50 Cent: Blood on the Sand: Audio PostmortemBy Rob Bridgett
[In this detailed postmortem, Bridgett discusses what went right and wrong during the creation of audio for the Swordfish-developed, hyper-stylized hip hop action game 50 Cent: Blood On The Sand.]
50 Cent: Blood on the Sand is a third-person action sequel to the commercially successful 50 Cent: Bulletproof. It was developed with the Unreal engine using a combination of third-party and proprietary audio technology.
Blood on the Sand features voice-overs from Curtis "50 Cent" Jackson himself as well as G-Unit members Tony Yayo, Lloyd Banks, and DJ Whookid, as well as a host of exclusive music tracks and hits, an original in-game hip-hop score written by producer Swizz Beatz. It also required a huge amount of over-the-top cinematic SFX work and a two-week off-site post production mix.
Joining the Team
My work as sound director on the project began in June 2006 when I relocated from Radical in Vancouver to Birmingham in the UK for six months to work with developer Swordfish Studios (then a Vivendi Games Studio).
A vertical slice featuring 50 Cent as the central character, with Tony Yayo as his AI-driven sidekick, had already been produced at Swordfish and greenlit by Vivendi executives, and so development got underway -- at which point I joined the project.
I was joining an already existing audio team, consisting of an audio coder, Justin Caldicott, and Audio Lead and sound designer, Mark Willott. One of the advantages of joining a team immediately after pre-production was that they had already proven a great deal of the technology for the sound of weapons and ambience in the game.
There was also an advanced physics sound system as well as extensive systems for placing sounds in animations (such as Foley and footsteps) using a variety of Unreal engine visual scripting techniques via Matinee and Kismet. All this made it very easy to hit the ground running after quickly learning the tools and pipelines.
Justin had built his own user interface application on top of the basic FMOD audio engine, called Noisemaker, which was our primary pipeline to get sound into the game. With this tech and pipeline in place, the team were in an ideal situation in which to begin discussing and designing two big-ticket areas that had not yet been considered -- namely, music and dialogue.
These were, in terms of content and implementation, to be the touchstones of a 50 Cent IP game. It was clear that the division of labor between sound designer Mark and myself needed to focus on our areas of expertise and experience, Mark continuing the responsibility and ownership of the in-game sound effects and myself taking on the as yet undefined areas of dialogue, music and post-production.
A breakdown of the areas of responsibility for the game's audio. Green sections show areas of responsibility already established and under the implementation of the audio lead. Blue sections show areas that had as yet been undefined at the time I joined the Swordfish audio team in early production.
Given the amount of content that was already in production at the time I joined, there was a huge amount of sound effects, ambience implementation and design work that Mark Willott could continue while I concentrated on prototyping ideas for music and dialogue.
It became very clear that the production needed a narrative direction as soon as possible, as production was already underway on in-game assets prior to the existence of any written story. We brought in a local writer, Adam Hamdy, to begin to flesh out some very high-level story ideas and get an idea of characters in the absence of an official writer.
These initial ideas and characters were subsequently handed over to the chief writer who was hired in LA, Kamran Pasha, who worked closely with the core IP group (myself, art director, exec producer and lead designer) in Birmingham via several weeks of conference calls.
Together we sketched out the story and characters that were to populate the final game, while all the time working within the constraints of our in-game locations (which were fixed due to the amount of art and level design that had been already produced).
The grand-concept and art direction for the game established by executive producer Julian Widdows and art director Michel Bowes was that of an over-the-top "music video" and arcade-driven in style.
It is always great, as a sound director, to have such a clear brief and this bold visual statement had distinctive inspiration for much of the sound design, especially of the HUD (points accumulation and call-outs), as well as the weapons and explosion sounds being greatly exaggerated and over-the-top.
The game's overall direction was also not to take the subject matter too seriously, and to have more fun with the license in order to get away from hip-hop -- which is all too often is unable to make fun of itself. The dialogue, particularly the inclusion of a taunt button, was designed to fit this brief, and to create a feeling of an over-the-top hip-hop arcade experience. It also introduced a lot of fun and humor into the gameplay -- an ingredient often sorely missing from other hip-hop licenses.
