Geekly Articles each Day
📜 ⬆️ ⬇️

Creating a design of replayable stealth levels for Payday 2


Making Murky Station: Payday 2


Payday 2 is a cooperative first-person shooter with RPG elements for up to four people. The game is dedicated to bank robberies and theft of rare trophies. It was released on August 13, 2013, since then 50 DLC has appeared to it and new ones are still being produced. Payday 2 was constantly in the top ten most popular games on Steam. Today I would like to talk about my experience in creating stealth levels for Payday 2 before leaving Starbreeze in January 2018. Although the article discusses problems and solutions related to the design of Payday levels, I talk about them in a broader sense. This article will be useful for mid-level and junior level designers. If you are a senior developer, then something in this article may seem familiar to you.

To begin with, the Payday stealth mechanic is not ideal and fails in some areas, but I would like to focus on the decisions made when creating the design of the cards, especially the Murky Station card. I will also analyze how we used the RNG (randomization), and how it was used for the purposes and mechanics to maintain the level of freshness and replayability. The creation of this card involved two people for six weeks. My partner took on the role of a level builder / environmental artist, and I was a script designer / creator. The two of us chose the scale of the project based on the needs of our studio. The idea was to create a script of a small robbery for 10-15 minutes with a high proportion of replayability. We have much to learn, so let's get started!



Let's start from the beginning


Before creating sketches or building schemes, we decided on whether we want to create a “loud level” (only for battles), a “stealth level” (with avoidance of battles), or a map in a mixed style. Due to lack of time, we decided to stick with only stealth. Due to the fact that we made this choice in the early stages, we were able to create better options for the movement of players and focus on balancing the patrols and the location of targets. We decided that the theme of the level would be a small wagon depot, managed by a group of mercenaries selling heavy weapons. The main task is to penetrate the depot and steal the EMI bomb. Creating a simple and intuitive goal is important in multiplayer games, where players can enter and exit gameplay at any given time.
')
We decided to aim for 10-15 minutes of gameplay and break our main goal into several subtasks, each would take about two minutes (we were based on our good knowledge of the gameplay Payday 2). It is worth noting that this estimate of time should change after a player passes the level several times. These values ​​are usually reduced by a third, and in some cases double. Given our main goal, we could create a simple outline of how the robbery would take place, and we began to think about possible dynamic and RNG elements that can be used to create replayable gameplay.


(This is an example script from our editor; each entity has its own function)

Test ideas before writing a script? What is it like?


Since for 90% of Payday levels, scripts are written by hand, it is important not to waste time on creating non-working things. Testing targets and complex RNG elements should be fast and efficient. The last thing we want is to create a system entirely and understand that it does not work. Most of the time for testing our ideas, we didn’t need animations or even models. Therefore, in the early stages, we simply had a debug text. You may wonder - “what if I don’t have debugging text, or I don’t know how to write scripts?” When testing Payday 2, we often prepared a simple template, and then ... pretended to do the tasks.

It may seem insane (and not everyone was able to play it without laughter), but one of our designers played the role of an outstanding mission for Bane, who simply showed us goals. We were moving in space, pretending to see security or hacking laptops, taking into account the necessary time delays. In fact, you can simply break the system into separate parts and try new features. For example, knowing that you can have two escape places on each side of the card gives you enough information to pretend that you are making decisions. By telling fellow developers where the truck arrives and specifying a place for folding trophies, you can determine whether the place will be interesting to escape.

Let the artists giggle, and people passing by from other departments stop and are perplexed, not seeing crowds of enemies. It actually works: you can often direct the level in the right direction and avoid investing too much time in the wrong goals. Of course, I understand that this approach will not work for all studios or situations, but until you try, you will not know.



