Chapter 4.1. Video Game Music Analytical Methods
The purpose of analysis is to identify how the system that governs the development of music functions as the player progresses through gameplay. I seek to critically examine the conflict between interactivity and narrative-driven musical structure by developing my own analytical model and methodology. Understanding this conflict informs my compositional practices when designing DMSs, but may also be more broadly beneficial as a new way to view DMS design and function. Chapter 4.1 reviews existing methods and models developed in the field for analyzing game music and outlines my own.
In Chapter 4.2 I conduct three case studies. Each case study reveals a different approach to DMS design. The case studies include DMSs from “Tomb Raider: Definitive Edition” (Square Enix, 2013), “Dark Souls 3” (From Software, 2016), and “Halo: Combat Evolved” (Bungie, 2001). Each game features a different type of gameplay and aesthetic, but they all feature a narrative progression. My analysis seeks to form a deeper understanding about how the conflict between interactivity and structure effects music’s ability to communicate the development of a story in congruence with player actions. The experience of a story and its narrative can be defined by a perceptible progression of characters and events that aim to communicate meaning to the viewer:
“Narrative is a basic cognitive structure (Schank & Abelson, 1995; Wyer, 2004; see also Mar, 2004), and can be formally defined as “a representation of connected events and characters that has an identifiable structure, is bounded in space and time, and contains implicit or explicit messages about the topic being addressed” (Kreuter et al., 2007, p. 222)””. (from Nabi & Green, 2013; citations as is in quote).
Immersion is integral to a game’s design, which attempts to create an empathetical bond between the player and the gameworld. Through this bond the player is emotionally responsive to the game and can impact player behavior (Mendonça & Mustaro, 2012). This can also be described as the experience of transportation. “Transportation is an integrative melding of attention, imagery, and emotion, focused on story events” (Nabi & Green, 2013). This means that to assess the experience of a story there is the subjective element of describing one’s own sense of immersion. That is to say, one may state that because a DMS is designed in a particular way, it may contribute to their sense of immersion.
DMSs are designed to function alongside the indeterminacy of player behavior. The assessment of the music’s ability to communicate progression requires an assessment of the DMS’s linear musical output. Sloboda states that the development of elements that define musical structure (in some western music) like harmony, melody, rhythm, phrasing, texture, and dynamics are key contributors to music’s ability to tell a story and evoke a listener’s emotional response (Sloboda, 1991). I argue that when the game music forfeits smooth structure for the sake of interactivity at the cost of disjunction (see Medina-Gray, 2019), depending on the style of music, it may lose its potential to communicate narrative progression. This does not mean music is always perceived as being of a lesser quality when it is interactive. But to cite the passively bored tone of Service’s comment about the repetitive boss music in “Zelda: Breath of the Wild” (Nintendo, 2017), “…I wonder how many thousands of times I’ll have to hear this battle music before I finally vanquish him?” (Service, 2018).
4.1.1. Analysis Methodology
4.1.1.1. - Existing Method for Accessing and Recording Data
Summers’ chapter about methods from his book “Understanding Video Game Music” offers the most conclusive foundation for approaching video game music analysis (Summers, 2016, 36-50). Summers discusses different sources from which the analyst may uncover information for understanding how video game music functions. Summers’ method introduces how an analyst must engage with games as a critical player, interacting with the game as both a player and researcher (Summers, 2016, 34). Summers describes a specific experience from his process of analytical play, in which he notices short looping cues being triggered depending on the player’s journey. Summers replays the game to isolate which actions result in specific musical changes. He then tests the same actions on a different platform to be sure he is not being exposed to platform-specific occurrences. Summers states that analytical play allows for general observations, but “playing the game can often do little to reveal the specifics of the music programming in very complex systems. In these situations, it can be helpful to investigate ‘behind the scenes’ in the game’s programming” (Summers, 2016, 36).
Summers suggests that an in-depth understanding of complex systems is potentially not feasible through analytical play (Summers, 2016, 36). I believe it is not only possible to dissect complex DMSs through gameplay alone, but also necessary, as it is not possible to investigate the programming of many games. My third case study is an analysis of the Pillar of Autumn level from “Halo: Combat Evolved” (Bungie, 2001), which contains fifty-five individually implemented music segments. Summers has analyzed the same segment of gameplay as I did in my case study. Where he states that it is important to observe source files and code to reveal a complex system (such as the one from Halo), I have illustrated an arguably more detailed picture of how the DMS is designed (compared to his model) without initially referencing the game’s source files or code. If the analyst is unable to access the source data, or if doing so violated the game’s terms of use contract, then the analyst would have no choice but to analyze the DMS through gameplay alone.
