How sound was created in Pathfinder: Kingmaker


    Pathfinder: Kingmaker (hereinafter PF: K for convenience) is a computer role-playing game created by Owlcat Games and released this fall on Steam and GoG . The project, inspired by the classic games from Bioware and sharing with them such features as the use of the rules of the popular system of board games, real-time combat with a pause, a camera with isometric projection, non-linearity of passage and the ability to get some unique endings.


    In this article I will talk a little about how we worked on audio in the process of creating the game - on task planning, creative search and problem solving. An experienced specialist is unlikely to discover unique discoveries here, but for beginners and those interested in the topic there will be interesting details.


    The first steps


    The work on audio in the project begins with the search for key ideas, which then will form the basis for the production of all resources: music, sound effects, acting sound, the overall balance of all components. As a rule, at an early stage, these ideas are demonstrated with the help of other works that have something in common with our ideas.


    PF: K was supposed to be a project that, unlike the new wave role-playing games, could actually prove that he is the spiritual heir to the Baldur's Gate series . We had a time-tested role-playing system ( Pathfinder appeared as a partial reworking of AD & D version 3.5 rules), game designers put together the first chapter of the future project as a module for Neverwinter Nights 2, and Viktor Surkov, our art director, drew an excellent concept that showed how the future game will be perceived and felt. The whole team also played Pathfinder every week for a long time, delving not only into all sorts of rules, but also carefully listening to their own feelings - as in the process of throwing dice and moving chips, a sensation of an exciting adventure is created.



    Concept art of the future game.


    With this baggage, we discussed a couple of weeks from which games, movies and books you can get inspiration. As a result, the following theses were adopted:


    1. General mood


    Unlike the Pillars of Eternity and Divinity: Original Sin already released at that time , our project had to adhere to the classic high fantasy style, in which approximately equal attention is paid to seriousness and entertainment, without any hard bends in one direction or another. We were going to recreate the feeling of an old, already slightly forgotten tale, which we suddenly got from the attic and decided to read it on a winter evening.


    2. Musical accompaniment


    The basis for the musical atmosphere of the game was to be the work of Inon Zur , who already had extensive experience working on games of this kind (Throne of Baal addition to the game Baldur's Gate 2, Dragon Age series of games). The starting point for the creation of the first compositions was the music of Basil Poleduris from the film “Conan the Barbarian” , in which the traditional orchestra and ethnic instruments sound equally.


    Inon creatively rethought the visual and plot materials transmitted to him - he wrote several key musical themes for us, which were then used as reference for sound and mood when working with other composers (unfortunately, the budget of the game did not allow Zuru to transfer all the music).



    Inon Suhr talks about working on music for Pathfinder: Kingmaker.


    However, we allow ourselves to move a little away from the reference, where appropriate. For example, in the game there is a so-called “First World” - the sphere of being, where the Gods initially only tried their hand at creation, because of which everything that exists there is subject to constant transformation. Such a place required a unique approach to music, and this is not the only example. We decided that the rules should not restrict creative freedom, where appropriate.


    3. Principles of sound design


    Reference examples of sound design for PF: K were the games Diablo 3 and Dragon Age. The first project boasts expressive combat effects (blows of weapons, sounds of magic, voices of opponents) and, in general, an excellent balance of all the audio in the game. The second game helped us find the right approach to how residential communities should sound - first of all, our capital. We approved the following criteria:


    • Battles take up most of the gameplay, and they have to entertain: therefore, the sound of the battle also needs to be bright, rich, even a little bit tough, but without too much realism. Underlined voicing of metal, crunching of bones and splashing blood when finishing off the enemy.
    • The sound of magical effects should be bright and well “readable” in close combat, for which most sounds for magic have a narrow stereo field and have a bright impulse in their composition. It is considered acceptable to add to the natural effects of characteristic synthetic sounds and textures - this increases the interesting and visible effects.
    • The sound of natural locations should be unobtrusive, have good detail, do not contain frequently repeated patterns, and regularly hint at what may exist “behind the scenes” (for example, howling wolves in the distance in the night forest, rare gnash and groans in abandoned dungeons, growling and swearing trolls in their fortresses).

    4. Acting dubbing


    Our budget did not allow creating cinematic cut scenes for the dialogues with detailed facial animation (such as what Dragon Age was famous for). To compensate for the player the lack of facial expressions and non-verbal gestures, we decided that we would try to convey as much as possible with the help of the voice of the characters: each character has a certain bright character trait that is constantly emphasized through the actor's game in his remarks.


