Greetings, A lot has happened in the last few weeks, if you did not catch our [url=https://store.steampowered.com/news/app/1454960/view/4128182933494568950?l=english]last developer update[/url] we recommend checking that out. In that update, we announced a new name, our publisher partner, and our upcoming demo and participation in Steam's Deckbuilder's Fest on March 25th! For this month, we wanted to focus on Paul-Drauz Brown, the Sound Designer for Dark Sky, and give him a chance to share some of his insights and process on building out the audio for the game. He covers capturing the essence of the characters, layering multiple sounds to form ambience, and more! Take it away Paul! [h3]Developer Focus[/h3] Hey everyone I'm Paul, the Audio Designer here at Ganymede. When approaching the audio design for this project, I felt it was important to reflect the hand drawn and cartoonish aspects of the characters and creatures that inhabit the world, without falling into the trap of making ‘zany’ sounds which would undermine the story. To move in this direction, the bulk of the audio design is intended to feel grounded, but with each element having a little extra flavor that stops it feeling mundane and instead a bit more fun. As the project has evolved, more emphasis has been placed on creating punchy transients within sounds, again to enhance the bold, slightly cartoonish art style. Exaggeration of certain sounds was key in this; for instance, some carnivorous plants in the environment, rather than sounding like pretty much other plants, instead give off squelching and gurgling sounds I recorded while feeling hungry. This still feels in-world, but is a bit more interesting than the standard rustling you’d expect. That’s just one example, but this approach is layered throughout most of the sounds in different ways. With many assets to create and a game that has kept evolving, maintaining flexibility when it comes to implementing sounds was key. On Dark Sky, using the audio middleware tool, FMOD, has been very helpful in this regard. It allows me to layer up and randomize sounds in many ways without requiring additional assistance from our programmers. It does have some limitations when it comes to things like controlling music states and more complex, layered systems. However, it has been of great help on this project. An area where it shines is in designing ambiences. I made heavy use of scatter containers in FMOD, which allow you to trigger spatialised one-shot sounds around the listener object, in this case the player. Most ambient beds were created in this way, with a small selection of base loops and then layers of these scatter containers to add randomized details and background sounds. In the shipyard for instance, the sounds of ships flying offscreen and crashing were placed in such a container with a modulated pan applied, so they would randomly streak left, right, back and forwards before exploding [Fig. 1]. A concern with using these is keeping track of the voice count, which can climb very quickly, so I made sure to keep these at acceptable levels using voice limiting and stealing. [img]{STEAM_CLAN_IMAGE}/41495898/ef5c5261474bdbd061b051f5bab87cc60e0fa74b.png[/img] [Fig. 1] An example of the ambience layers and scatter containers. At the bottom is an LFO controlling the movement of ships as they fly past The second component of the ambiences are the emitters in the level itself, which are dotted around to add more specific details. For instance, in Bloody Tears Gulch, there is a derelict ship which is being hacked to pieces from the inside by the Runshaka. This emitter is actually a set of scatter containers which play various mechanical hits, whirs and groans, creating a fully randomized sound of busy workers hidden inside. As mentioned, these containers and ambient emitters in general can spike the voice counts easily, so every level emitter also controls its play state based on the player’s overlap. I set the radius according to the sound’s attenuation distance, so it can turn on and off during gameplay while retaining the smooth volume fade as you approach. For simpler, looping emitters I would either create variations, or set random start points and pitching in each event, meaning that even if the same source sound is playing for two level emitters, they will sound quite different. [previewyoutube=tBF13tzkKX0;full][/previewyoutube] For oddly shaped and long environmental features, there is a custom emitter which follows the player on a spline, such as along the banks of a river. In Bloody Tears Gulch, there are three of these river sections which blend into one another, creating a transition from the fast flowing part of the river, down to the slow trickle and acidic quality it has as it reaches the edge of the crystallized, toxic Dead Barrens. This provides the best of both worlds; a more consistent volume along the banks, and fewer emitters in the level. To enhance the sense of a space, we make use of convolution reverbs, which produce reverb based on the recording of a real