The scope of Summers’ method is broader than my own research interests. Summers introduces several approaches for using the data to understand different properties of game music: “mapping motivic relationships and thematic development”, “harmonic analysis”, “topic analysis, “semiotics” and “intertextuality”, “psychological effects”, “hermeneutics”, “form analysis”, “ethnomusicological study”, and “performance” (Summers, 2016 39-41). In comparison, my method is much more focused on illustrating how the technical design of the DMS allows for music composition to develop structurally in congruence with the gameplay narrative. This can be seen from a comparison of our two analytical models of the Pillar of Autumn level from Halo. My representation of the DMS looks more like a blueprint for how each of the music segments are programmed into the game with very little description about the music itself. Summers’ model simplifies the illustration of individual musical segments and how they are implemented, and instead has large amounts of text that describes the musical material in much greater detail. My model can be seen below in the third case study. Summers’ model can be found here:
Download/view Summers' Halo:CE analysis
4.1.1.2. - My Method for Accessing and Recording Data
To understand how video game music works the game must be played. This requires hardware to play the game and some degree of ‘gaming skill’ to get through the game. If one was interested in analyzing the DMS used in a competitive multi-player game then the analyst must be able to conduct the study while other players are actively competing against them. If one wanted to analyze a DMS from the game “Dark Souls 3” (FromSoftware, 2016), a game infamous for its difficulty, the player must survive long enough to investigate the DMS. If one wanted to analyze the DMS in the final scene of “Red Dead Redemption 2” (Rockstar, 2018), then the analyst would first need to progress through sixty hours of gameplay.
The analyst must investigate all possibilities of musical change. This will require the player to repeatedly play small segments of the game attempting every possible gameplay action in every sequence. From doing so they must observe the triggered musical material and figure out how it is organized into stems. The analyst must be goal oriented with each gameplay action by attempting to observe specific music behaviors.
Interviews, articles, and presentations are sources for understanding how a specific DMS could be designed. But these types of resources aren’t available for the majority of games. Non-disclosure agreements often limit what is allowed to be said about how games are made. Data-mining or accessing source material within the game offers objective proof of how a DMS is constructed, but is often a violation of a game’s terms of use. Therefore it is essential that the primary method of analysis is centered on gameplay.
4.1.2. Represeting DMSs
4.1.2.1. - Review of Existing DMS Representations
DMS representations have been created by ludomusicologists to visualize DMSs. Below are some existing models that have been developed to understand game music. Reviewing how other researchers visually communicate DMS functions has influenced the design of my own model.
Brame’s representation of game scores plot connections between different parts of the game that have different musical segments associated with them. His model describes how the music associated with each module can interact with its connections (Brame, 2009). Brame’s model is useful for understanding different pathways that the player can navigate and what music will accompany their play. Brame’s model shows what types of musical changes are connected across the larger branching structure of the gameplay. Each module that Brame illustrates is described by an object in the game that correlates with, or causes, the musical change. The model does not attempt to show what happens within each of the modules where any number of music segments could be interacting with the player’s actions in interesting ways. Being that this is what I’m most interested in, Brame’s model is not sufficient for what I want to discover.
Figure 1. Brame's model of how various game scores are connected (Brame, 2009, 55).
Medina-Gray’s model illustrates individual loops, layers, and segments, and how they could potentially be assembled to create different possible musical outcomes. Medina-Gray’s model requires the reader to reference a toolkit of symbols used to describe all the different components of the DMS. Her model portrays the DMS as a non-linear, modular, three-dimensional construct (Medina-Gray, 2014), which I feel makes it difficult to conceptualize the potential linear musical structures that the player could experience. Medina-Gray’s model includes a system for understanding the smoothness of transitions as well. Her general treatment of DMSs is inspired by how analysts have approached understanding contemporary modular art-music (Cage, Brown, Feldman) (Medina-Gray, 2016, 66). Medina-Gray’s model is very comprehensive for illustrating all aspects of musical properties and possibilities. However, I find Medina-Gray’s models to contain a surplus of information than I require.
Figure 2. Medina-Gray's model shows how individual stems are connected together to form a DMS (Medina-Gray, 2016, 66).
Karen Collins has created an illustration of how the DMS in the game Russian Squares develops over a span of gameplay. The illustration shows a notated score that contains a numbered list of short musical phrases, which are then layered together as different blocks. Each of the blocks are labeled with a number associated with the notated musical phrases (Collins, 2008, 154). The analysis resembles what a performance of a modular art piece such as “In C” (Terry Riley, 1964) would look like. Collins’ model illustrates a singular outcome from a gameplay session of Russian Squares, which is helpful for understanding the outputted musical structure. The advantage of Collins’ model is that it very clearly illustrates the musical structural output of the player’s gameplay, and allows the analyst to suppose what may have been different if the player behaved different. The disadvantage is that it does not include any technical information about how the DMS was constructed.
Figure 3. Collins' model shows how the music from a playthrough of Russian Squares is assembled (Collins, 2008, 154).
Tim Summers’ model resembles a storyboard containing screen shots from the game (Summers, 2016, 25). The storyboard contains three sections: one section is for describing the trigger that causes music to change, the second section is for showing the musical structure, and the third section is for describing the music. The advantage of Tim Summers’ model is that it allows for there to be paragraphs of text to articulate what is happening musically and what the player is doing. This is useful to the broader field of ludomuscology, as it allows for analysts to discuss various ludomusicological topics within the model. The simplicity of the model makes it easy to understand how the DMS functions. For the analysis of “Halo: CE” (Bungie, 2001) Summers simplifies the ‘musical structure’ illustration of the DMS by reducing the organization of random sequence containers used for horizontal resequencing as ‘fragment sets’. Summers’ description of the music in each section discusses the semiotics and intent of the musical material. The disadvantages of Summers’ model, compared to Medina-Gray’s or my own, is that it does not describe the technical construction of the DMS (how the music is segmented and programmed).