Design of a suitable dialogue system with gratifying yet simple interactions was worked on primarily with the design and AI teams. This work began with breaking down enemy reactions and outbursts into AI categories such as "taking cover" / "throw grenade" / "attack" / "covering fire", etc.
We also broke down all the categories of dialogue that the player character (either 50 Cent or a G-Unit member) would require, in total each had around 50 categories. Each of these categories was analyzed for repetition in actual game play, and corresponding numbers of variants were mapped out for the more common categories, the biggest being the player's taunts.
The taunt button was going to be central to getting the feel of "being 50 Cent". Being able to use comical profanity at any time, totally at the user's discretion, certainly upped the fun factor, and was a technique that I had used previously on the Scarface game.
Also key to getting the taunt button integrated into gameplay and away from simply being a swearing button meant that when used at certain times after certain kills, as part of a combo, it allowed points to be multiplied with timed use of a taunt. The dialogue content of the taunt button was also made upgradable via the unlocking of extra themed taunt "packs" which increase the amount of things that the character can say on the button as the game progresses.
Because the game was set in a fictional Baltic / Mediterranean war zone, we wanted vaguely authentic and indeterminate foreign voice assets to be shouted in original dialects -- not in English with a foreign accent, which can often be repetitive and irritating.
This also meant that we could get away with much less offensive dialogue content, because once it was translated and shouted in an angry over the top performance by the actor, it sounded a lot more aggressive than it actually was. Myself and the core IP group all settled on a mixture of Russian, Serbian, and Croatian voices to obfuscate any idea of a middle-eastern location and to deepen the idea of foreign fighters and unknown forces with which 50 Cent and crew find themselves in confrontation.
There was already a fairly rich and complicated dialogue implementation system in place on the project, left over from Swordfish's previous release, Cold Winter. However, it had not had much attention for a long time and needed a re-write to handle much of the 50 Cent design requirements.
Because the dialogue we were going to be implementing for 50 Cent was nowhere near as in-depth as Cold Winter and didn't require large branching tree structures, I recommended that to save time and manpower we replace this with a new piece of technology we had recently developed at Radical for the Scarface, Crash, and Prototype projects in Vancouver.
The dialogue tool, called UDO (Universal Dialogue Organizer), is a piece of technology originally developed by audio coder Robert Sparks and myself (as well as the audio team at Radical) off the back of the Scarface game.
It is a pipeline-agnostic, stand-alone dialogue database that not only contains all the spoken content for the game in text form, it also allows you to organize, search and flag characters and lines with values for implementation and re-appropriation in the game. It is essentially a dialogue engine and dialogue database in one. It turned out to be fairly simple to integrate this into the Unreal pipeline at Swordfish.
50 Cent dialogue in U.D.O. (Radical's proprietary dialogue database)
During mid-production we also welcomed a new member to the audio team at Swordfish, Andrew Green, who pretty quickly took on sole responsibility for implementing the often complicated build steps and implementation code for the entire dialogue system.
This involved taking the data from UDO and introducing a build step which bundled the data into Unreal engine packages. The loading system also relied on these packages being of small enough size to load and unload as required, so things like dialogue variations were bundled into packages of around five to 10 lines each -- which for things like 50 Cent's taunts, a category consisting of a total of 99 variations, allowed us to load small variation packages which would be unloaded and replaced with new packages once they had been used up.
The total number of in-game dialogue lines was around 10,500, the majority of those belonging to player and co-op characters (50 Cent and G-Unit). Each of the twenty or so foreign enemy characters had around 120 lines each.
Interactive Music Player. With access to many unreleased 50 Cent and G-Unit tracks, there were many ways in which we could integrate 50 Cent's music into the game. However, many of the ideas and suggestions we explored in early-mid-production turned out to be overly complicated. Some of the ideas explored rhythm-based game play centering on the BPM of tracks that were playing, scoring extra points etc if kills were performed to the beat.
