0
Hello, my name is Chuck Jackman. I direct the motion capture and animation of the cinematic conversations in Guild Wars 2, and I am here to give you a little insight into the process. What exactly is a cinematic conversation? I’m glad you asked. If you’ve played the Guild Wars 2 demo, you’ve seen the brief cutscenes that take place between the story NPCs and your player character. These are cinematic conversations.
Until now, you have only seen our placeholder versions of the conversations. This is because, here at ArenaNet, we have a very iterative development process, which allows us to layer in content at various levels of completion. This lets us see how everything is jibing and then tweak and refine as we go. We took advantage of this approach with the cinematic conversations so that our writers and designers could still get a feel for the pacing of the story and events while we worked on identifying what we wanted to get across visually. This also allowed us to refine the tools and process.
When we started production of Guild Wars 2, we knew we wanted to do more with the story through cutscenes than we had in the original Guild Wars. It was decided early on that we would have three tiers of cinematics in the game. Tier 1, or “Full Cinematics,” would be the high-action, Ascalon-destroying, dragon-smashing moments of the story. These would be handled in our proprietary tool, called Cameo, and done by the same amazingly talented team that handles the beautiful trailers you’ve seen. These use a marriage of 2-D and 3-D art and animation.
Then we have tier 2. These are the nuts-and-bolts story moments, all the twists and turns, and the proverbial meat of your character’s personal tale. These are not pre-rendered and take place when the story needs more than in-game action to progress. These are the cinematic conversations.
Scenes occur in the game as text spoken by NPCs and are usually more for giving direction and information than they are about story content.The third and final tier we refer to as “scenes.”
We knew that we had more life in Guild Wars 2’s cinematic conversations than we had in the original Guild Wars, but we also wanted more life than we had in the temporary versions already in place. These placeholders were basically two people standing there playing minimal, seemingly random, animations. They featured some very basic lip-synch that resembled more of a duck’s bill than a human’s mouth. This was because these temporary assets consisted of nothing but an automated mouth-flap animation and a very limited pool of basic idle and talk animations. This, of course, did not allow the characters to portray the emotion we wanted in a realistic or believable manner. It also made it impossible to have characters gesture during ideal moments of dialogue because these were fully “canned” animations.
We needed a versatile solution that allowed us to reuse and recycle animations from a controlled pool of content across a broad and vast story line without things getting stale. So we got together with some technical artists, programmers, designers, and writers, and determined what we wanted to achieve and how we could do it. After some prototyping and experimenting, we decided to use a layered approach to the cinematic-conversation animation system, utilizing motion capture, hand-keyed animation, additive animation, and very robust lip-synching software.
Once this plan was mapped out, we then blotted out the sun and the stars, summoned forth fog, thunder, and lightning, lit ritualistic candles, and handed our final request off to the tools programming team. They, in turn, invoked their most sinister black magic and tore the needed code straight from the minds of the ancient gods, trapping it in an awesome tool set, allowing us to accomplish all our wicked—er, cinematic conversation needs.
This new tool set enables us to have idle animations that remain constant, based on a character’s emotional state, and still give us the ability to layer in gestures and facial expressions. The flexibility of additive animation gives us a way to create small and quick gesture animations that can be dropped on or added to any other idle animation we’re using. For instance, we can animate a shrug animation, apply it to a confident idle, and it will play correctly. We can also use that same shrug on an angry idle and it will still look and feel correct.
We also use a layered and additive approach on the face. This allows us to animate a mood or emotion-based “face idle” and layer on the lip-synch animation as well. We can then drop additive facial gestures at the appropriate time in the dialogue. For example, we can have a character looking timid or frightened while they talk. Then maybe something happens to scare them and we can drop an additive flinch animation onto the character’s face, body, or both independently at just the right moment so that they react to what is happening to them in the scene.
So here’s a quick of how we put a cinematic conversation together: After the final voice-over work is done and the audio team has worked their voodoo on it, I play through the scene a few times to determine the emotional state of the characters and find any spots where we can place gestures to accentuate beats of dialogue. I then take these notes and head over to our motion-capture rumpus room with our animation lead Heron Prior and other motion capture guy, Igor—uh, I mean Andrew. We put our chosen victim into our motion capture gear and have them bend and contort to our whims and wishes until they either give us the performance we want or break down into a sobbing mass of despair.
We do our own motion capture on site using a slightly different method than what most people may be used to. We use a camera-less motion capture system called the Xsens MVN system, which is a suit with a bunch of little inertial sensors in it that track the constant movement of the body and relay that info back to the computer. This system gives us a lot of flexibility because we can use it outside rather than just inside the studio.
Once this is done, either I or one of the other animators put the data into Motion Builder. We then do extensive cleanup and reposing, and then it’s off to Maya for some additional love and polish. This becomes the “base” idle, or foundation, animation for the emotion. Think of it as the body language or the overall pose and movement of the body. Once we have the idle layer in, we drop in the gestures—both motion captured and hand keyed. Then Heron and I watch them again and we determine where, if at all, we should add some additive gestures. We create these through traditional hand-keyed animation.
Let’s take a look at an example of a cinematic conversation in Guild Wars 2:
If you watch the video above of the cinematic conversation between the player and the Priestess of Dwayna you can see an example of this. The player’s head rub, or confused gesture, at the beginning is a full body gesture blended out of and into a neutral emotion idle. The hand movements that the Priestess of Dwayna performs while talking to the player are small gestures that are added into the idle.
While animators are busy at work on facial expressions and facial gestures, the lip-synching is generated by running the final audio files and text from the writers through a program called FaceFX. This marries the phonetic sounds of the dialogue to shapes created by members of our animation team. As the pieces come online they are layered into the cinematic conversations, refined, polished, and tweaked until we are happy with them.
Lip-sync is not the only thing we are doing with the faces, however; we also make facial loops and facial gestures. These work similarly to the body idles and gestures I mentioned above, but a key difference is that the facial loops can be used across a broad range of body idles, which gives us the ability for subtlety and extremes.
For instance, in a scene with an enraged character you might put one of the angry body idles on him as well as one of the angry face loops. If a scene called for a stern character, we don’t have a “stern idle” per se, but we could pair a confident body idle with the furrowed brows and sharp, tense facial angle of an angry face loop in order to portray that emotional state.
Let’s take a look at another cinematic conversation to further illustrate these improvements:
Well, that about does it. By the looks of things, my “brief” synopsis of our process was not that brief after all, but I hope you enjoyed reading about our cinematic conversations as much as I like to make them. Take care, and hopefully I will see you in-game!
View the full article