    For example, the dwarf Harrim is in constant melancholy, which in exceptional cases flares up to religious ecstasy; Jeital's elf comments exude contempt 24 hours a day; The half-hearted bard Linzi's frivolous chatter demonstrates her frivolity and inspiration.


    Such an approach is markedly different from what can be found in modern games - it is often called “theatrical”, but it is quite traditional for old games and cartoons that were released in 1970-1980, which also should help us in creating the right atmosphere.


    Let's now see how these ideas evolved further.


    Sound natural locations


    In our game there are about 200 locations of various types, and all of them require environmental sounds. We began by breaking all the places that the hero can visit into “settings” - typical combinations of natural and plot conditions. If several locations belong to the same setting, then they use the same typical set of sounds with some variations.


    The game features the following settings: forest, plains, hills, dungeons, ancient ruins, swamps and First World. Artists and narrative designers selected photos and art that most clearly conveyed their mood, made a preliminary list of locations and quests. Since our game is based on the desktop module, we did not experience a lack of material.



    Reference for setting plains.


    The sound team took these documents as a basis and began to think about how to complement and deepen the perception of the setting for its part. It was important to make sure that the sounds of the environment did not bother you for 50-80 hours of play. This required making them as “alive” as possible, which means trying to use as many random and diverse parts as possible.


    As a result, we developed a sound scheme for each location based on the following main components.


    • The overall sound of the setting: several (from three or more) sound layers, which are played in a uniform distribution over the stereo field. In these layers, there are usually common sounds for the entire setting that do not depend on the character's position in space: several varieties of light air movement, occasional gusts of wind, insects, distant birds and animals. For some locations, specific elements are added that occasionally sound somewhere outside the camera: shedding stones, rustling leaves, and the like.
    • Invisible localized sources of sound: individual birds and insects, amphibians, the creaking of trees, the howling wind.

    With the help of such sounds, we make the location more interesting to study - the player selects individual bright sounds in the general background and may try to trace where they come from. Figuratively speaking, we use them to create a sound landscape design: by singing a bird or chattering a grasshopper, you can decorate an unremarkable piece of the map or draw attention to some secret that is hidden there. Or, for example, if you take two copies of the same card, but on the same day jays will sing, woodpeckers will knock and cuckoos on different edges, and crows will croak everywhere on the other side, the sensations from them will be noticeably different.


    • Sounds that are specific only to a specific location: usually these are its possible inhabitants or hints of events that occur in it. This includes trolling sniffing, cursing and hissing of kobolds, tapping with hammers in a dungeon, strange and frightening sounds in ancient ruins. Usually this all sounds also somewhere far away and only hints at what we may encounter, and also does not have a clear localization.
    • Visible sound sources on the map: characters, animated birds and animals, rivers, bobbing boats, fires, signs.

    The general combination of these components allows you to create a fairly rich and varied sound design, which is not boring, even if the player spends several hours in any place.



    An example of a general sound location: a fishing village.


    How to revive the city


    We needed a slightly different approach to the capital and other settlements: unlike natural locations, here we had to create a feeling of constant life and activity, taking into account not only the time of day, but also the worldview of the main character.


    Worldview (English alignment) is one of the main elements of games based on D & D. It reflects the ethical and moral standards of the character on two axes: Good - Evil and Order - Chaos. Each axis also has neutral states, when a person does not clearly lean toward something and his decisions are most often individual in each specific situation.


    After the completion of the first chapter, our main character establishes his small state and builds the capital on the site of a dilapidated fort. The actions that he performs during the game affect his worldview, and this implies that he will establish rules in his state that correspond to his worldview. Depending on the rules and laws, both the appearance of the capital and the behavior of the people who live in it change. And I wanted to convey this in her sound.


    Naturally, there was no ready-made material anywhere that would help us solve this problem. Therefore, we agreed with ten actors (five men and five women) about the recording session, during which they portrayed for us the noise of a crowd of different nature - what the movies call “gur” or “valla”. Also, each actor recorded an individual set of voice reactions for game events. All these recordings served as a kind of designer, from which we then collected the sound atmosphere of the capital.


    Before starting work, it was necessary to decide on what language we will use to voice the crowd. Russian did not suit us (all the dialogues and remarks in the game were recorded only in English), and working with foreign actors would be too expensive for such a task. As a result, we asked our actors to choose a few syllables for themselves, with which they could invent words and phrases on the fly, giving them the necessary emotion. In general, it worked - players from different countries do not recognize the word, but feel their meaning.