Creation of the scheme of our sandbox


After we pretended to run through goals and got used to our stocking, let's talk about the scheme we created for the Murky Station. We chose what I can call the “bulb method”. It is necessary to create multi-layered rings, giving a sense of movement towards the center (or towards the goal). In fact, we use the outer layer as the starting point of the player, and each subtask is located inside the next layer until the player reaches the main goal (in the so-called “center”). This approach is very useful when working on sandbox type levels, especially when the player can move to almost any place.

Note: also at the end of the subtask, we changed the layers of the music track, creating more suspense and a sense of involvement.

You can see that the outer layer of the onion is the player's spawn (shown in green) at the overpass, where you can fully inspect the depot. From here, the player can learn the patrol routes, the location of the cars, and possibly the target locations. In addition, the transition can be used by a sniper rifle player to mark guards in different corridors, helping to convey accurate information about guard positions to ground level players.


The next layer is the penetration into the depot through the fence around the perimeter. The fence here is needed in order to send the player and show him the visual boundaries of the "safe zone" (in which there are no patrolling guards). The next layer is a search of cars to find out where the main target is hidden, after which a breakthrough is made through the storage doors inside the trains themselves. These layers of the bulb must be properly managed to convey to the player the necessary sensations. If you make too many layers, you can confuse the player or he can forget what he does, if too little - the task will seem too simple and he will not get a sense of completeness.

The most important thing is player mobility!


Mobility is a key factor that allows players to express themselves and make optimal decisions when passing a level. I thought that in the case of the Murky Station it was very important to ensure the use of different styles of play - from slow and methodical to fast and dirty. I absolutely didn’t want to force players to play in a strictly defined manner and create predictable and linear routes. To do this, I spent the first week of developing the level for prototyping and testing various design ideas that could maximize the number of routes and choices.


Simplified top view of routes in the wagon depot area

It became obvious that we must necessarily allow players to pass through and under the trains, because they occupy most of the level space. Unfortunately, the old models and train schedules were not designed to pass under them, but thanks to the prevailing night level, all the details are hidden. Only the two of us worked on the project, so I quickly learned the basics of Maya and corrected the lower part of the trains, removing collisions and editing the models to improve readability.


The next difficulty was this - you need to teach the player that he can hide under the trains and be safe there. Until this point, Payday players had never had the opportunity to climb under trains, so we needed to draw attention to this and also show them that this was a safe place. By making these places dark and hidden, we were able to create the illusion of security, but also complicated the process of their search for players.

To solve this problem, we added yellow stripes and dim red lighting under the wheels, attracting the player's attention. The combination of these elements was then used in other parts of the level as a visual “dictionary”, allowing players to show what they need to explore.


Another way to add new routes at the level was to create a ventilation system in the lower tunnels. The level is designed for stealth, so we created more compact spaces. Ventilation paths allow players to safely watch patrols, search for targets, and move loot. For prototyping such moves, I built a ventilation system out of the modules, which made it possible to perform rapid construction and testing. It's quite funny that in the first iteration of the moves they turned out to be too small, so the bodies of the players passed through the floor. I managed to rework the modules and as a result we chose a height sufficient to move while standing, rather than just crawling. Here we again used the yellow strip indicating the entrance to the ventilation on the wall.

Modification of trains and moves - this is one of the factors. affected the success of the card; he gave new players confidence in exploring the depot and lower claustrophobic tunnels. So, having studied various ways of movement and giving players more options, we can proceed to the construction of a randomization system.




Randomization of targets for replayability


RNG is one of the fundamental foundations of Payday, so every decision we made was necessary to view through the prism of RNG. We were quite confident that randomization should be meaningful, and not just used for its own sake. It is important to ask the following questions: is it worth changing the location of each mug at the level? Did we get something by swapping all the cars and buildings? Was creating a third entry point valuable for the level? Perhaps one day we completely randomize every object in the building, down to the smallest circle, but in my opinion, in games like Payday, such aspects do not justify themselves and often destroy the planned design.

