Answers to the level design questions you’ve always wanted to ask

War Robots Universe
MY.GAMES
Published in
18 min readSep 22, 2023

--

In the last article in our series for level designers, we look at common questions that didn’t fit elsewhere: on career, best level designs, the boundaries of level design, 2D vs. 3D, playtesting, career tips, and much, much more.

Hi! My name is Vasiliy Skobelev, and I’m a Lead Level Designer at Pixonic, MY.GAMES. I’ve been promoting level design for a long time; this has included teaching, speaking at conferences, conducting live streams and master classes with Q&As, participating in podcasts, writing articles, maintaining a blog, and editing books. Over time, readers and listeners have asked me a lot of important level design questions (a good chunk of which I wouldn’t have thought about otherwise).

So, this article will attempt to answer those. It’s an addition to my series of articles on level design, and I’m also going to cover everything that wasn’t included in my earlier articles. In the process, we’ll again talk about ideal level design, the entire process of developing a location, the skills a level designer should have, and much more.

Previous articles in the series:

On ideal level design

Which game do you consider to be the gold standard of level design?

Well, if you want to see examples of cool level design, you can go through almost the entire portfolio of Arkane Studios. They have absolutely amazing solutions and guidelines. It’s also very interesting to read their materials about the pillars they’ve basically built their designs on; if someone has played, for example, the relaunched Prey, then you might recall (drum roll) that there are no vertical ladders throughout the entire game.

The developers didn’t like the “empty” gaming experience where players would just stare at a wall while on a ladder. Thus, a player can move across all height differences, all elevations, without the use of ladders. Arkane has many examples with thought-out solutions like this.

If you’ve played the second Dishonored, then of course, you warmly remember Jindosh’s manor, that is, Clockwork Mansion. For those who haven’t played this one, this mansion features a bunch of complex, transforming rooms that completely change the layout of your surroundings. That’s great, but, in my opinion, this is not the most interesting part. That distinction belongs to a different mechanism: there is a point later in the game where the character receives a piece of glass that allows them to see the same place in a different time period — and literally, just like that, by pressing one button, you can travel in time.

As a result, we have three versions of the location from the player’s perspective: initial present, past and changed present (which function simultaneously, since we can also see opponents moving through the glass). Again, kudos to Arkane, who was one of the first studios brave enough to experiment with this kind of feature.

A side note: while it’s fun to play Dishonored both in stealth mode, and while beating everyone black and blue, in fact, Dishonored shares a problem common to all immersive sims (and it’s better to point this out sooner rather than later): the game claims to support a bunch of playstyles, but in reality, it has one effective build that allows you to get to all the available locations. Let’s recall, for example, Deus Ex, where you can find a door that requires some high-level upgrade to use.

Naturally, this makes you feel frustrated, and then you find a ventilation entrance somewhere nearby and can still get into the room behind the door. And then you see that in order to actually explore all the available locations, you need to have: hacking skills of the 3rd level, arms’ strength of the 4th level, and so on. There may no longer be replay value here, because the meta comes into force, which implies a single effective build. In Dishonored, the situation is similar, because going into a head-on collision is much less beneficial than upgrading a teleporter or, for example, slowing down time, and crushing opponents one by one.

Back to level design, there is also the second Titanfall — it has excellent game design and level design. There is an absolutely amazing level with time travel via parkour, and that’s pretty cool. There’s also a location with a conveyor that collects modular portions of the play area piece by piece, and this feature also says a lot about their design abilities. In general, I’m not going to spoil your gaming experience — try it out and you won’t regret it.

There’s also Cyberpunk 2077, and despite all the shortcomings of the game, I think that the design of the city itself is great. (Although, it’s a pity it wasn’t released in the originally planned form, for instance, there was always supposed to be a subway, but it was pretty much cut out of the final version.)

Another note: I also like the fact that the car mechanics take up a decent part in the game: that is, these mechanics are not rudimentary. But the game seems to have quite serious problems with combat, because, first of all, not all builds can effectively go through certain combat encounters, and second of all, the general principles of level design don’t work everywhere. All major projects have this common issue when something changes in the production process, which leads to a reworking of all levels/maps. But developers usually don’t manage to fix everything, and that seems to be the problem here as well.

I went through the entire game and saw all the endings. If you’re OK with the current bugs, go ahead and try it out. Oddly enough, the most stable client is on PC.

Maps for MOBA games like DOTA are very old and similar to each other. Are these examples of ideal design or just tradition?

Well, somewhat both. First, I’ll note that such games often have one map, and this has actual historical reasons. For example, Call of Duty can have two dozens of online multiplayer modes, not to mention the number of maps. But most of those disappear after sales have peaked. The fewer the modes, the more popular they are among players. If we follow this logic — if there’s one map, it will always be played.