space, and can create very realistic results. These are expensive to run, as they require a lot of continual processing. To offset this cost, however, almost all of these reverbs are placed on separate reverb buses in FMOD, so other tracks can be sent to them. In order to avoid requiring a lot of micromanagement, all spatialized sounds have a reverb send set inside their attention settings preset. Those then all send to a single reverb master track, which in turn sends to the various types of convolution reverb bus. The level of these sends is controlled using ‘Snapshots’ in FMOD, which can be triggered from inside unity upon entering a new ambience zone or collision shape. Using an inbuilt parameter for the distance to the sound source, means that the degree of reverb can be increased as the distance increases too, this is handled using a custom curve for each source’s attenuation setting, so behavior can be customized for special cases. In and outside of combat, there are many characters and creatures to encounter on Wolf Prime, each with its own visual and sonic identity. When approaching the audio for these, I based my aesthetic decisions heavily on which natural element or feature stood out to me the most in the visual design and VFX. This helps distinguish them from one another more easily, while also opening up creative avenues for exploration. [previewyoutube=dQJL2MK9x0o;full][/previewyoutube] For instance, the sound for the Arocknid is made entirely out of earth elements like stone, gravel and sand. Even the vocals are made by squeaking stones together and processing them to sound like small, shrill vocalizations. Other creatures, like the Runshaka, are associated most with the idea of metal scrap, given their scavenger lifestyle. This comes through in both the audio design for their tools and contraptions, which are rickety and unstable, their power sources which buzz and fizz, and their vocals which are made with a broken kazoo. Thinking along these lines also made it easier to categorize different sounds that characters could share, which helps a lot with keeping the scope manageable. Each attacking ability has a ‘cast’ and ‘hit’ component in separate FMOD events, sometimes with an additional projectile traveling sound. Each hit sound for every attack takes a parameter for the body type of the hit enemy (Flesh/metal/feathers/etc), then plays this material layer under the hit sound, providing more feedback and variation without requiring additional asset creation or exploding the feature scope [Fig. 2 & 3]. This same technique was used for when characters move on the grid, where the appropriate ground type is layered into their unique movement event. [img]{STEAM_CLAN_IMAGE}/41495898/8d81ea8c169d11cba073593d00c33e50f667c347.png[/img] [Fig. 2] An example of the character flesh type layered with an attack hit. [img]{STEAM_CLAN_IMAGE}/41495898/f95df5cae3db5bae5dda94e8c72046cab2bf42d9.png[/img] [Fig. 3] An example of one of the shared flesh hits, in this case 'Heavy Slash'. The Music System makes use of several nested events in order to play continuous music. At the bottom level there is a single event for each location, with internal transition markers and regions to control the moment to moment gameplay, whether we are in combat or not, what is the current threat level, etc [Fig. 4]. These are controlled by parameters and events set from other events or from the game engine itself. These are used to determine which music tracks should play, and which layers should be added or removed. These individual music events are then wrapped up in a main music event, which handles the transitions between levels and the main menu, using an additional set of parameters [Fig. 5]. Upon launching the game, we play the master audio event, and then control its state throughout the course of the game for a more seamless experience. [img]{STEAM_CLAN_IMAGE}/41495898/9b8ff706aa460e564614bf14defa898079df2cf8.png[/img] [Fig. 4] An example of the music setup for one level and its possible gameplay states [img]{STEAM_CLAN_IMAGE}/41495898/d405952b9296e266152ccc57de6d7d3d6750f348.png[/img] [Fig. 5] A look at the master music event, which controls all referenced music sub-events. There are many more features and aspects of the design which could be discussed, but I hope this gives some insight into the overall processes and concepts applied. Coming soon will be new settings and creatures which start to detach from the material world, and I’m excited to find new ways to bring them to life while retaining a sense of physicality. [h3]Conclusion[/h3] Thanks Paul for that look into the audio design, we hope this has been an informative look into everything that goes into designing and creating the audio that you'll hear more of on March 25th with the release of our newest demo, and later this year upon our full release. Be sure to check out the new Demo and the Steam Deckbuilder's Fest on March 25th! Thank you all for your continued support and enjoy the brief look at Wolf Prime! - Ganymede Games & Midwest Games Teams