When working with RNG, it is important to ask yourself as many questions as possible in order to start with a solid foundation, especially if you want to finish the game in time. I often notice that juniors and media designers forget to decide on the scale and prioritize their goals. This may seem trivial, but if you forget about priorities, you can create a black hole that will suck all your time.

So how do we add RNG to the Murky Station? After analyzing our goals, we began to choose those RNG options that we could manage to implement within one month. In addition, I set my own priorities for them (from low to high).


I mainly focused on randomizing train configurations, storage doors, and key layout. These goals were critically influencing the way the player moved through the main space and how he would behave in the same zone during different passages. To achieve these goals, I smashed my subtasks into digestible points of interest and divided them into self-made blanks (see the figure below). Thanks to this, it was possible to write a script for one piece and teleport it to any number of points. This approach has increased the manageability of writing scripts randomization and reduced the number of bugs that would have arisen in the construction of all manually.

Note: we have given each of the key / vault stock a unique look and sound so that players can recognize them from afar or listen when they are close. It was very important to provide players with such a level of feedback so that they could choose the right decisions in the process of passing the level.




Having dealt with the storage doors and keys, it was possible to begin planning the process of placing them at the level. Before placing them you need to make sure that each place meets certain conditions. The main goal is to create complexity for the player and the motivation to be creative when working with the surrounding space. Due to the fact that I created a scheme with a lot of interesting bottlenecks and paths, it was quite obvious where to put them. Key collection is one of the most randomized goals in the Murky Station - sometimes all the keys are in different parts of the map, and sometimes all are close together. Later we cleaned up the script to avoid too strong points or terrible probabilities of the RNG, but in general this system became a huge level success.

Usually we placed the keys closer to the center of the map and tried not to put them too close to the players' safe zones. From time to time we have several keys in the backyard to change the pace of the game and give players a different type of difficulty, caused by the openness of the level.

This is how the upper depot looks like and how the keys are distributed. In the lower tunnels are located the same number of keys.


The same method we used to create the interiors of cars and storage doors. Having created one preset and a script by assigning them four times inside the level (one for each type of storage door), we were able to randomize the location of the main goal of the players. In addition, the engine allows us to rotate blanks, which makes it possible to swap interiors of cars. This added a whole new level to their configurations, because some of their interior designs were asymmetric.

As a result, we got about 600 railcar configurations, 2000 combinations of storage doors and 256 subtask configurations. In addition, with each passing, one or two exits were randomly selected, which radically changed the decisions made by players. It also influenced how they would move through the level and enjoy the benefits of a wide range of movement options.

On top of this, we used non-linear goals: in fact, players can fulfill several goals at the same time, and in some cases in a different order. At the Murky Station level, players can simultaneously search for keys, search cars, tag guards, look at them from above, and collect additional trophies. This allows four players to conveniently split up to cover more space and play independently of each other. In a well-coordinated team, two players can search for cars in an EMP bomb, while others look for keys to the vaults. I believe that it is very important to give each player the opportunity to contribute to a common goal.

Note: You may be wondering how, with all this randomization, the QA department could do the testing. The short answer: he did not have to. We ourselves had to effectively build a level to guarantee 90% of its reliability, and then catch as many boundary cases as possible. In the Payday development team, the vanguard of QA protection is a level-creating designer; thorough testing of his own work is his task! Due to the method of constructing the systems described above, in each example, it was enough to support only one blank. This gave us the opportunity to experiment when adjusting the level, because we knew how small the probability that it will affect our blanks. Therefore, thanks to clever leveling, we were able to reduce the number of aspects that need to be tested by QA, and accelerated development.

Having finished with the goals, let's now take a look at how the RNG can influence patrols and cameras.


Guard and RNG patrols


Randomization strongly influences how smooth / annoying the level will be. When developing stealth levels, we should have considered the irritation of the players from the poor location of the patrols, the number of guards and its stops at each point. The task was to create an interesting puzzle, not a terrible game about waiting. A bad RNG could force a player to sit in a corner for a minute, waiting for the guard to leave, only so that at the end of this minute another guard would take his place. Our goal was to eliminate such situations by adjusting the time, processing the level scheme, and possibly the level mechanic. Therefore, it is important from the outset to create a solid foundation for the options for moving players.

