[In the third installment of his Action Adventure Level Design series, Lara Croft creator Toby Gard examines how the design process should incorporate discussions of pacing, structure, and mood -- and how leads can hone their feedback to the team to make it all work. Part 1 described how to create a Level Flow Plan to hand off to the level team. Part 2 described a variety of tools to help turn those Level Flows into detailed, immersive and interesting levels plans.]

By the end of the process described in the last article -- building through fiction -- you will most likely have a mixture of paper maps, written stories, detailed flowcharts, concept art and possibly some 3D mockup spaces, depending on how each level team prefers (or has been instructed) to represent their plan.

Those levels will have taken shape in surprising and unexpected ways. Levels that we had assumed to be straightforward action levels may have revealed rich veins for puzzles, and many levels are likely to have prompted ideas that fall outside of the current game mechanics.

Evaluating the Big Picture

To structure their feedback, the creative leads need to validate all level plans in relation to each other. Because the levels are likely to be pretty complex, it is useful to create a simplified representation of the whole game so that you can assess the pacing and emotional consistency of the experience.

Extraction of Mechanics

The first step we need to take is to identify all of these special case interactions and ideas that the level teams have come up with while fleshing out the level plans. Inevitably they will be some of the coolest in the game:

Ken Kong falls down a 30 story lift shaft, doing frantic mid-air kung-fu until there is a pile of zombie bodies beneath him thick enough for him to survive the drop.

It sounds awesome, but the fight system simply cannot accommodate this "fall fighting" mechanic, so the level team has suggested it as a cutscene.

In a couple of other levels, Ken Kong has to destroy some walls and the level teams have proposed different McGuffins to allow him to do this, such as a convenient, precariously balanced heavy object that will break through the wall if triggered.

It is this list of ideas that can produce the neat and original game mechanics that will set your project apart from everyone else's. By promoting ideas that have the flexibility to be expanded into the core mechanics and peppering them throughout the game, we can create a richer more coherent overall experience.

For example:

How could destroying walls become a reusable mechanic? Would it require a consumable, or is it a readily available ability? How rich of a vein is it to be tapped for more applications? Does it have synergy with other player abilities?

Let's say that we can integrate destroying walls with a new survivor type, a demolitions expert, who carries around explosives that can be put to all sorts of uses, but who also explodes when attacked by a zombie -- potentially taking out a large proportion of your crowd. This could make for an interesting risk/reward mechanic and with some standard "explodable" barriers and/or enemies could be used in several levels.

Perhaps the "fall fighting" could also be used on several levels, but this seems more like a mini-game than a new mechanic. While the idea is interesting, the question is, could you make the gameplay deep enough to justify three or four "fall fighting" sequences throughout the game? It potentially seems like a large investment for too small a gain, but if we could make it work, it would be really cool.

These mechanics are generally gold, because they were not forced into the game design from a desire to tick boxes based on competitive products, but were discovered organically through an exploration of its unique themes and the thoughtful exploration of its world.

Once we have integrated the new mechanics and rejected or noted all the new set pieces, we will have adapted the character to live in this more clearly defined world and gathered a major part of the information needed to give feedback to the level teams.

Gameplay Types

Most games have a basic mixture of elements. For instance, an FPS might have 70 percent shooting on foot and 30 percent vehicle combat.

If every level in the game had exactly that mixture of gameplay, it would get dull for the player pretty quickly. But if you have levels that are entirely on foot, interspersed with a few levels that are predominantly or entirely involving vehicles, then they will act as palate cleansers, changing up the experience enough to keep players interested.

By looking at the mix of gameplay types over the course of the game, you can isolate points where the experience might be too flat.

A great example of a game that keeps the player constantly interested is Half-Life 2. Almost every level has a new central theme, whether it's a new weapon, a new vehicle or a new type of enemy, your experience changes dramatically every thirty minutes or so.

Example: KFZK

Let's carry on with the imaginary game Kung Fu Zombie Killer, discussed in depth the last installment. The variety of gameplay in that design comes from the types of survivors that you rescue.

Let's assume these were the locations we settled on for the levels:

We know from the story that the game has to start in Ken's Dojo and that it has to end with camera men filming Ken as he rescues jenna126xyz.

We have goal mix of 80 percent fighting, 20 percent puzzles for the whole game and we had ordered things like this:

But during the detailing phase two things happened. (More likely a massive number of things would have changed, but let's keep it relatively simple.)