    The list of topics we asked actors to improvise.


    So, the sound of the capital consists of the following elements:


    • Crowd: a constant background hubbub of people with a uniform distribution over the stereofield. He has three options: calm (neutral), lively positive (with a predominance of laughter, drunken shouts, good-natured intonations), lively negative (with a predominance of frightened cries, whispers, rare moans).
    • Conversations of the inhabitants: if you come closer to some houses, you can hear the muffled voices of their inhabitants (as a rule, married couples). In neutral states, they just discuss something, not breaking into harsh emotions, in positive states they often rejoice, laugh and argue, in negative states they swear, cry, get angry.
    • Shouts from afar: a set of random replicas that sound in different places of the capital and are selected under the worldview. Their main feature is that we will never see the people who pronounce them: their voices are tuned so that they can be heard only from a distance. This allowed us to save on animations and stage setting, but at the same time, we noticeably revive the overall sound “picture”.
    • Urban activity: this includes pets, tapping the blacksmith’s hammer, a horse-drawn carriage somewhere, and similar sounds. They do not have a clear localization, they just add a sense of the population of the place.
    • Sounds of nature: our capital begins with a small village, which then grows to a stone city. Nature does not retreat completely: we can still hear the voices of birds from the roofs of some houses, the chirping of crickets in the evening, and feel the breath of the wind.


    An example of the sound of the capital with different worldviews.


    Weather and seasons


    Approximately one and a half years after the start of development, we began to implement various weather conditions in the game, including rain and snow of different intensities. By this time, more than half of the locations were ready and configured. How to add weather effects to them now and avoid problems with what is already working?


    A small digression: all the sounds of the environment in the game exist in Unity's own scenes, which are loaded along with the location. Before we thought about the weather, there were two scenes for each location: day (it sounds morning and afternoon) and night (it sounds evening and night). Naturally, all of them were created with the expectation of fair weather.



    An example of the contents of an audio scene.


    It became obvious that we would not have time to make on time two more variants of each audio scene (taking into account rain and snow), and it would be more correct to play these effects "on top" of what is on the location at the time of loading. In this case, we need to adapt the individual elements of the already running ambient sounds so that nothing interferes with perceiving the effects of the weather.


    To solve this problem, we introduced the global states of Rain and Snow into the game, and also added a controller for the intensity of weather effects. With their help, we taught the game to turn on the sound of the weather on top of what was already running when loading the location, and dynamically change the volume of the other sounds of the surroundings location.



    A set of audio buses through which environment sounds are played.


    In the picture above, you can see that we play the sounds of insects and birds through our own bus called LocalLive, which, in turn, is divided into two more tires - for resident birds (AllSeasons) and migratory (Summer).


    During the winter, insects, toads and migratory birds are completely muffled, but we still hear those birds that do not fly away to warm lands for the winter (in our game these are mostly crows and owls). At other times of the year, when it starts to rain, all insects and birds smoothly “fall out” from the general sound picture and then come back when the weather is getting better. All this is controlled by the weather effects controller, which controls the volume of the above-mentioned audio buses.


    The controller also helps us manage thunder during a thunderstorm: when light rain falls, the sound of thunder lags about 2-3 seconds from a flash of lightning, and the more it rains, the faster and more powerful thunder. At maximum intensity, the delay between lightning and thunder disappears completely.



    Demonstration of weather effects.


    Dialogues and character cues


    Actor dubbing is one of the most expensive games in terms of game development: the cost per hour of a good actor usually starts at $ 200 and has no limit. This amount is often added to the cost of renting a studio and the work of the director of sound.


    For PF: K, over a million lines have been written, of which more than half are dialogues. To voice them all did not allow the budget, and after long disputes we stopped to record only 10% of the dialogues with the actors in the entire game, but at the same time to completely sound the standard reactions of the characters and their conversations around the fire - unlike the dialogues, we hear them Often.


    Standard reactions


    Standard reactions are pronounced by our characters when performing any action: almost everything that can be done in the game can provoke a replica of your hero or companion. First of all, they inform us that our order has been accepted or rejected, or they signal the dangerous state of the character. But at the same time, all the remarks are written taking into account the character of each character and mean to them no less than dialogues - ideally, having listened to several phrases, we can already get a more or less truthful idea of ​​the character. Our companions, in addition to this, have also comic remarks that can be achieved by long and persistent clicking on their portraits.