4.1.2.2. - Constructing my own DMS analytical model
The primary goal of my DMS model is to show how a system is constructed. The visual representation of the system must show how the music is segmented into individual audio files, what those audio files sound like, how long they are, and how they are programmed to move from one to another depending on player actions. From this type of representation the analyst can focus on playing the game, or having someone else play the game, self-reporting their experience of the music, and then pin-pointing the causes of their experiences within the model. I refer to my model as the DMS flowchart representation model (DMSFRM). My model attempts to show all potential pathways of musical flow; meaning how the music is segmented and organized, what compositional techniques are used, a description of what the music sounds like, and the gameplay parameters that cause musical change.
The flowchart representation model that I developed reads left to right. Each music segment is represented as its own node, or module. While video game music arguably contradicts a linear depiction, I felt it was the most natural way to gain an understanding of how different modules could potentially flow from one to the next to form a linear structure. A left-to-right format also allows me to stack modules vertically. Stacked modules represent how different modules are treated as vertical layers. This duality (between horizontal and vertical) creates a side-scrolling branching tree-shaped diagram.
The modules are connected by lines or arrows that show the direction of flow. The lines and arrows contain text for describing the gameplay parameters that trigger the movement from one module to the next. Since the programming logic that governs how music triggers work are built from if/then/else statements, I felt it was appropriate to use that vocabulary. Below is an illustration of the construct I’ve described.
The case studies in this chapter refer to each segment as a numbered ‘cue’. The label should identify which branch of the tree the module belongs to.
For example:
A1, A2, A3; each “A” module would be part of the same linear pathway, being small segments that make up “track A”.
T1, T2, T3; each “T” module would be a transitional module used to transition between different tracks (or branches). Since they are transitional “T” modules they do not combine with each other to create a continuous musical set (like the “A” modules).
B1, B2, B3; each “B” module would represent a second linear pathway, being small segments that make up “track B”.
I have connected vertical layers with a dotted line.
Each module should contain a description of the musical material so that a reader can identify what is happening musically without having to reference the gameplay. Understanding how long each module is, what the music sounds like, and how each module connects, helps clarify what the DMS could potentially be doing depending on how the game progresses. I have included this type of descriptive text within each module.
As I used the model I found that there were additional components of DMSs that I wanted to include information about. For example, if the player pauses the game and all the music is passed through a low-pass filter. Or when the music uses horizontal resequencing, does the music transition immediately by crossfading, on the next beat, on the next bar, at the end of a specific module, or at the end of a predefined musical phrase?
For effects like what happens with the player pauses the game I use a free-standing ‘auxiliary’ module that describes what the effect is and when it happens. To describe the timing of transitions more precisely I have used a descriptive terminology common to music production. Quantized-to-beat means that the transition happens on the next beat. Quantized-to-bar means transitions happen at the beginning of the next bar. Quantized-to-segment means the transition begins when the current module ends. This terminology describes the transitional process between modules.
By looking at the model a reader can understand the premise of the gameplay scenario, how the DMS is constructed, how the music changes from player actions, and what the music might sound like.
I developed some methods for helping to discern different aspects of the DMS to create the DMSFRMs. When critically playing the games for my case studies, I used an external capture card to record the gameplay from my console onto my PC. I recorded all of the gameplay segments. The actions in each of the gameplay segments were either exploratory (looking to observe where music changes), or observational (looking to understand what triggers musical change). The video recordings were later referenced in order to analyze the musical material or DMS behavior. The video recordings are available to view in the “Misc” folder.
To help hear the music more clearly I adjusted the volume settings in each of the case studies. I have raised the volume of the music, and lowered the volume of the sound effects in the game’s menu. The observations and conclusions from each recorded gameplay video was logged on the data tables. Often times, when an observation was made, I would continue to play the segment to make sure that my observations were consistent. From the video recordings and the information logged on the data tables, I began to construct a visualization (the DMSFRM). After forming the DMSFRM, I would replay the gameplay scenario several more times to make sure the visualization was consistently representative of how the DMS functioned regardless of different types of gameplay behaviors.
After creating the DMSFRM I looked for other resources to check the accuracy of my models with. Resources included interviews, documentation about the DMS design, or data-mined source files. I could not find additional resources related to my first two case studies. However, with the third case study of Halo, I was able to find the game’s source audio files organized into folders. By looking at how the assets were organized into folders I was able to see if my analysis of the music’s organization and implementation was correct.
4.1.2.3. - Summary of Analytical Method and Model
My analytical model is a tool for understanding how a DMS is designed and functions, which I call the DMS flowchart representation model (DMSFRM). Collecting the data for the model requires the player to play the game while attempting to perform every possible action in every possible sequence in order to uncover how the different segments of music are organized and implemented into the game. The model I have developed for illustrating this data prioritizes the visualization of vertical layering techniques and the branching horizontal resequencing possibilities. It also includes text that describes the music, and connecting lines that represent triggers.