Why are they like this? This is a combination of “survivor bias”, luck, and really good design, and in DOTA this is a classic “clover” — a three-line level. It’s well-balanced, easy to understand and to memorize. This was a considered decision that was done for a reason and this is a design that works. Even in first-person shooters, they often reference a DOTA map and make a two-three-line location with a clear combination of segments. In fact, if you look at the DOTA map and then at DE_Dust2 from Counter-Strike, you’ll spot some similarities.

What are the most unusual solutions you’ve seen in level design, like Silent Hill 2 with its fog?

Good question. As I said, in Titanfall 2, the time travel thing mixed with parkour and the level with an assembly line. What other interesting things? The Psycho Krieg DLC for Borderlands 3 has absolutely amazing level design; the entire DLC, from top to bottom, is just the best. I won’t go into details, because I might spoil your playthrough. Here’s a screenshot from my gameplay anyway, and I’m sure you won’t be able to understand where the top and the bottom are located from this.

In addition, if you want a really cool level design that relies on player choices, pay attention to the single player campaign of Call of Duty: Black Ops Cold War. The last 2–3 locations are rearranged on the fly depending on how the player chooses to navigate them. Hats off to the developers — no one has done such a cool thing for a long time.

From concept to playtests — all about level production

Where does level development begin?

Level production pipelines are different, and this very much depends on the studio. When a level designer comes in and says, “I have an idea, let me show you a blockout,” — is a perfect situation which rarely actually happens. It’s good when there are opportunities like this for level designers, and this is an amazing growth opportunity. Then the team can start the concept stage from completely different moments in time. But it also can be the other way around: especially big studios can start with concepts when they’re looking for direction.

There are also those who start with art passes first: they show artists the locations and the artists then sit down to work. Art directors and, in general, everyone who then gets involved in the process should accompany the artists to these locations to scout them out. Rockstar is famous for using this approach.

A person who decides what is more important in the game (graphics, gameplay, or something else) should also be involved at this stage. This does not mean that everyone involved in the process should be fighting each other. If graphics are more important, first you create a visual part and try to keep it and not break it in the process. If the gameplay is more important, at first you create everything using flexible tools with very primitive graphics that favor several easily-implemented iterations.

Should a level designer focus on the small details, that is, arrange small stones, puddles, and so on?

The bottom line is, if that puddle doesn’t affect the gameplay, and the level doesn’t get more interesting because of it, it’s not your job. The level designer’s job is to make the level interesting, and the artist’s — to make it nice-looking. If water slows down a character or activates swimming, of course, it’s your job to make it. If you need a puddle to make the light reflect nicely off the surface, that’s the artist’s job.

You can keep improving a map forever, so when do you realize it’s time to stop, that the map is ready?

To solve this issue, while in production, clearly indicate under what conditions the map is considered ready. If your game is partially complete and it doesn’t have finalized locations, you’re doing something wrong. You can’t develop all levels in parallel before release — you’ll shoot yourself in the foot. So, make life easier for yourself and state that you consider a location to be “complete” when it meets certain requirements. Afterwards, contact the level designer only to restore or support the gameplay of this location. Next, work with art to make levels meet the art requirements.

It’s good when there is a lead level designer or a lead game designer who signs off on your locations and says: “The LD pass is ready. The next LD passes will be for support only.”

It’s also good when there is an art director or an art producer who says: “The art for the location is ready, if something changes then we’ll just re-bake the light and navigation.”

Can an artist “break” my level?

You shouldn’t hold out hope that a product, upon being sent to another department, will keep everything intact. Other specialists might have their own ideas, and their management might also see things a different way on some points. It’s not your job to make sure that no one “breaks” anything. Instead, you need to identify the limits within which they can break things. To illustrate, I’ll hand over a level and say: “You can change this object as you like, it doesn’t matter the shape or anything. But this is an art dominant, so it should be visible from anywhere.”

Remember: we’re responsible for functionality. The visual elements are an artist’s task (and anyone who will change the map in one way or another), you just need to agree on the limits.

How do you know if players like a level?

Playtests, playtests and more playtests. First, carried out within the team, then with the studio, next with a slightly wider audience (some kind of test group) and then, for example, via all sorts of soft launches, CBT, and so on. Let someone unbiased (who isn’t a developer) play — let your grandmother, mother, or some other relative play. And observe how all the basic principles of level design apply to them: what catches their eyes, where they move, and so on.

Different forms of level design

Does the complexity of developing 2D and 3D levels differ?