    An interesting detail: all the usual (sounding out of combat) replicas were also recorded in a deliberate whisper - this was required to support the mechanics of stealth in the game, which allows you to behave relatively quietly and imperceptibly when required.



    Replicas of characters in the usual and secretive modes.


    Talk by the fire


    Talking around a campfire, in fact, is a large set of miniature dialogues that help us learn companions from different angles and provide some additional knowledge about their past, habits, or attitudes towards recent events. If there is no one to talk to the party (the main character does not participate in conversations around the campfire), then the lonely companion can simply speak on some topic of concern to him with a couple of sentences.



    What will the characters decide this time?


    Magic effects system


    PF: K uses about a thousand magical effects. When we calculated how much time and resources work on them can take, we decided to optimize the process in advance. Since many magical effects are similar in action and visualization, it would be wise to use the re-created sounds for them, combining them “on the fly” right in the game: from one effect to take a fire flash, from the second - a small explosion, from the third - a bright tonal overtone. Such an approach would help to quickly deal with new magical effects that are not completely unique.


    To do this, we created a Unity component called “Magic Constructor” in which you can set an unlimited number of prepared sounds in advance and adjust the volume, pitch and delay for each of them to start playing.



    An example of a customized component with a magic constructor.


    After we decided on the mechanical part, we started working on a set of basic elements from which the final sound will be assembled. The magic effect is usually divided into three main layers:


    1. impulse : transmits power and speed of propagation;
    2. domain : element or school of magic - fire, water, air, transmutation, enchantment, etc .;
    3. decorating elements : low-frequency shocks, crystal modulations, modulated air gusts, and the like.

    It is not necessary to use all three layers, but in general, this separation gives some flexibility and significantly simplifies the work when new variations are added to the main schools of magic, where the only main difference is domains (for example, acid instead of fire).



    Constructor magic in action.


    Select important in battle


    Real-time combat with a pause is a rather peculiar mode, when up to 12-18 characters can participate in a battle at the same time, and even the death of one hero can lead to the victory of the enemy. To help the player to better navigate what is happening, we tried to make the battle as informative as possible from the point of sound: select the right, remove the extra.


    Using High Dynamic Range Audio


    Audio in PF: K works with the Audiokinetic Wwise audio engine - one of its main advantages is the High Dynamic Range Audio (HDR audio) system, which allows you to expand the dynamic range of audio in the game and indicate which groups of sounds are louder / more important than others.


    At each moment in time, Wwise analyzes all sounds started in the game and adapts the sound mix so that the most important events “move” the less important ones down when the sound engineer needs it. For example, if a grenade explodes next to you or a newly killed dragon falls, then you absolutely do not need to hear birds singing in the forest or your own steps at this time — the game mix is ​​“cleared” for important events and then restored to its normal state.


    We used this system quite carefully, trying to make the player notice her work only in critical cases. All sounds were distributed in groups with different virtual loudness, and for some time we checked different settings, looking for the most comfortable combination. As a result, we have built such a hierarchy of priorities of sounds in battle (from maximum to minimum):


    1. The magical effects of tremendous power or the voices of huge creatures.
    2. Critical replicas of characters (low health, loss of consciousness).
    3. The usual magic effects (flashes, explosions, transformations, and the like).
    4. The pronunciation of magic spells.
    5. Regular replicas of characters and enemies: reactions to the beginning of the battle, given orders, attacking screams, etc.
    6. The sounds of weapons: waving, shooting, causing damage.
    7. Other sounds related to characters: steps, armor, clothing.

    Thus, the player always has a chance to notice the magic effect or cry for help in the midst of the battle, have time to pause the game and issue the appropriate orders to the characters.


    The priority of the volume of the sounds of our weapons over others


    A rather modest detail that later turned out to be very useful. When the owners of melee weapons (swords, axes, daggers, maces, etc.) mainly participate in combat, the sounds of combat, for the most part, boil down to swing and strike.


    In order to at least somehow highlight our characters in the “dump”, we decided to lower the volume of the enemies' weapons by 3 decibels: the player hears his own attacks and hits better in battle and focuses on his success. Well, in those cases when we ourselves receive a sufficiently large loss, then the problems will be communicated to us by the voices of the characters and the game interface.


