Alpha Protocol Blog Update from Audio Producer Mikey Dowling

Obsidian Entertainment audio producer Mikey Dowling has paid a visit to the Alpha Protocol blog on IGN for a quick discussion about the steps required to produce time-critical choices and “reactive callbacks” in the game from a written script.

Alpha Protocol – From Analog Script to Digital Vision

Mike Thorton walks into a dry cleaner’s shop; he hears muffled yelling to which he calls out, “Hello?”

“In the back.” He follows the voice to the source to see a man tied in a chair, tape over his mouth, along with another man – an informant for Thorton in the Taipei hub, Steven Heck – walking toward him with a huge bottle of bleach. Clearly the man tied to the chair is about to be tortured for incredibly important information; yet Thorton needs information of his own from Heck -but he’s (obviously) preoccupied. Alas,Thorton needs to say something and what he says could affect how Heck looks at him from then on. Does Thorton hesitate, offering to come back later? Does he instead offer to help, possibly gaining some reputation points with Heck? Or does Thorton go the professional route and get right to business, thus possibly angering his would be informant?

The three choices – or Stances as they are referred to in AP – will come up in every major cinematic in the game. Matched with a timer, the player will have to make their choice quickly to keep the conversation going toward what they think would be best. Yet how do we, as the developer, incorporate what can be a spiderweb of choices and reactive callbacks into the game with relative ease?

In Alpha Protocol, each dialogue starts off as a script. Once the script is finished, a table read is done to make sure there isn’t anything that just doesn’t flow right. Once any needed adjustments were made, the scripts were then handed off to some of the members in our QA department to start putting into templates. The template process was integral for several reasons. 1.) It gave us the WAV file names so the recording studio (Womb!) would be able to accurately name each file. 2.) It allowed us to test the flow of a conversation to make sure it worked correctly. 3.) It gave us the files necessary to create not only our subtitles and FaceFX animations, but also our Kismet (the scripting language for the Unreal 3 Engine) .t3d files so we could put the scene into the game itself.

Once the script is tested in the template to make sure everything links correctly, a flow chart is then created to help with setting up each of the scenes in Kismet itself. While some of the cinematics in AP flow in a rather straightforward manner (like the scene with Heck) others can literally look like a complex web of choices, events, and reactive moments. It’s for these conversations that the flow charts help greatly.

In a previous blog, written by Obsidian’s Matthew Rorie, a character named Grigori is mentioned. While the big moments of the Grigori conversation are either acting calm (for the most part) or slamming his head into the bar, there are a lot of subtle moments in that conversation that won’t be apparent to the player on the first playthrough. That scene has quite a few reactive points that are very subtle, but show how much we wanted to focus on rewarding players for their choices.

At the beginning there are three possible starting points for the scene that either goes along in the standard matinee to Stance Choice to matinee format; or things get more complex dependent on the variables the player has already set throughout the course of the game. Which hub has the player gone to after Saudi first? What activities did the player do in said hub? And once in Moscow did the player go to Grigori first or not? Each of those variables are accounted for in the first few moments of the scene.

This is why the flow charts come in handy. If a scene gets to a point where it is this diverse and reactive, being able to map it out before setting the scene up in Kismet is incredibly helpful. You’re able to make sure all of the logic is in place, then go right into building the scene once the correct assets are available.

The assets in the case of the scenes are the level in which the scene is going to take place, the proper actor proxies for the characters in the scene, and last but certainly not least, the WAV files for the scene. Once the WAV files come in from the recording studio, the writer for the scene (either Chris Avellone or Travis Stout) will choose from the alternate lines for which ones they feel flow with the scene more appropriately. Once alts are chosen, the WAV files are put into what we called VO (voice over) packages in the engine. The VO packages consist of a WAV file package, a SCUE package (or sound cue), and the FaceFX package (so lips would animate correctly to the wave form).

Once the VO packages are made, the scene is ready to be set up in Kismet.

Once the scene is implemented in Kismet one of the members of the cinematics team (under the guidance of Joe Bulock and Shon Stewart) will do a pass on it. This will include adding expressions and animations to the actors to help bring the scene to life. Then finally the scene is passed over to our audio department so our audio director, Scott Lawlor, and one of our sound designers, Andrew Dearing, can do a music and foley pass on the scene respectively, thus giving it the “movie feel.”

It’s a process that takes time and a lot of input from a few departments, but the finished product is quite satisfying to see come to fruition. This is really apparent when it comes to a scene like the Heck one mentioned above. Having gone from reading the script, hearing how our incredible voice actors acted it out, to how our cinematics and audio teams injected it with awesome was a process to behold.