Yes, definitely. Actually, I would consider three categories:

  • The first one is purely 2D games
  • The second is what used to be called 2.5D. That is, when your gameplay takes place on a plane, but in fact, the game is three-dimensional. For example, Duke Nukem: Manhattan Project is a very clear example of this genre
  • 3D games, where players move in all three axes — whether a third-person or first-person game

Here’s an example from my personal experience: we made locations for Vector, there were some nuances, difficulties, and unique/rare things which later weren’t oftenly repeated in 3D games. For example, we had parallax backgrounds that shifted with character movement. We had different silhouettes that were molded from different sprites and assets. That is, not all art was unique, some of them were combined. Further, we had colliders like in super-old engines, they were physical entities with which the character interacted — and graphics were separate entities. In modern 3D games, this is usually a combined entity — the collider and graphics are one object, and it’s been like that for quite a long time.

In 3D games, no matter what kind of camera spring you have, first-person or third-person perspective, there are many nuances in 2D games such as Ori that don’t exist as a class — for example, baking light, working with perspective, and so on.

Is it harder to work on procedurally generated levels or on handmade levels?

This is a great question that few people ask. I worked with procedural generation on Vector 2, and the load on level design was many, many times higher than on the prequel.

Let’s bust one very strong myth right away: procedural generation doesn’t mean that you have a magic fairy who generates great, nice-looking playable spaces — it doesn’t work that way. In procedural generation, you have level segments that need to fit together and be playable. (Preferably with a minimum number of restrictions so that you can connect them in any sequence perfectly with each other.) This ensures stable pacing.

In Vector 2, we had a rather raw generator with a bunch of conditions. Each run takes place on several floors. Each floor has about nine rooms with an entrance and an exit, and inside each room we had a different number of segments, at least two: start and finish. I think we even had rooms with four: start, middle 1, middle 2 and finish. Each of these segments had an average of ten options, and each of them had to match with each other. Can you imagine how hard it was to design it?

Let’s give a bad and a good example illustrating how procedural generation works.

Bad example: No Man’s Sky on release — it’s alright now, but at launch it wasn’t very good. They made this standard error that procedural games developers make: 80% of the awesome content appears within the last 20% of progress. Those who reached the center of the galaxy saw some awesome things that were featured in the trailers, and those who didn’t felt rather frustrated.

On the other hand, a great example of procedural generation is Remnant: From the Ashes. Each biome is shuffled segmentally, so that in one playthrough you won’t see all the enemy types. And the generator works in such a way so that you’ll only see all the content if you go solo and periodically join a friend to help go through the same maps because there is enough content for at least two different playthroughs. The generator also takes into account which bosses you’ve already met, and those your friend met. It’s great, check it out.

What is important to pay attention to when developing an open world?

Let’s say we have some projection with an open world — let’s define what it is at the start, technically, it’s not one large map, but a mosaic of map tiles. Those who say that there’s one map in Fortnite, are fundamentally wrong: it’s a lot of maps, just distributed along a matrix (a grid if you like), and the load on developers is not much less than when developing a single game with a bunch of consecutive levels.

What to pay attention to? First, when you get a request to make this type of map, check your health insurance and what it covers, because not everyone goes through this trial safe and sound. As for the constructor and automation, very often in the gaming market, designers are required to make everything nice-looking. But, don’t be fooled because the location should be delivered for fast iteration.

In terms of high-level things where you distribute points of interest around the map, there is a thing I call “POIs Diversity Rule”, when a character standing at any point in the world, sees two or three locations, which offer different gameplay experiences. Before starting full-fledged production of points of interest, take a map, mark it up, and see how many of them you ideally need. This is better to put together manually, and preferably involving not only one lead, but several designers. It’s also important to avoid having too much content to bottleneck the pipeline.

Open worlds are top-class maps if done well, but they have a matrix structure of levels with no guarantee where the player will go, so you have to make it as interesting as possible regardless of the route. This is a very significant task. In my opinion, recent releases like Assassin’s Creed: Valhalla and Ghost of Tsushima have done a very good job. Things are very well calculated: no matter which way you go, you’ll come across adventure, for sure.

On the ideal level designer

How in-demand is a level designer these days, and should they be able to do something else: concept art, 3D, game design?

I’ll give a quick overview of the market situation: there is no shortage of level designers. Instead, oddly enough, there is a shortage of specialists who can interview them and check their skills.

Unfortunately, if you post a resume on LinkedIn and tick an “Open for opportunities” box, recruiters will start asking: “Where is your ArtStation?” And then you’re like, “But I’m a level designer.” The most interesting part begins after that: now, everyone thinks that level designers should have static screenshots in their portfolio — and they must be nice-looking. Yes, there can be screenshots, but they should also display clear layout, navigation, how the level looks from the player’s character perspective, and not how “nice-looking” it is. In an ideal scenario, there should be a walkthrough video, and even better, a playable build. There are candidates that meet this criteria out there who just need to boost their self-promotional soft skills in these terms.

