-work for this game. Especially for the level Crisis Scar which was the first level shown to the public.

What Does It Mean to Be a Level Artist

Bioshock Infinite

In the old days the level designer was also the level artist. But with every new generation games are becoming more complex. Today I think most companies have Visual Designers on one side and Level Gameplay Designers on the other side.

What the exact difference between these two roles is depends on the company you work for. Some companies leave most of the level layout work to the level artist while other companies let designers build a grey box for the initial layout before the artist takes the visual side over.

The major difference is that the visual designer is responsible for the visuals in the level while the gameplay designer mostly works on enemy placement, objectives and most actions the player can do. Every company I worked on had a slightly different process. Sometimes the responsibilities can be a bit blurry.

The environment artist doesn’t design anything. He just builds the environment to the requirements given to him by a designer.

Principles of Level Design

Bioshock Infinite

This is a bit of a million dollar question. Visuals and design are subjective to a large extend. So what one player likes another might dislike. But there are a few basic rules for good level design of cause.

Most of the time there is a story for your level and your main goal as a level designer is to make this story work well for the player. The player is your client. I always try to play the level as someone who has no idea what will happen in it. The gameplay being implemented early on helps in this process. Most players will play your level only once so it’s important that the design is easy to read. But there should also be a lot of detail in it that can be discovered when the player takes the time to explore it. It’s a 2 way street. If there is nothing to discover than the player won’t even try to explore.

The levels hould be built around something interesting that sets it apart from the rest of the game. This could be something physical like a large structure on a volcanic lake or it can also be an idea like a fun park where environments are changing during the play through. This will most likely be defined by the game’s story script for this level.

Once I got a rough layout I try to identify the main structural features and then try to simplify them and make them stand out more. A clear simple structure is often better than a mess of small ideas. Good visuals need to be somehow cascading (not sure what the best expression would be). The player needs to be able to read the main structures from the distance. The closer he/she gets the more detail unfolds down to the very minor detail like grunge on the floor.

Visual storytelling is also an important part. Make the level tell its own story. Often the spaces or rooms have no actual people inside but someone left a cup of coffee and a newspaper behind on the table. The chair fell over and parts of the paper are laying on the floor because he left in a hurry.

Also make sure it feels and looks like a real and bigger place. The player can’t go everywhere but there is more out there if he/she would be able to open that door or climb over that wall.
Finally lighting can help putting focus onto some parts of the level or direct the player along the path. Like often less is more here. Rather have 2 distinctively lit areas than a hundred little lights everywhere. And last not least the post processing will provide the mood for the level but can also help merge elements together.

Working on games Borderlands

It was really good actually to work on Borderlands after a short period of getting used to the style. It’s very forgiving when you don’t need to be super realistic. We were able to do things in Borderlands that would not work with another style. We could kit bash together pretty much whatever we wanted.

The visual style in Borderlands depends a lot on the textures and the silhouette of the structure (everything is outlined). The lighting and post processing are quite strong and can merge objects together that would look out of place or out of scale in a more realistic game.

Borderlands has different factions like Dahl, Hyperion or Bandits which have clear design rules. Your level would be either Dahl or Hyperion but not both. The faction defines the style of the level and makes everything consistent within this area.

Randomness is very limited in games for several reasons. In Borderlands it was mostly the type of enemies that spawn.They do so from a pool of possible enemies suitable for the environment. There are a number of possible locations set up by the designer and the enemies would always spawn in sight as far as I know. As so often in games it is about the illusion of randomness and not so much about being really random.

Changing daylight is also something that could be considered random. We just had to make sure there is good lighting during the night phase.

The Advantages of Unreal Engine

Unreal’s biggest advantage is that it has successfully been used for a long time and for a lot of games and in different styles. It is tailored towards the content creators. Anyone can start building a level in Unreal Engine without the help of a programmer. Artists and designers usually love Unreal for exactly these reasons.

Epic games do develop their own games. It’s made by people who understand what’s important and what isn’t during developing something as complex as a AAA game can be.

Not to forget the huge amount of free and good tutorials and content out there for this engine.

I started working on projects that simulate real environments for architects and property developers in Unreal Engine 4. I think the engine works very well for that. Especially the new physics based material system seems to be a huge improvement.

Occasionally I am having trouble finding some of the features from Unreal 3 but most of the things are still working very similar in Unreal 4.

Blueprint seems is a logical step forward from Kismet. Gearbox developed something similar for Borderlands 2.

Tools for the Level Designer

I’m not very tool savvy. Having worked in quite a few different company I try to focus on the major tools everyone seems to be using. 3ds Max is a pretty good modelling tool for my needs. Photoshop is commonly used for textures.

Then I use Crazybump for detailed normals or spec maps and substance designer to bake low poly objects from high poly objects and for the AO and normal map.

For the initial design I use paper and a pen or occasionally a tool like illustrator or CAD. Once the simple grey-box is built I do paint-overs of screenshots or just try to design directly inside the level with large temporary structures.

Verticality in Level Design

I’m not totally sure if this is true for many games. Borderlands 2 for example has huge height differences. Maybe they are not so visible to the player. Dishonored, Mirror’s Edge and Assassin’s Creed are examples were the game play is specifically designed to use verticality. Other good examples are Uncharted, Tomb Raider or God of War which use a lot of verticality. But for a very realistic game it might not make much sense to climb up walls.

Some levels are unnecessary flat. The only explanation that comes to my mind is that designers often start a level on paper. It’s difficult to design a 3 dimensional space on a 2 dimensional medium.

For ThePre-Sequel we based our initial design too much on the Borderlands 2 gameplay. How we would use low gravity wasn’t always that clear until late in the project. So some of the areas are not as vertical as they could be. We did a better job in The Claptastic Voyage DLC.

The Modern Structure of Levels

Good pacing between sequences of action and more quiet areas is very important. Enemy AI works best in larger arenas and it’s not much fun to fight in a tight space. So that’s why we often have slower paced corridors and high action arenas.

A connecting space doesn’t need to be a corridor however. And every rule should be broken once in a while to make the game more interesting.

Some developers don’t want to take the risk and try something new and rather repeat what seems to have been working for previous games. Another bad habit is that lazy developers just try to pick a few ideas that work in other games and try to combine them into their own game. This can become obvious and predictable.

Or it might just be that we designers are sometimes trapped too much in our own rules. Level design is very complex and you need some basic rules or you get lost. But always keep in mind that rules shouldn’t be followed just for the sake of following them. This can block creativity.

Finally bad design could be a result of a personal crusade of one of the developers. Players will have a hard time understandingit because the decision wasn’t made for the player. It’s sad but it happens. Game developers are only human after all.

Simplicity in Level Design

Artists or designers can’t really replicate or create a super detailed environment. Often it’s hard to understand the engineering behind a complex structure and the end result is a simplified version of it. In most cases it works because the player doesn’t understand it either.

There are also technical restrictions like the amount of detail that can be rendered or fits into memory of the console. And it might be a problem of the budget both time and money. It just takes time to create detail so we have to compromise.

An indie game can’t really produce the level of quality a AAA game can. Some ideas work very well on their own and super complex environments would only distract and reduce the fun.

The developer or publisher might be scared that the player wouldn’t understand a more complex game. I’m all for taking the player serious and trust him or her to figure out difficult tasks. It’s a walk on a thin line and sometimes it’s too much and sometimes it’s not enough. Focus tests or letting some friends play can help here a lot.

Still – games are complex and a lot of different personalities are working together. I’m always amazed when everything finally comes together. But it is really hard to predict the outcome if you don’t have the luxury of unlimited time and money.

Stefan Doetschel, Level Architect