Since we did not want the RNG patrols to get out of control, we had to carefully consider their movements in space. This requires attentiveness and multiple iterations. Move the balance too far in one direction, and you will get empty areas without protection, move too far in another - there will be too much protection, and the possibility of maneuvering will disappear. The last thing we need is the probability of a chain reaction to death. It occurs when you kill one guard, after which the body immediately notices another, located ten meters away ... which forces you to kill this guard, who is immediately noticed by the next one, and so on. In Payday 2, players have a limit: they can kill no more than four guards, after which an alarm is triggered (at all levels). At the levels, we actively had to manage the number of routes crossing and the frequency of meeting guards.

In the first test run of the Murky Station, I got a good level coverage for the guards, but there was a drawback: in some parts, up to eight guards could randomly accumulate. After game testing and design changes, I decided to break the patrols into small cycles and add more points. This increased the coverage of the guards, and the patrols became more coordinated. It also reduced the maximum clusters to about four guards and significantly reduced the number of possible chain reactions to death.

Places of patrols in the first run:


Places of patrols in the second run (new routes provide the same level coverage and reduce the likelihood of congestions):


New look at the old mechanics


Most of our stealth levels use random static cameras that allow players to use their avoidance or sabotage skills. Players have a lot of mechanics to handle cameras, but when discussing options for the Murky Station we came across a brick wall. Because of the “corridor” style of the map and the surrounding structures, we have very few options for camera placement. Because of this, the cameras no longer changed the level in a positive direction. We also found their inconsistency with the design level, because the player must look for a specific car. If cameras will be sent to him. then the player will be able to find him too quickly, which will adversely affect the complexity of the search.


How do we deal with these problems? We could not abandon the cameras, so we began to brainstorm and search for resources that can be used. It was important to keep the main camera functionality intact and continue moving to the fundamental basis of the levels - randomization. We found a drone model created for one of the previous levels and began to prototype new ideas. The resulting design provided the coverage we needed, and also created new challenges for the players.



All trains can spawn up to two drones, each of which flies around the perimeter of the train and scans the space for the presence of bodies and players. At all levels, three or four drones are activated at random each time for scanning. The cycle takes about 30 seconds, after which they return to their trains and deactivate. The cycle continues in this manner every few minutes until the level is completed.

At high levels of difficulty, more drones spawn and become indestructible.

From the point of view of design, drones turned out to be a tremendous find, because they allow us to dynamically influence the gameplay and prevent players from feeling comfortable, using the same routes with each passing. Some players will avoid the corridors with drones, more experienced players shy away from them using the available movement options, and some players may even drive into a trap and force them to invent new routes.

(GIF accelerated about eight times; the complexity of Hardest allows you to show routes and move drones)


Thoughts in conclusion


It was very interesting to work on the level of the Murky Station, and even now I sometimes play on it. If you analyze the players' goals and their mutual influence in the space of cooperative transmission, you will begin to see a big picture and be able to understand how a well-planned level with controlled elements of the RNG can remain fresh and replayable. I find very interesting experiments with different types of RNG, especially when combined with level design. I hope my article gave you an idea of ​​how we build levels with the RNG and why we chose it as the main pillars of design.

Thanks for reading, here is my contact information:

Twitter: @generalvivi

Email: generalvivi [at] gmail. com

Website: www.generalvivi.com

In addition, I have a speedran (1 minute) level so you can rate it, as well as passing on Hardest (10 minutes) difficulty from one of the pro-community players.

Fastest passage, 2018 (carefully, loud sound)



Ten-minute video gameplay , showing the great variability of the robbery.

Source: https://habr.com/ru/post/359326/

More articles:


All Articles