By understanding how a DMS functions and visualizing its potential structural outputs, one can note their own experiences from playing the game and identify which parts of the music system were, in their opinion, effective. The judgment of effectiveness is purely subjective. As a game composer it is a useful exercise for informing my own creative process. Below are three case studies that analyze different gameplay scenarios using my methods and model.
__________________________________________________
Chapter 4.2. Analytical Case Studies
4.2.1. CASE STUDY 1: TOMB RAIDER: DEFINITIVE EDITION (XBOX ONE VERSION)
4.2.1.1. - Preface
“Tomb Raider: Definitive Edition” (Crystal Dynamics, 2013) is an action adventure game. The game is a story about Lara Croft, who crashes onto an island where she becomes stranded. The story of the game starts with her being held captive by what seem to be island natives. After she escapes captivity she investigates the mysterious island while trying to find a way off. The gameplay requires the player to solve puzzles and combat enemies in order to progress. As the player discovers new areas, participates in combat, solves puzzles, and progresses, the narrative develops. Below is a recording of the beginning sequence that I analyzed for this case study.
This full gameplay session is an important first impression for attempting to understand the gameplay (audio, visuals, and mechanics). Below is the data table containing observations made during a more experimental way of playing.
4.2.1.2. - Data
Download/view Tomb Raider analysis data table
For this case study most of my observations describe how different actions trigger different musical stems. I refer to an action as a single isolated gameplay test to discover something specific about how the DMS works. As can be seen in the data table the DMS is mostly constructed with stingers that respond to player actions. On the data table additional rows were used to make multiple observations about a single action. The additional rows are not labeled as new actions. Every individual action has been video recorded. The folder containing all videos can be found in the folder labeled "Misc".
From the data that I collected above I have constructed a DMS flowchart representation model. The DMS flowchart representation model is below.
4.2.1.3. - DMS Flowchart Representation Model
Download/view Tomb Raider analysis DMSFRM
4.2.1.4. - Assessment of DMS
The music from the beginning of the video is a sustained metallic pad that accompanies a cut-scene. This is shown in the DMSFRM as Cue 1. Just before the gameplay begins there is a sudden percussive impact with a number of bowed cymbals and gongs. As the player swings Lara back and forth, Cue 2 begins looping. The player must swing Lara into the fire in order to burn the rope that is suspending her upside down. As Lara falls Cue 2 fades out. At the moment Lara hits the ground she is impaled by a stick. Cue 3 is triggered just as Lara is impaled. The sudden percussive impact of Cue 3 is followed by a high pitched metallic ringing, which seems to communicate the agony and disorientation that Lara is experiencing. As the player is prompted to pull the arrow out by pressing “x” on the controller there is a sudden metallic scraping sound with string harmonic effects (Cue 4). Cue 4 very abruptly cuts off the metallic ringing of Cue 3, breaking the continuation of music and revealing a seam between segments. A solution that I think would help hide the seam would be to start fading out the music of Cue 3 with a long fade time when Cue 4 is triggered, and then emphasizing the transition with a bigger initial percussive strike at the start of Cue 4. This would allow the music to feel more continuous. These first four cues are primarily textural with no pulse and features a dissonant harmony. The opening passage defines the musical language and sets expectations of the musical development.
As Cue 4 fades out Lara stumbles deeper into the cave. Cue 5 is triggered just as a strung-up body comes into view. The orchestration of the material is similar to the bowed metallic textures that precede it. The next musical cue in the video is Cue 6, which occurs when the player picks up the torch. The music here is a stinger with another sudden impact. The continual use of sudden impacts for every action makes me wonder why picking up a torch warrants such drama, and what the music is trying to accomplish by being this way. A pattern is revealed about the overall design of the DMS: when the player does something, play a stinger.
In the video, Cue 6 comes after Cue 5, however, from my observations (in the data table), I noticed that the player is able to trigger Cue 7 directly after Cue 5 (before Cue 6). Cue 7 is a loop that contains airy flute-like instruments alternating between two pitches, and is accompanied by a soft electronic pad. Cue 7 is triggered when the player approaches a pile of debris blocking Lara’s path. As the player progresses through a small opening in the rocks Cue 8 is triggered, which resolves Cue 7 with a downward glissando. The player is required to relight their torch, which triggers a low brass and woodwind crescendo (Cue 9). Cue 7 and Cue 8 seem to act as a way of breaking up the consistent use of stingers every time the player does something. However, when Cue 9 is triggered, it again gives unnecessary drama to the act of lighting a torch. Perhaps this is to remind the player they are in an unsafe environment, but I feel silence could be more effective. Silence would also allow any real dramatic moments (such as Cues 11.1 and 11.2 coming up) to have a much greater impact.
After lighting the torch, the player must light a bundled roll of fabric that ignites a barrel of explosives. As the player lights the fabric Cue 9 fades out and Cue 10 plays. Cue 10 references the earlier high metallic instrumentation.
