Wednesday, 3 October 2012

Non Repetitive Design - Why it matters


Non-Repetitive Design – Why it matters

One of the main aspects that game developers set out to achieve when creating games is that the experience of playing it remains immersive throughout. The idea of immersion is that you can play the game without being distracted towards something that seems unnatural in its context. This is achieved not only through the game itself but also through the audio. With the audio, we ideally want realism and spontaneity in its design, much like the ‘real world’. Having minimal repetition with the audio in a game can help to indulge the gamer into complete immersion and fulfil their expectations, without interruption.

RAM budget and streaming allowance

Something that can be considered responsible for repetition within games, is the streaming allowance and RAM available for the audio content. As a sound designer in games, keeping to the correct memory usage whenever possible and ensuring you are saving space whenever possible can help to keep the size of the audio to a minimum.  A couple of ways of achieving this are as follows:
  • Using audio compression to decrease the size of a file. (UDK has an in built compression available, named ‘Vorbis’.)
  • Editing out silence in a sample. Silence takes up just as much space as audible sound does, so by carefully editing it out of samples you can save on file size.
  • Being meticulous with Sample Rates and down-sampling files when necessary. A concept known as the ‘Nyquist Shannon Sampling Theorem’ states that you only need to sample a sound at twice its frequency in order to accurately reproduce it. For example, if the frequency content of a sound peaks at 22 khz then you would need to sample it at 44khz in order to ensure fully audibility of the sound. These method of memory saving can give promising results.
Non-repetitive design considerations

Once the above rules are adhered to, you can begin to look at ways in which you can create an unpredictable and non-repetitive soundscape for a game. The following are a couple of ideas that can be considered:
  • You can use pitch shift to determine the weight or size of an object instead of using multiple samples. In most cases, a higher pitched sample can be used for a smaller object and vice versa. In terms of footsteps you can vary the pitch dependent on the surface or size of the character.
  • Re-modelling a sound using pitch shift to create something else fit for another source. For example pitching a drum hit down to create a low frequency rumble.
  • Pitch up/Play down – Pitch it up in your DAW and play it at half the speed in UDK to save half the size.
  • Non Standard Randomization – Sound playback system that has a bias against recently triggered sounds.
In UDK, you can randomise sounds using [AmbientSoundNonLoop] or SoundCues. You can apply slight pitch, filter and volume variations to create many outcomes. Furthermore, if you deconstruct a sound of its various qualities, you can then split these qualities into individual categories, and this then allows more control over both sounds and they can be randomised and played back simultaneously. This significantly increases the combinations of playback and therefore decreases the repetition in the sound heard. An example of this could be proposed with the fundamental sounds within an explosion ; you have the sound of the initial explosion, followed by falling debris (and some other stuff). Consider that there are multiple samples of the explosion and multiple samples for the falling debris, each branched off into different randomised nodes. Using a concatenator the sounds can then be mixed and matched creating a numerous array of combinations to significantly decrease repetition.

Middleware can be used in this nature aswell, to randomly playback sounds to create different combinations. This FMOD tutoral video demonstrates on a basic scale how you can get sounds to playback differently each time to prevent repetition, in this case an explosion debris sample.


References:
-  R. Stevens, D Raybould (2011). The Game Audio tutorial. Oxford: ELSEVIER


Monday, 17 September 2012

Fallout: New Vegas - The Mojave Soundscape

CRIMES Function - Orientation

One aspect that strikes me about the soundtrack for Fallout: New Vegas, is that the music, sound design and overall ambience in the game all blend seamlessly in between one another to create a unique and fully immersive environment for the player. By working in unison so effectively, the resultant audio helps to express the desolate and post-apocalyptic qualities that the Mojave Wasteland sets out to exhibit whilst reacting to the landscape as accurately as it can.

Through further research into the implementation of the audio in the game, it became apparent that the audio team had thought hard about the music before the game was produced. They had devised a music system, which consisted of 5 'types' of music, as a way to aid the interactive outcome of the game. These five types of music are categorised as 'scripted', 'incidental', 'location', 'battle' and 'hostile'. One of the types in which I found of great interest is the 'location music', which operates in relation to where the player is situated on the map and thier distance from an art object within the game. There are usually three layers of music radiating outwards from an art object and these layers help to blend the music into the virtual atmosphere. See below.



This process within a game relates to the idea of 'Orientation', where music is used as a navigational device to direct you towards points of interest (in this case a town). 

An example of this system can be heard in 'Goodsprings', the first town you encounter. Arranged music can be heard in its full entity when situated inside the heart of the town, this can be referred to as Layer 1. As you move the character away from the town and into the Wasteland, you hear the music becoming more sparse in its consistency and less audible via seamless transitions through Layers 2 and 3. However, If the player is deep inside the Mojave Wasteland and isolated from civilisation, 'incidental music' can be heard instead, which is basically short segments of the music played randomly. The incidental music integrates with the ambience of the wasteland, blurring the boundaries between the music and sound design, whilst creating an unpredictable and ambiguous atmosphere to suit the game itself. This approach to adaptive/interactive music aids the natural flow of the game and toys with the player's subconscious, subtly informing them of locations and other things nearby. Using music sparsely also keeps the soundtrack fresh and prevents it from becoming too repetitive. 



The following youtube clip demonstrates the layers idea in detail:






A detailed article by Scott Lawlor (Audio Director for Bethesda) which looks closely into the implementation of the musical systems within the game can be found here.