The simplest solution, and the one which offered the most choice and ease of use for the player, was to offer a music player interface in which tracks could be unlocked, and organized into a custom playlist. The actual music player design was copied over directly from the Scarface game (in which I had implemented a similar music player concept), and given an updated art direction and streamlined implementation.
Original Score. It was decided from the outset that we not only have a music playlist of licensed music, but also an original hip-hop score for the game. The setting and style of the fictional Baltic war zone implied quite a few different directions that could be taken, and we began to explore original orchestral action-movie style scores to hear how they would sound in the game.
Although the style felt great for the setting, it did not really lend itself to conveying 50 Cent's personality or enough of a hip-hop feel, which absolutely needed to be stamped all over this game if it was to succeed with the 50 Cent fan base. 50 Cent's record company and management agency, Violator, came to the rescue in this regard with a roster of great producers, the hottest of which was Swizz Beatz, on whom we settled to provide the whole of the original underscore for the game.
Direction overview and documentation was drawn up and work began in the early summer of 2008. After around three months Swizz passed over mixes of the finished 24 tracks (one track per level of the game) along with individual instrument stems including exact BPM information.
These stems made it very easy to re-appropriate sections and prepare the tracks for implementation in the game. Each of the tracks had to be cut and re-arranged so that they represented musically different sections of gameplay, such as "gangsta fire" modes, and side-mission modes, as well as beginning, ending and conventional gameplay sections.
Balancing Score and Licensed Music
The choice between score and playlist in the game was resolved by having licensed music play whenever there were items in the player's playlist, and original score play when the playlist was empty, in this way players could choose which version of the soundtrack they wanted to hear at any time.
One of the key objectives for the soundtrack was to get the player's blood pumping right from the start of gameplay, so a variety of the best tracks that matched the action were chosen to be auto-unlocked and already populating the playlist by default, this way giving players the best of the "50 Cent" feel right out of the box. Then, as the player worked through the missions, they would discover a further depth of music player features and then access the hip hop score.
Visual Style and Sound Design
There is a striking richness and hi-resolution detail to the art direction of the game that initially surprised me when I first saw it. Finely detailed, high-definition particle effects gave clues to the appropriate audio direction for the sound effects in the game. There needed to be a lot of corresponding detail and richness in the sounds that were created for the game, namely in the destruction, bullet ricochets and key explosions.
Adding many layers of rubble and fine debris to the tails of the effects to reflect and underpin the visual density of the game was one of the key directions established for the sound effects design. Another key direction was for the sounds of the weapons that 50 Cent uses in the game. These needed to follow not a realistic weapon model, but one of over-the-top cinematic power.
One of the things learned from the weapons work on the Scarface game was that it is not about having a wholly authentic weapon sound or weapon recordings, but that the feeling of shooting a weapon (something we did at a shooting range in the Nevada Desert for Scarface) that was the key to giving the players the fun and impression of overwhelming power that comes from holding and firing an automatic weapon.
In this sense we chose to concentrate very hard on getting all the various weapons in the game scaled correctly and feeling lethal and fun, which meant that no original recordings needed to be made. We instead concentrated on layer upon layer of creative sound design using only content from Vivendi's sound effects library.
This over-the-top direction in turn creates an important point-of-view effect for the player, in that they are hearing the weapon from 50 Cent's perspective and getting his larger than life bulletproof personality communicated through those sounds.
To achieve this, sound designer Mark Willott and I worked very closely on these particular aspects of the sound design, focusing on and fetishizing many of the reload, shell-casing, and bolt-action sounds to augment the firing sounds. In the end we achieved fairly quick cyclical iteration on the explosions, weapons and ricochet sounds, reviewing weekly during production, and these sounds turned out to be crucial to the overall action of the game.
We often found ourselves laughing out loud at some of the gunplay in Blood on the Sand due to its sheer over-the-top nature, which to me is always a great indicator of a solid, fun action game.
With the sound direction firmly established on-site at Swordfish in Birmingham and the focus of production moving onto assets generated in LA, in November 2007 I returned to Vancouver, continuing sound direction duties on 50 Cent remotely via regular conference calls. As I was now responsible for the dialogue and music content in the game, it made sense that I was closer to the center of operations at Vivendi LA in order to work on the same time zone with the executive producers and our LA-based voice-over studio.