    Finishing off enemies


    Another small detail: each final blow, which sends the enemy to the forefathers, is played by an additional low-frequency “kick”, to which blood splashes are mixed in (or whatever plays the role of blood from the opponent) - the player receives a kind of sound reward in the fight.



    Demonstration of sound in combat.


    Work on music


    How much music is needed for a game designed for 60-80 hours of passage? A strange question - as much as possible! However, for us the actual figure turned out to be about an hour and a half of musical material. From this it followed that first of all we had to avoid the feeling of repeatability, when the game makes you listen to the same thing for hours, and after a while you just turn off the game music and put on your favorite playlist.


    How we solved this problem:


    • All music was divided into three categories: subject, research and combat themes. Story music plays during dialogues and important scenes, setting the desired emotional background; research topics are heard during the study of locations outside the battlefield; and combat, respectively, played during battles.
    • Research music takes the most time in the game, therefore, it plays a deliberately supporting role in the overall balance: these compositions are written in such a way as to not stand out more than required and organically intertwine with the natural sounds of the location. Story and martial music is allowed to be vivid and memorable: we hear these themes less often, and usually due to strong emotional experiences.
    • In the predominant part of the locations, research music plays with long pauses between tracks (40-50 seconds), which allows the player to better feel the location through two equal streams of information - music and a carefully crafted sound atmosphere, while the music does not have time to get bored even for a long time. .
    • Battle and storyline music always sound in a cycle until the end of the event, maintaining a constant voltage.

    Otherwise, the music in PF: K uses a rather traditional approach: we sacrificed interactivity in the name of clear structure and memorability. The only thing that changes from time to time in compositions is the introductory part for combat themes: as a rule, for each of them there are 2-3 variations.


    Figures and graphics


    In conclusion, some common numbers:


    Music


    • Six composers worked on the game: two staff members of the studio and four specialists on a contractual basis.
    • A total of 60 songs were written, 105 minutes of music.
    • The main volume of music was created using virtual instruments, some compositions partially use live vocals and wind instruments.
    • All compositions for the tavern were recorded predominantly live.

    Sound


    • Eight sound designers took part in the project, six of them were brought to work on a contractual basis. Home help was needed in creating sound effects for a huge list of monsters and magic effects.
    • The project uses more than 23 thousand sound files in the following categories:

    1. locations (sounds of the environment);
    2. VFX, magic effects;
    3. monsters, NPC;
    4. dialogue events;
    5. weapon;
    6. interface;
    7. interactive action;
    8. character reactions;
    9. the dialogues.

    Actor dubbing


    • More than 15 hours of dialogue were recorded, which, in general, took about 150 hours of work in the studio.
    • 45 people were brought in to work on the actor dubbing. Most of the actors collaborated with us with the help of two studios in New York, and with some we worked directly.


    Conclusion


    For two years, a lot of work has been done on PF: K audio. Did all this give a return? I believe that yes - most of the reviews call sound, music and acting sound one of the main advantages of the game, people are happy to note the small but pleasant sound details and share them on the forums. In my opinion, the following helped us in achieving this first:


    1. We started planning audio at the pre-production stage of the game, when all key ideas could be discussed and agreed with other team members in advance. During this period, it was decided what the game sound will ultimately be and how important it will play, the basic aesthetic principles are set. We knew from the very beginning that we wanted to receive, and we respectively built the whole process from the beginning to the release.
    2. We licensed to play a third-party audio engine with rich functionality (Audiokinetic Wwise). This allowed us to quite easily integrate and maintain sound in a project without the frequent need to distract programmers. For example, elements of gaming audio, such as the weather system, required only a few lines of code calling the necessary procedures — all the rest of the work was done right in the sound engine.
    3. Sound designers had the opportunity to work with the game directly and study it "from the inside", which allowed them to offer their own ideas for improving the sound in the game with reference to its various mechanics. Such work is difficult to outsource, as it requires a deep knowledge of the project and operational interaction with the team.

    That's all. I hope that in the process of reading you found something new or useful for yourself, or at least you did not get bored. I, in turn, will be glad to hear your comments or advice.


    About the author:


    Composer and sound engineer, working on games since 2004. I put my hand to several dozens of projects of various kinds (casual and mid-core games for PC, mobile games, MMORPG). In my free time I do recording and synthesizing material for my own commercial sound libraries, and I also write music that is released under the pseudonym Silent Owl . Now he is the sound engineer of the Owlcat Games studio as part of the Mail.Ru Group.


    Also popular now: