Making Hitman 2's Best Level

One of the pleasures of stealth games is the large, open-ended level. A good stealth stage does not feel like a narrow ride with scenes arranged in order. It feels like a place that existed before the player arrived.

The player can watch from a high vantage point, choose a route, enter through a side door, change plans halfway through, and deal with the same problem in several different ways. That sense of place is what makes Camp Omega, the shifting mansions of Dishonored 2, and Deus Ex's Palisade Bank so memorable.

Hitman pushes that idea especially far. Its levels are fully explorable spaces filled with people, routines, props, locked doors, disguises, overheard clues, and opportunities. Hitman 2's Miami mission, The Finish Line, is a strong example of how that kind of stage is built.

The first useful question is not the layout. It is the fantasy. What would be interesting about sending Agent 47 to this place? In Miami, the answer is a huge coastal race event, complete with food stands, paddocks, VIP lounges, garages, a corporate office, a podium building, and a race constantly looping around the level.

Racing games usually put the player on the track. This mission does the opposite. It lets the player explore everything around the track: the public stands, the backstage areas, the corporate facility, the pit buildings, and the spaces that racing games usually treat as scenery.

That inversion gives the level immediate identity. The race is not just decoration. It creates noise, spectacle, geography, timing, target behavior, and a big physical barrier that the rest of the design has to solve.

Once the location is clear, the next major design problem is the targets. Miami has two: Sierra Knox, the driver, and Robert Knox, her father and the head of Kronstadt. They are not simply people to eliminate. They are moving centers of gravity for the whole mission.

Robert is mostly a dweller. He stays inside the Kronstadt building, moving through a tight loop that includes labs, balconies, scientists, and his office. His fortress is heavily restricted, so reaching him means learning the building's security structure.

Sierra is different. For much of the mission she is sealed inside a race car, circling the track for almost twenty minutes. Later, if the race state changes, she becomes a roamer who moves through more public VIP spaces. The two targets create different kinds of access, timing, and problem solving.

The recurring design question is simple: what could possibly go wrong here, and how can the player get control of it? Robert's routine is predictable, but the level is full of ways to disrupt it.

Sabotage the air conditioning and he may leave to use eyedrops. Turn off the satellite and he may walk to the roof. Break his prized car and he may come to inspect it. Find the right inputs for a military robot and the target can become part of a very different setup.

These assassinations are not isolated buttons. They usually require items, disguises, access to restricted rooms, and a chain of actions. That structure turns a kill into a plan rather than a command prompt.

Hitman often places an object, line of dialogue, or visible contraption in the world as a little promise. A military robot suggests that someone could be targeted. An explosive golf ball suggests future mischief. A trophy, a champagne glass, or a pyrotechnic rig quietly asks the player to imagine how it might be used.

The game can point toward those possibilities in several ways. The player might overhear a useful conversation at exactly the right time. They might wander into a room and see an interactable setup. They might notice a challenge name that hints at a possible outcome.

Those hints can feel slightly artificial, but they produce an important stealth-game emotion: the thrill of overhearing useful information and turning it into a private plan.

Miami's most interesting wrinkle is that Sierra does not win the race by default. If she comes second, she never goes to the podium building, which means several possible assassinations will not happen unless the player changes the state of the race.

That creates a point-and-click style puzzle inside the simulation. To put Sierra on the podium, the player has to help her win or make Moses Lee lose. The target schedule is not just something to wait for. It becomes something to manipulate.

This is a delicate balance. Long schedules make the world feel indifferent to the player, which is powerful. They also risk making the player wait. The level answers with shortcuts: alternate starting positions, ways to disqualify drivers, and actions that speed up the desired state.

Miami is divided by the track. On one side are the stands, food trucks, paddocks, medical tent, motel, and public event spaces. On the other are the multi-storey Kronstadt building, the podium building, and the marina.

A race track through the middle of a stealth level is an awkward obstacle if the player needs to move freely from one side to the other. The design turns that problem into navigation: overhead walkways, underground passages, a parking garage, and multiple crossing points spread across the map.

The split gives the level a clear large-scale shape. The crossings keep that shape from becoming frustrating. The player can understand the two halves quickly, then learn the faster routes over time.

A useful way to describe this kind of layout is a snail house with Swiss cheese. The snail house is the winding main path that lets a small footprint feel large. It is the route that takes the player naturally through the public spaces, major landmarks, and intended flow of the level.

The Swiss cheese is everything that punches holes through that path: fences to climb, windows to enter, shafts to scale, back doors, staff-only corridors, parking entrances, and tiny connections between larger areas.

Together, the two ideas create both readability and mastery. A new player can follow signs, colored floor lines, maps, and obvious routes. A returning player can move almost like they are teleporting because they know the holes in the structure.

A stealth level becomes more generous when most rooms have more than one exit. Dead ends can create panic, but too many of them make experimentation feel dangerous in a bad way. A player who slips into a restricted room should usually be able to improvise a way out.

Multiple exits also mean multiple entrances. The obvious way into the Kronstadt building is the front door, but the player can also approach through the parking area, the podium building, a walkway door, or another route discovered through exploration.

This structure makes the player feel less like they are following a scripted path and more like they are choosing angles of attack. The room is not just a destination. It is a node in a network.

Hitman is not mainly about hiding behind walls. It is about hiding in plain sight. That means the level's geography is also a social structure, with each costume changing which spaces are safe, suspicious, or forbidden.

In Miami, Agent 47 can walk through public stands and parts of the marina as a regular visitor. VIP areas need a badge. Security spaces need a guard uniform. The Kronstadt building escalates from public showroom to staff floors to guarded upper levels. Racing paddocks are split by team colors.

The order matters. If the best disguise appears too early, a large part of the level collapses. Good disguise design layers access so the player keeps discovering new permissions, new risks, and new routes.

A large open level still needs moment-to-moment friction. Hitman does this by scattering small stealth situations across the map: an engineer who needs distracting, a doctor blocking access, a guarded item, a generator that can pull someone away, or a uniform that can only be taken after a small setup.

These micro-problems are not the whole level. They are texture. They make the spaces feel authored without forcing the whole mission down one narrow solution path.

Because the level is open, these little puzzles can be solved, skipped, reversed, or combined with larger plans. The player is constantly moving between local tricks and global strategy.

After the major structure is in place, the level has to be played and adjusted repeatedly. Designers look for moments that are too easy, too hard, too slow, or too invisible.

If a target is alone for too long, the assassination may become trivial. If a sniper sightline is too obvious, the challenge becomes waiting instead of planning. If a suit-only route appears impossible, the designers may still leave it for players to discover, because expert players often find solutions the team did not expect.

That final tuning is what turns a dense system into a fair one. The level has to give players enough information to plan, enough friction to make the plan interesting, and enough freedom that mastery still feels personal.

The danger of formal level-design rules is that every stage can start to feel the same. The better approach is to treat these ideas as guidelines: clear locations, targets with routines, manipulable schedules, layered access, readable routes, shortcuts, few dead ends, and repeated iteration.

Those guidelines can flex from one mission to the next. One stage can be dark and claustrophobic. Another can be dense and public. Another can center on a fortress that limits safe movement. The chain of missions matters as much as any single map.

The broader lesson is not just for Hitman. Characters on schedules can make a world feel alive and give players a basis for planning. Snail-house layouts and Swiss-cheese shortcuts can make compact spaces feel large. Social access tiers can turn level design into a puzzle about identity. Miami works because all of those ideas support one thing: the player making a plan, breaking it, and finding a better one.