It also transpired that the cinematic cut-scenes for the game, on which I was responsible for cutting sound and mixing, were being outsourced to FX-house Rainmaker in Vancouver, and again being on site in Vancouver allowed me a close working relationship with the cinematics team.
Voice Casting and Recording
February and March of 2008 saw the scheduling and recording of the majority of the voice assets in the game. Eric Weiss, voice-over and talent director at Vivendi Games in LA, was instrumental in getting the best performances out of all the foreign voice talent and, most crucially, out of 50 Cent and the G-Unit.
Eric and I had worked together previously on Scarface and he had also directed 50 Cent and the G-Unit for the 50 Cent: Bulletproof game back in 2005, so there was already a familiarity and respect in place with 50 Cent and the Violator crew. Eric flew to New York for a week to direct the sessions, where I listened in and aided with context direction via a phone link to our studio in Vancouver.
All the G-Unit members recorded for the game. Tony Yayo, Lloyd Banks, and DJ Whookid were incredibly professional and accommodating to our script, offering suggestions and very often more appropriate "street" alterations where necessary. Curtis Jackson himself was fresh from working with Robert DeNiro and Al Pacino on the set of the movie Righteous Kill, and as soon as he entered the voice-over booth his professionalism was evident.
Curtis also has incredible stamina, more than any voice artist I have ever heard; he did four straight hours of shouting lines from our in-game script and did not once want to let up for a break. Among the other actors for the game were Lance Reddick (The Wire) who plays Carter, and Dwight Schultz (The A Team) who plays "The Harvester" Wilder.
We also made use of Side UK for a few of the voices such as Omid Djalili (The Love Guru, Pirates of the Caribbean: At World's End) who plays likeable hip-hop wannabe Eddie in the game and who was recorded in the UK under the on-site direction of our own Game Director, Julian Widdows.
50 Cent was fresh from working with Robert DeNiro and Al Pacino on the set of Righteous Kill
The cutscenes in development at Rainmaker were delivered on the production Alpha date, giving me two weeks to complete the primary pass of sound effects cutting and pre-mixing for our Sound Alpha date. The cinematics were are all pre-rendered FMVs, and because the in-game characters, sets and effects are all of such high resolution, it is often hard to tell the difference between in-game and cut-scenes.
Due to the high quality rendering of these scenes, there was great incentive to get them to sound as cinematic as possible. A week of Foley recording was carried out at Sharpe Sound in Vancouver, and these Foley tracks were integrated directly into my Nuendo sessions back in the mix studio at Radical.
Working on FMV cinematic visual assets offers much more creative inspiration for the soundtrack than NIS in-game cut scenes in games, as the latter tend to give very low resolution, grey-blocked assets to the sound designers to work on.
The more detail that is visible in the final movies that go out to places like Foley outsourcing, the better the quality of work you get back in the end, due mainly to it being obvious what materials the characters are walking on and interacting with. As well as this, having all the visual and particle effects as part of the movies adds additional clues for the sound designer.
Post-Production and Final Mix
Music and dialogue assets were edited and mastered at Radical by our senior studio engineer Lin Gardiner, ensuring that all the levels of both the score and the licensed tracks from various sources all had consistent volume levels.
Right from day one of my involvement in the project, I wanted to repeat the success we had with the Scarface game in terms of post production time allotted after Beta so we could take time to polish and finesse the game's sound with everything in place. Sound Alpha and Sound Beta dates were established with the project managers to occur two weeks after both production Alpha and Beta.
It was also always the plan to use the newly constructed 7.1 post-production mix suite at Radical Entertainment, which had recently been set up and calibrated by THX, as the ideal location for the final mix of the 50 Cent game, taking full advantage of the inter-studio sharing between Vivendi studios.
The studio had been designed from the ground up to not only handle the mix of internal projects at Radical, but also to accommodate private VPN based projects from other studios. Audio Lead Mark Willott flew out from Birmingham to Vancouver to join me for the final two week post-production audio phase on the project during August of 2008.