As Lara progresses forward through a small hole an enemy suddenly and grabs her legs. This triggers Cues 11.1 and 11.2. The two cues play simultaneously, with one or the other fading out depending on whether or not the player successfully escapes the enemy’s grasp. The jump scare stinger seems appropriate, but I wonder if the previously unnecessary use of dramatic stingers has lessened the effect. Cue 11.1 and 11.2 introduce the first clear metered rhythmic pattern.
After the player escapes, the music fades out. As the player continues forward they enter a flooded tunnel, triggering Cue 12 to play. Cue 12 is nearly two minutes long. The music starts with soft airy flutes alternating between two pitches, then becomes more tense as a high metallic sustain and an electronic pad fade in. Cue 12 ends while the player is solving a puzzle.
After solving the puzzle, a crescendo (Cue 13) leads into Cue 14, which contains a driving rhythmic percussion pattern and low brass glissandi. The loop continues until Lara falls down a hole triggering Cue 15 to interrupt Cue 14 mid-phrase, before leading into Cue 16. The interruption of the rhythmic phrase in Cue 14 reveals a seam in how the DMS is designed, distracting from the musical experience. Cue 16 is a development of the previous driving rhythmic pattern in Cue 14. As the player progresses forward they must successfully make three consecutive jumps. Each jump triggers a short muted brass swell. Because of the way brass was introduced previously in Cue 14, then removed in Cue 16, and added once more each time the player performs an action, the illusion is created that the player’s actions are responsible for the continued development of the brass motif. This DMS design is effective at supporting the intensity of the gameplay while also allowing the music to develop a structure in congruence with player actions.
As the player progresses, Lara must dive under a boulder where she is grabbed by another enemy. This triggers Cue 20, an eight second segment with a half-time rhythmic feel. In the gameplay I failed to escape within the eight seconds so I had to reload from the most previous checkpoint. As Lara gets crushed by the boulder the music stops. On the second attempt, after successfully escaping, Cue 21 is triggered, which plays a new rhythmic figure and brings back the high metallic bowed cymbals. Cues 16, 20, and 21 are problematic in how they abruptly interrupt the current musical phrases, revealing seams in the DMS design.
Following Cue 21, Cue 23 is triggered to begin playing as Lara jumps to the final area. In my observations I noticed that there was an additional cue, Cue 22, which will begin looping if Cue 21 finishes before the player triggers Cue 23. Cue 23 leads directly into Cue 24, which is a loop that features a low ascending brass melody over a driving drum pattern. As the player climbs up the slope towards the end of the level, rocks come tumbling down towards them. Each time the player dodges a rock a different stinger is triggered. When the player finally escapes the cave by reaching the top of the slope the music fades out.
Many of the stingers throughout the DMS feel unnecessarily dramatic. While there is a gradual increase in the use of rhythmic drum patterns throughout the gameplay, the sequencing of various textural themes do not create the illusion of a continual structural development. The reuse of various textures seems relatively arbitrary to me.
4.2.2. CASE STUDY 2 - DARK SOULS 3 (XBOX ONE VERSION)
4.2.2.1. - Preface
“Dark Souls 3” (FromSoftware, 2016) is an action role-playing game where the player plays as a character called The Ashen One. The player must navigate through a hostile world seeking to collect the souls of specific enemies by defeating them in combat (we’ll refer to these fights as boss fights). By collecting all the required souls, The Ashen One is able to keep the world from spiraling into darkness. The Dark Souls trilogy almost exclusively uses music to accompany boss fights. Each boss throughout the trilogy has its own unique music that attempts to portray something about its character. For example, the music from the boss fight against an antagonist called ‘Aldrich, Devourer of Gods’, contains many themes and motifs from previous Dark Souls boss fights. These themes are woven together to form a new composition. Aldrich’s backstory is very dark and disturbing and the music hints to the player which of the previous Dark Souls bosses that Aldrich has devoured, suggesting that they are somehow still inside of him. Music in the Dark Souls games is used as a story telling tool and a way to evoke player emotions. Boss fights often have multiple phases and the music adapts alongside the progression of the fight. This creates a type of adaptive musical experience that is very different from the Tomb Raider analysis.
I have chosen to analyze the music from an encounter with the boss called the Abyss Watchers. Because the music is an integral part of communicating the game’s lore it is important to understand who the Abyss Watchers are before assessing the effectiveness of the music. Empowered by magic, the Abyss Watchers were a group of undead warriors once responsible for keeping the world from being taken over by the darkness (the abyss). They were later betrayed, leading the group to enter an eternal cycle of killing each other. The story of the Abyss Watchers describes a fall from a heroic and noble post into madness, agony, betrayal, and eternal suffering. The music is used to communicate this aspect of who they are.
The Abyss Watchers fight is split into two different parts. The first part involves the player fighting a primary Abyss Watcher, while other watchers rise from the dead to partake in a free-for-all melee. After the player defeats the primary Abyss Watcher a cut-scene begins, where it rises from the dead and becomes imbued by flame. This starts the second phase. In the second phase no other watchers rise from the grave to join in, however, the primary Abyss Watcher is more difficult to defeat.