First, someone came up with a really cool teacher survivor who can put zombies to sleep by lecturing them, which changes the gameplay mix at the college to involve more puzzles.

Second, someone has proposed changing the cinema into a film studio, whereby the zombies and the survivors can be based on clichés like Wild West or Godzilla films. People are very excited about this idea and enough crazy mechanics have come from it to justify potentially splitting it into two levels.

Consequently things are now looking a little less balanced and we have one too many levels:

(For full chart, please click on image)

We have found enough new mechanics that we can nearly introduce a new mechanic every level. By cutting the supermarket and moving the power station a bit earlier we can adjust the level order to create a better gameplay rhythm:

(For full chart, please click on image)

This can still be improved; we can look to either find a new survivor type that can be added to the town hall level, or we can try to replace it with something else that gives us more opportunities to do so.

Mood Map

There are potentially a host of emotions you will want the player to experience over the course of the game. The main character may experience things like unrequited love, revenge, sadness, and anger. These sorts of emotional events are important to track but they are not as important as the overall emotional tone or mood that you want the player to experience.

By "mood", I mean a basic emotional concept that can be passed to the audience. So panic, fear, trepidation, awe, and excitement would be considered moods, while higher order conceptual emotional themes such as revenge, jealousy, or nihilism would not be.

Generating the mood map has two purposes. It is used to assess that the level order and content will not interfere with the emotional journey of the player but more critically it is a fundamental tool for aligning the whole development team towards creating a holistic experience.

For instance, let's say that the story of Ken Kong will go like this:

Ken fights his way across the city saving the loved ones of his crush, but it takes him so long that by the end when he reaches her, she has been bitten and become a zombie herself.

If I define the mood map like this:

Kick-arse awesomeness - farcical chaos - mounting triumph - dark comedy

By defining the moods specifically over time you will guide the whole team more precisely than you might imagine. For instance "mounting triumph" implies a growing crescendo. It is likely to encourage a ratcheting up of music intensity, increasingly outrageous level end victory animations, and a general tendency to try to up the pacing each level.

While you probably assumed that the tone of KFZK would be defined as something like "zany", the act of stating it over time has a dramatic impact on the whole development.

For instance, if I instead define the mood map for the whole game like this:

Panic - horror - increasing trepidation - tragedy

Every aspect of the game will be completely changed by this mood map:

With exactly the same game design, these two mood maps would generate utterly different gaming experiences. When the whole team embraces the mood map and diligently tries to express it in all the assets and creative decisions they make, the mood will be successfully instilled into the player.

What normally happens, though, is that every team member has a slightly different idea of what mood or tone the game should be creating, and rarely any idea at all of what mood the player should be experiencing at any given point in the game. Is it any surprise that most games fail to move people, when the development team are all communicating slightly different messages?

The mood map can be as simple as the above four stage progressions, or it can be as detailed as putting several mood chunks into each level. It is worth bearing in mind that literally no story-based game has only one mood. Even horror games oscillate between building tension and outright terror.

Once you have the gameplay types laid out and the moods defined you can see how the current level plans fit together.

(For full chart, please click on image)

In our case we have puzzle levels late in the game that are clearly going to slow the pace where we want people to be experiencing "mounting triumph." By reordering levels, or shifting ideas from one level to another, we can better support the emotional goals:

(For full chart, please click on image)

Luckily KFZK's level order is very flexible, but most games are not. In most cases the answer is to give feedback to the individual level teams to try to reach the desired mood and gameplay mix.

While the above example is probably not the best order, or even the best mood map, the point of the exercise is to try to force yourself into examining the entirety of the plan so that feedback on each level is given relative to its place in the whole experience.

Block Mesh and Prototype

The next step is to start building the levels in 3D, and I argue that the best people to do that are artists, not designers, if you want believable and interesting spaces. Block mesh should validate whether the level as planned will fit into the technical and production limitations while demonstrating that they can be compelling enough spaces.

As these levels are prototyped, inevitably things will end up being slightly different than planned. Designers will adapt their plans based on the art, so throughout the block mesh and prototype phase, the leads have to continually update the game rhythm chart and validate the levels within the context of the mood map.

By continuing to extract new mechanics that arise from the block mesh phase and staying open to level re-ordering you can continue towards a balanced game plan without restricting the creative process of the level builders.

Final plan

All the information gained by building the block mesh should have refined the game design significantly.

Once all the levels are prototyped and one level has been polished to act as a vertical slice, production can begin from a very solid basis.