The post-production plan that we followed allowed for a week of sound effects replacement, during which we played through the entire game, flagged priority sound effects that we felt could be improved and used Radical's in-house sound designer, Cory Hawthorne, to re-work any sounds that we needed to replace.
Cory would create several versions of each new sound, each with an identical memory footprint to the sound that was being replaced, and we would then try it out in the game. More often than not we would decide on one of those replacements there and then to be the new sound.
Post-production sound design and mixing took place on site at Radical Entertainment in Vancouver. One of the first titles to be mixed on the newly built 7.1 surround mix stage.
After a week of sound replacement, we moved on to a week of mixing for the Xbox 360. One of the unique aspects of mixing video games is that, as there are no standards in place for reference level mixing and monitoring, as there are in movie post production, most of the competitive games are of dramatically varying output levels.
It was decided that we wanted to mimic the output levels that Gears of War had used, as this was the game we had been most closely modeling in terms of gameplay and target audience.
To this end we attempted to get our output levels as close to Gears of War as possible, listening at slightly underneath the -79dB reference level as we mixed and referencing the output surround waveforms generated by the game, generally this is somewhat quieter than we would prefer to mix a game, but it matched the expectations of the audience for this kind of game.
As well as matching cinematic levels with those of in-game sounds, much of the actual in-game interactive mixing involved ducking out explosions, ambience, music and physics sounds whenever important dialogue was installed. This was achieved via the installation of mixer snapshots that are triggered to coincide with the event in the game, and then un-installed on event completion.
After playing through and mixing the entire game in surround, final checking of the stereo and mono down-mix was achieved using the Studio Technologies Model 79 monitor controller built into the mix studio's console desk.
Proprietary mixing technology, Noisemaker, developed by Justin Cadicott at Swordfish Studios, allowed for the installation and live tuning of mixer snapshots during game play.
Finally, after mixing the Xbox we cloned all the mix settings for the PS3 version of the game and tweaked a few levels for the discreet PCM 7.1 mix output. The game had been running in 7.1 all through development on the PC, so there was very little extra work to do in order to support the two extra channels required for the PS3.
There are always sacrifices to be made as production deadlines loom closer, and there were a few features and areas of content that we could have improved given more time.
We did plan to record a significant amount of new and replacement revision lines for the game with 50 Cent and the G-Unit. We managed to get these pickups recorded with all of our actors except the G-Unit in the end, purely due to scheduling conflicts and requirements which meant that we needed to cut off our content implementation before we could get any of the new assets.
This is one of the areas that we all felt the quality of the game's dialogue assets could have been improved, offering tighter integration with the events of the level designers. In the end, what is in the game could certainly have been improved, but is still of high enough quality that we were happy we could complete development on the title.
In terms of music, some of the features were very late at going into our code. Beat-mapping that would allow us to transition the Swizz Beatz score on the exact beat was very late going into the game, but an essential feature that needed to go in.
One feature that we did have to drop was random selection of parts within a looping track which we had to forgo due to the risks involved with introducing new FMOD code updates into our Alpha build.
As such, it was designated as something that we didn't need to have in order to ship the game, and so rather than have randomly looping parts with each track, we simply had a single .wav track made up of the four previous random parts.
Given these slight improvements and missed opportunities, the whole audio team is very happy with the final game we shipped, especially given the pressure to deliver quality in a high-profile license IP such as this. I'd like to extend my personal thanks and congratulations to everyone in Birmingham, LA and Vancouver who worked on the audio and helped to create a great sounding hip-hop game.
Rob Bridgett - Sound Director
Mark Willott - Audio Lead
Justin Caldicott - Senior Audio Coder
Andrew Green - Audio Coder
Voice Director - Eric Weiss
Music Licensing - Steve Goldman
Recording Engineer - Mike Patterson
Post-Production Sound Design - Cory Hawthorne
Music and Dialogue Mastering and Editing - Lin Gardiner
Return to the full version of this article
Copyright © UBM Tech, All rights reserved