AGC: BioWare on 'Writing for Digital Actors'

Author: by Wendy Despain, Austin

Bioware sent two representatives to the Austin Game Writers Conference to present 'Creating Characters for Games: Writing for Digital Actors,' about the way they approach writing for games on the new consoles. Mac Walters, Designer/Writer for Jade Empire and the upcoming Mass Effect and Mike Laidlaw, lead writer on both titles, discussed the differences they've found as they've built a title for the Xbox 360. They said it was both a blessing and a curse to be able to work on a platform that could render nearly photo-realistic characters in nearly an unlimited number of situations. Walters started out by explaining their problem. In the old days, RPGs involved lengthy conversations with characters mostly seen at a distance, where facial expressions didn't matter and everything had to be conveyed in words displayed on the screen. But in Mass Effect, they have changed both the graphics and the dialog engine to the point where they're achieving nearly television-quality close-ups on characters in conversations. Now facial expressions and body language have to be put into use, or the game looks stiff and unreal. Bioware has started talking about their characters as "digital actors" and writing in a new way. The Planning Process Laidlaw traded off with Walters throughout the presentation, as they explained that their process begins with the concept that great stories make great games. From there, they build up the world by addressing where the story starts, where the story ends, and how the player moves from one point to another. This is when they develop a design bible, which can get quite lengthy but serves as a planning tool to keep everyone on track. This design bible defines the setting, the story arc, the protagonist, the antagonist and every character with a speaking part. It breaks the narrative into chapters or worlds, including any big bosses the player needs to face to move ahead in the story. Then they plan out subplots using the earlier planning they did on the setting as a starting point. This is how they keep the minor characters with smaller problems - the missing puppy, the monastery in trouble - consistent with the theme of the larger story. This process hasn't changed as much as the level of detail required. In television, the writer can provide a basic sketch of a character's looks and actions which the casting director uses to hire an actor, and the wardrobe department dresses and the make-up artists work on. But in the next generation of games, the writer starts to fill all these roles and that of director as well. The artists need a detailed description of the character's features and mannerisms. For each conversation, other team members need to know if someone pulls a gun or walks around while they're talking. Of course, those team members make choices and contributions too, but to keep everything organized a lot of documentation needs to be done and that job usually falls to the writers. In the past, it was pretty easy to take some basic dialog and tweak it a little for each new NPC. Changing the tone of the text made the little boy into a grumpy old man. But now Bioware is doing all of their dialog in voiceovers with lip-synched digital actors, so every writing change impacts lots of other team members. Laidlaw said their new motto has become, "If you fail to plan ... you can plan on having your fingers broken by artists and programmers." He advocated copious notes, good team building skills, and each writer really taking ownership of specific characters so the voice is kept consistent throughout the game. Where Does It End? In this process, a problem of scale quickly develops. As dialog trees branch out and players are given choices, it's technically possible to have even minor characters built up into elaborate digital actors requiring months of animation and voiceover work, not to mention the documentation requirements in the planning process. So Walters addressed, "How do we deliver a good product without blowing our minds and our budget in the process?" First, he said they define the spectrum for the player and be consistent: what the player can and can't do, how far they'll go with the humor or the violence, clearly setting boundaries on the tone and playability of the game. This is something, he said, they lock this spectrum down early in the process. Then they define the voices, starting with the player character. For each major character they move to a new point in the spectrum. One character is the straight man, or the gritty cop, while another cracks jokes all the time. This differentiates the characters and fills the whole spectrum as defined earlier - keeping the tone consistent throughout the whole game. Don't Be Afraid Of Subtle Differences Both speakers emphasized that players enjoy making fine-grained decisions. To illustrate, they used the example of a main character choosing to be either good or evil. They said you can give the player one big checkbox at the beginning or the end of the story, or you can give them several throughout the game at the ends of levels, or other natural breaks, but they prefer - and they think the audience prefers - to give players many small choices in the little actions they take, adding up to the ultimate sum of good or evil. They used the "killing the puppy" example to show how it provides flavor. Instead of a character giving three neutral responses and then jumping suddenly to, "I'm going to kill your puppy." They explained that it flows more naturally for characters to possibly start out with a neutral response, then provide a few hostile responses before proclaiming, "I'm going to kill your puppy." They encouraged writers to pay attention to these minor differences in dialog. This not only becomes more interesting to the player, but having specific directions to move in can help keep the scope of the work under control. Observing The Mass Effect Laidlaw then showed the audience a demo of their upcoming Mass Effect game. It uses their new style of character animation and dialog engine. He showed how conversations with NPC's are handled with a radial interface of summarized dialog choices. So, instead of the player reading three different possible responses to the NPC and then picking one, they're presented with paraphrased choices on a wheel, with a generally good response at one point on the dial, moving around through neutral responses to aggressive or mean or shy responses. The words displayed on the screen give the player an idea of what response they're choosing, but it doesn't give away the exact wording or the way that sentiment will be displayed in actions. This allows for a more complex mood system and camera changes. From the player perspective, this is dialog interface is a faster, more seamless experience and provides more entertainment value, since they can be surprised by the delivery of the chosen line. Walters also said this results in more replayability because a player can't read all the responses the first time through. The net effect feels more like a continued game experience, rather than stopping the gameplay to read. The example they showed included the player character choosing an aggressive response from the radial menu, which turned into a sequence where guns were drawn and threats were made. "Think of all the people who were involved in that," he said when it was over. He did explain that not every conversation is animated using individual motion capture sessions. Instead, elemental actions are built into a library, which are then blended together - a turn blends into a lean, blends into raising the gun. These looked pretty seamless in the demo, but he said they aren't hand-tuned except on major characters in important situations: yet another way they manage to keep the scope of the project from getting out of control.