The fight has many mechanics that raise questions about how the music could potentially be adaptive. The first is a question about whether or not music changes when additional Abyss Watchers rise from the dead in the first phase. When additional watchers join in the difficulty for the player goes up and a part of the Abyss Watcher’s story is revealed to the player, as previously described. The next question is about how the music changes when the player enters the second phase of the fight.
Boss fights in the Dark Souls trilogy are notoriously difficult and can take several tries to succeed. The difficulty means that there is a great sense of excitement or anxiety when the player is close to victory or when the player is close to failure. Given the emotional reaction to nearing victory or failure, I wondered if the music was adaptive to player or boss health percentages. Below is a video of the full gameplay. Note that I die twice (on accident), and each time I have to run back to the boss room to try again. This creates a type of repetition that I think is worth considering given that failure isn’t simply an expectation, but rather a primary mechanic of the game.
4.2.2.2. - Data
Download/view Dark Souls 3 analysis data table
Play-testing the fight revealed that the scenario was only comprised of two linear looping tracks (one for each phase of the fight), and one transition. The difficulties in collecting data for this gameplay scenario were, 1) trying to make sure no other layers were being used when different mechanics occurred, 2) trying to understand how the two linear tracks transitioned from one to the other, and 3) trying to stay alive long enough to conclusively make observations about how the music behaved under different circumstances. To make this analysis possible I had to recruit help from my fellow game audio friend, Ricky Westray, to team up with me, and help keep the Abyss Watchers from killing me while I made observations about what was happening. He can be seen running around the background in the observation videos. The folder containing all videos can be found in the folder labeled "Misc".
4.2.2.3. - DMS Flowchart Representation Model
Download/view Dark Souls 3 analysis DMSFRM
From the data I was able to determine that the music from the first phase was split up into eight different musical phrases. The music transitions to Cue 3 via Cue 2 when the player enters the second phase of the boss fight. The transitions are quantized to occur when the currently playing segment finishes.
As opposed to the Tomb Raider DMS, which attempts to immediately respond to a variety of different player inputs with short stingers, the Dark Souls 3 DMS smoothly adapts to the second phase of the boss fight by using horizontal resequencing techniques and longer ‘linear’ sounding compositions.
4.2.2.4. - Assessment of DMS
The music for the Abyss Watchers analysis begins when the player enters the boss fight arena. The music is of a Romantic style with a full symphonic orchestra and choir. Compared to the other two DMSs, the Abyss Watchers DMS is much simpler. As previously mentioned, music is often used to explain the lore of different characters. For the Abyss Watchers, their story is a tragic tale of being locked into an eternal civil war with their own kin. Through the gameplay there are many opportunities for the music to adapt to events, such as when the Abyss Watchers are resurrected, killed, or locked in combat with each other. Instead though, the DMS is comprised of two linear compositions and a single transition. The first linear composition plays during the first phase of the boss fight, and the second is transitioned to at the beginning of the second phase, which is marked by a cutscene.
All of the music is melancholic and represents the bleakness of the Abyss Watchers’ story. In the second phase, when only one remains, the music becomes much darker and less energetic, potentially representing the loneliness of the last Abyss Watcher’s final moments. Through my data collection process I was able to identify that the first track was split into eight segments, which I refer to as phrases. At the end of each segment the music will either continue to the next phrase, or horizontally resequence the transitional segment (Cue 2). Cue 2 is able to be seamlessly horizontally resequenced after each of the phrases. This is possible through clever composition, as it means there does not need to be a different transition for each exit point of Cue 1. This is an idea that I have used in part of my own DMS prototype in Chapter 5.
In the gameplay video it took me three attempts to defeat the Abyss Watchers. Dark Souls is known for being a difficult game where players often get stuck on boss fights, having to replay them countless times. From my experience with the franchise, after repeatedly losing, the music can become tiresome. One criticism I had of the DMS was that the music starts exactly the same way every time the player enters the arena. I wondered if it would be possible to start the music with an intro that is unique to the number of times you’ve failed the boss. Since it is likely you will repeatedly run into the arena and quickly die, the beginning of the music is what becomes most repetitive. This is another idea I explored in my Chapter 5 prototype.
The music from each phase will loop indefinitely until the player loses or progresses. After defeating the boss the music fades out. In general it is difficult to criticize the adaptive capacity of the DMS, as it does not attempt to adapt to the narrative created through the player’s actions. Rather, the music seems to describe the story of the Abyss Watchers’ past.
4.2.3. CASE STUDY 3 - HALO: COMBAT EVOLVED (XBOX ONE VERSION)
4.2.3.1. - Preface
“Halo: Combat Evolved” (Bungie, 2001) is a science fiction first-person shooter game. The game centers around a war between humans and an alien race called the Covenant. The player plays as a super soldier referred to as Master Chief, who ends up on a ring shaped artificial planet created by the Covenant. Throughout the story the player discovers that the planet is a weapon called Halo, and the Covenant on Halo are being invaded by a parasitic species called the Flood. In order to prevent Halo from being weaponized, and the Flood from escaping, Master Chief must destroy it.