I would suggest focusing on the requirements of studios like Insomniac, Arkane, etc., and focus not only on art, but also on visual scripting in order to build a full-fledged gaming experience. If you do that, you’ll score extra points, and when working on the project pre-production, you won’t constantly need to bother the engineering department asking them to implement some kind of game mechanic; the implementation might be all crooked or imperfect but we’ll be able to try it out early — and this will help a lot on any project.

And how in-demand is “pure” level design now?

I often come across questions like: “Should I do a 180 and start working with art, or learn how to work with the light?” This can be done, and very often companies require the level designer to do the dressing, art, environment, and so on. But if you think that you should be hired as a level designer, then you should be hired as a level designer — and only as a level designer.

What have you done to let your ideal employer know you exist? A lot of people usually don’t do anything about it. Get to know people from the industry. Ask people from a particular studio. Maybe they haven’t posted the job posting yet or there is an unannounced project. Stick to your line and sooner or later you will find your job.

Another important point: many believe that you first need to build up a significant portfolio before trying to get somewhere at all. However, it’s better to search for employers right away, because otherwise you may miss the moment when a slot is open. You can work in the industry without nepotism and word of mouth and without an open portfolio at all — I’m a living example of this: you show your projects and say: “I’m ready to do a test for you,” you do it, and if everyone is happy with the result, then it’s done. Generally, an employer who is really looking for a specialist will definitely give you a test.

Is lighting up to the level designer or an art professional?

In short, lighting is an art task, lighting artist in particular. That said, level designers should at least understand how light is built. So, if you work with Unreal Engine, you should understand, for example, the mobility of lights, why they’re stationary or moveable, the engine’s limit for the number of simultaneous moveable lights in a level, how light is baked, static and volume light maps, and how to apply the light to the blockout to make the location playable and readable.

Those are basic things, but if you’re making a blockout of an interior space and it’s dark because it doesn’t have a light source, that’s on you — it’s not the lighting artist’s problem. Your task is to provide a playable design that can be understood without you whispering things over the shoulder.

On another note, if you understand how it’s technically arranged in the engine, you’ll be able to help the art department at many stages and, in general, you’ll be quite a useful specialist. This has helped me quite a lot in my career.

Does a concept artist need to understand level design?

This is the same thing when we discussed the limits of changing things in the location. Concept artists, as a rule, don’t work directly with game content, they simply take it as a reference and make concepts based on it. There needs to be a dialogue. This is crucial.

In the case of some jack-of-all-trades who also creates the game design, if they’re an artist, then these skills will be useful, but the main thing is that the concept should be their priority.

How deep does a level designer need to understand player psychology?

The basic principles of level design are like markings on a road. They point to how a player’s attention should be directed. You need to understand how to work with a person’s motivation via visuals.

In a nutshell, yes, you should understand the basics of how to keep the player’s attention. But this does not require an academic education. It’s enough to just play games — especially within the genre you want to work with.

What field has the easiest entry level at the initial stage?

Actually, any field. But, of course, now I’m looking at this with all my experience. But when I first got a gamedev job, the main thing was just to get any job, and I would figure out the details later. I don’t recommend this. Aim at a specific position, and don’t consider the positions as temporary; very often, for instance, people think that QA is a temporary position — it’s not like that at all. QA has a separate career ladder: testers, QC, QA, each with their own grades.

If you want to be a level designer, monitor jobs at sites like Linkedin, Talents in Games, stuff like that. Contact recruiting agencies — go straight into level design and gain relevant experience.

Yes, at first you might not join a project of the genre and engine you want to work with. But if you want to be a level designer, as my colleague said, get ready to be bad at it, and make a lot of mistakes. There are often job openings that require no work experience, like internships. Take advantage of them.

Recommendations on reading

In terms of game design, my colleagues recommend “A Book of Lenses” quite often. But I recommend reading it and questioning what is written there — that way it will be more useful.

For level design, there is “Designing Virtual Worlds” by Mikhail Kadikov (Level Designer at Crytek) — that’s a good one. But, if someone wants a slightly more advanced analysis, then it’s better to read his blog, in my opinion.

Max Pierce has two books: one about level design, the other about encounter design. There is also Tommy Norberg and his book. And don’t forget about the “Level Design Compendium” — the largest level design library on the web.

Also, take a look at the previous articles from our series, perhaps they will be useful to you. Links to are at the beginning of this article.

In general, if you want to set out and make levels, open YouTube and scour the playlists of the game engine developers, starting with Level Creation from Epic Games or some top notch designers such as Patrick Haslow, Max Pears and others.

There are no barriers now, and you can immediately start doing something of your own. Perhaps it will be simple and primitive, but the important thing is that you did it yourself.

Keep loving games, and see you soon!

--

--

War Robots Universe
MY.GAMES

Behind the scenes of gamedev. Creators of War Robots franchise from Pixonic team at MY.GAMES share their secrets and experience.