Music is used throughout the game to represent various characters, settings, and emotional states (combat, horror, sadness, victory). As Summers describes in his analysis (Summers, 2016, 25), the composer Marty O’Donnell has used different musical themes to represent the contrast between the Covenant and the human race. More exotic and electronic instruments are used to represent the Covenant, and more symphonic instruments are used for humans.
The game is structured as ten levels. Each level connects to the next by a cut-scene. Within each level are a number of checkpoints that the player can load their game from. Each of the ten levels can be started from the main menu. Each level has an objective driven mission that reveals part of the narrative.
The level that I have chosen to analyze is called the Pillar of Autumn. In this level, the player wakes up on the spaceship called the Pillar of Autumn as the Covenant are boarding it. The player must retrieve an AI called Cortana from the ship’s captain and escape before the Covenant discover it. Throughout the mission the player is given a tutorial, is introduced to both the humans and the Covenant, experiences combat, is given the opportunity to try and sneak past some enemies, explores dark maintenance tunnels with hidden enemies, and is continually reminded to escape before the ship explodes (even though it is not actually a timed mission). Each experience the player has is accompanied by a different type of music. The structure of this level is relatively linear.
4.2.3.2. - Data
Download/view Halo:CE analysis data table
The Pillar of Autumn’s music system was much more complex than the other two case studies. As I played through the game I found that there were a number of times when the repetitive music seemed to be constantly varying. One of the biggest challenges was figuring out how these sections of music were broken up into smaller segments within the seamless continuation of music. My solution was to record multiple videos of these sections of gameplay, play them back, and use a shorthand notation to quickly transcribe different rhythmic and melodic patterns that I heard. By cross referencing the transcriptions from each of the videos, I began to notice the segmentation of short consistent musical patterns. At some parts of the game where it was difficult to distinguish where segments began and ended, I found it possible to render the music separate from the videos and import the audio files into a DAW, where I could analyze the audio visually to see where patterns began and ended. After my analysis I checked the game’s source files to see if I had segmented the music accurately. Out of the fifty-five individual segments used, I had segmented two of them incorrectly. I have since gone back to fix the mistakes. The mistakes were inconsequential and did not effect the accuracy of the DMSFRM. The folder containing all videos can be found in the folder labeled "Misc".
4.2.3.3. - DMS Flowchart Representation Model
Download/view Halo:CE analysis DMSFRM
From the data that I collected I was able to determine that the music from the gameplay scenario was segmented into fifty-five individual assets. Some of the music exists as linear tracks that accompany cutscenes, but the majority of the segments are used throughout different sections of the game to be randomly horizontally resequenced. In the model I refer to these sections as random sequence containers. I show the segments that are able to be horizontally resequenced in random sequences as smaller modules that are attached to the random sequence container. The term random sequence container comes from the term used in the audio middleware, Wwise, to achieve this type of technique. Each of the segments within the random sequence container are seamlessly connected until the player either activates the next set of segments (in a new random sequence container) to begin, or the music ends. When the music is triggered to end there is often a specific musical segment with a cadence that plays. The alternation between random sequence containers and linear cut-scene tracks make up the entirety of the DMS in the Pillar of Autumn level.
4.2.3.4. - Assessment of DMS
The music for the Pillar of Autumn level begins during a cut-scene. Long melodic string orchestra passages are introduced over a bed of minor chords as Captain Keyes questions how they unsuccessfully escaped the pursuing Covenant fleet. The music throughout the cut-scene is linear and in a Romantic style. The music ends as the scene changes from the ship’s bridge to the hanger deck where troops are mobilizing for combat. The scene-change triggers a new linear track to begin, which I’ve labeled as Cue 2 in the model. The music introduces an orchestral percussion section playing march-like patterns on the snare drum, timpani, crash cymbals, and bass drum. The repetitive drum phrase ends abruptly as the marine, Sargent Johnson, gives a speech to his troops. The music begins again after the scene shows the troops shipping out, adding a staccato string pattern with a sustained melodic development (Cue 3) similar to the music from Cue 1. Cue 3 abruptly fades out mid-phrase as the scene changes to the Master Chief being woken from cryosleep. As the master chief wakes up a male choir sings a short sequence of chords with an ascending melody (Cue 4). While this entire sequence is a series of cut-scenes, Cues 2, 3 and 4, don’t seem to be written specifically for the durations of the scenes. In the context of the cut-scene, Cue 2 sounds like an introduction, Cue 3 adds strings, winds, and vocals to develop the music of Cue 2, and then it is suddenly faded out just as it starts to develop a melody. A few moments later Cue 4 plays and seems musically unrelated to the previous cues. My criticism here isn’t about dynamic music. However, the way the music unfolds throughout this section suggests how music is treated throughout the game, accepting that the nature of game music inevitably includes sequences of unrelated segments and abrupt musical transitions.
The dynamic music begins with Cue 5 when the player runs into an exploding doorway. This triggers what I refer to in the model as the Drum Sequence. The Drum Sequence is a repetitive drum pattern that chains together cues 6 through 10 in a random sequence. This prevents exact repetition from occurring. The technique is effective for creative the illusion of non-repetitive music. When experimenting with how the system works, I stood around for five minutes listening to the Drum Sequence. When not progressing, the illusion loses its effect as the music eventually starts to feel repetitive. This criticism can be applied to all parts of this DMS.
The next musical trigger occurs when the player reaches a door that slides open revealing a Covenant enemy. This triggers a stinger to play that emphasizes the element of surprise. Drum Sequence (Part 2) is simultaneously triggered. Drum Sequence (Part 2) includes an additional layer of music that features an exotic improvisational vocal solo and a choir. In the gameplay video the player never hears the exotic vocal solo. This is likely due to the drawback of horizontal resequencing requiring the current segment to finish before moving to the next. This drawback paired with the way a random sequence container works may have meant that when the music finally transitioned, it transitioned to a segment in the Drum Sequence (Part 2) container that did not feature the exotic vocals. Before one of the vocal segments was triggered the player had already progressed to the next musical trigger. This does not create any structural problems with how the music develops, because if the music never introduces the exotic vocal motif, it is never missed. However, Summers’ analysis (Summers, 2016, 25) mentions that the vocal line is likely meant to emphasize the exotic and unknown nature of the Covenant race. If so, then the DMS design used in this section has failed to serve its intent. A simple solution would be to force Cue 16 to be the first segment played when Drum Sequence (Part 2) begins.
The next musical trigger occurs when the player enters the bridge, which is a linear track that accompanies a cut-scene. Following the cut-scene, Cue 19 introduces a sustained bass pedal. The bass pedal sustains through the player’s first experience with combat where they must help a group of marines eliminate a Covenant threat in the cafeteria. The bass pedal fades out during the combat sequence. The bass pedal adds tension to accompany this initial encounter, however, I would question the creative direction of allowing the bass pedal to end at a seemingly random point through the encounter.
As the player proceeds, Cue 20 and the Ominous Strings Sequence are triggered simultaneously. Cue 20 serves as a bass pedal that various segments from the Ominous Strings Sequence fade in and out from. In the gameplay video the player only hears a few seconds of this musical material before progressing to the next musical trigger. When experimenting with this section of the DMS I was able to identify three different musical phrases that are randomly sequenced. Interestingly, the next trigger is a very short distance from the one that starts the Ominous Strings Sequence. If the player is to hear all of the music in the Ominous Strings Sequence they must not progress forward. I think this type of ‘hidden’ music that is only heard if you choose to not progress is a cool concept, and I use it extensively in my Chapter 5 prototype.
The next trigger occurs as the player turns the corner and sees the Covenant boarding the Pillar of Autumn. Cue 24 triggers the Ominous String Sequence to fade out and the Boarding Sequence to begin. This is an example that shows how the DMS is capable of guaranteeing that a specific segment leads the introduction of a new random sequence container, which is the approach I suggested for the introduction of the Drum Sequence (Part 2) where the exotic vocal line was missed. The music from the Boarding Sequence introduces many new musical elements such as synthesized plucking sounds and a different drum pattern. The section compliments the action of continuous combat throughout this section of gameplay.
The next random sequence container begins when the player progresses down a corridor. The Boarding Sequence (Part 2) segments add a layer of brass to the existing synthesized plucks and drums. The addition of the brass allows the music to have a sense of development as the player progresses. The Boarding Sequence (Part 2) concludes with a climactic cadence in Cue 37. The combat continues without music. This may be due to combat having become a constant state of play at this point, no longer requiring the extra dramatization provided by the music.
The next section of music is introduced as the player enters the ship’s ‘scary’ maintenance tunnels. The dark tunnels are a new setting for the player to experience. The segments used within this sequence contain short string passages that alternate between extended technique tropes (like harmonic glissandi and col legno textures), and eerie melodic phrases. The music continues as the player sneaks around, eventually engaging in combat. The sequence ends when the player enters the control room overlooking the cryo chambers from the beginning of the level. The Maintenance Tunnel Sequence feels like a standalone musical experience that prevents the over-use of continuous rhythmic patterns.
The final section of music begins when the player triggers Cue 46, which leads into the Final Sequence. As the player progresses to the next checkpoint, Final Sequence (Part 2) begins. The segments contain a mix of sharp angular string runs and contrapuntal melodic fragments. The segments in this section are poorly made compared to previous ones. The beginning and ending of each segment is audible, and when they are sequenced there is a perceptible disjunct transition. When the player reaches the end of the level Cue 50 is triggered, which resolves the Final Sequence (Part 2) sequence. The final music is Cue 55, which accompanies a cut-scene leading into the next level.
__________________________________________________
From the case studies I have uncovered how each of the DMSs are designed. I have observed how different techniques contribute to the development of music in congruence with player actions. I feel some of the ways techniques are used are successful at developing the musical structure in congruence with gameplay, but other times, throughout the DMSs, the music seems to develop in an arbitrary way with little attention to the structure of the larger musical experience. These subjective claims are not meant to be definitive evaluations. However, the analyses provide a window through which I am able to identify which parts of the DMS design are responsible for the positive and negative musical experiences I had. These observations inform the design of my own DMS prototype in Chapter 5.
__________________________________________________
Previous: Chapter 3: Contemporary Techniques
Next: Chapter 5: Prototype and Closing Remarks
Contents: Back to Table of Contents