Jim Boer on the Evolution of Cinematic Conversations
http://www.arena.net/blog/wp-content...ime-304x62.jpg
One of the things we game developers always have to struggle with is reconciling our ambition with the reality of a production environment. In the case of our cinematic conversations, there’s certainly been a significant challenge in designing a system that not only meets all of our design requirements, but that is also realistic enough in scope to allow us to meet our production goals.
First, it’s important to note that we’re talking about an enormous number of these storytelling elements —around fifteen hundred cinematic conversations are implemented or planned at this point. They’re used throughout the game, in every branch and offshoot of your personal story, as well as in both the story and explorable modes of our dungeons. And that number doesn’t include our larger, handcrafted full cinematics that require an entire team of artists and animators to create.
The bulk of our production resources are spent in many other areas of the game, so cinematic conversations needed to be reasonably fast and easy for designers to create. We were also given some specific design requests, such as the ability to branch dynamically, based on various story arcs, a character’s personality, or other criteria. Additionally, we didn’t want to “time restrict” the localized audio by requiring actors to precisely time their deliveries to the English track. Avoiding this allows for more natural non-English performances while cutting costs, which in turn means that more lines can be recorded. And finally, we wanted to adhere to the overall cinematic style we’d established for Guild Wars 2.
So, what sort of system would work best to meet these requirements?
The visual style of the cinematic conversations went through a lot of design and technology iteration. At one point, we considered posing the characters statically and applying filters to try to make them look more hand painted. The idea was to create a more comic book type of look, and the intention was to visually indicate that these were not our full cinematics, but a simplified way of presenting routine conversations between characters in a more compelling manner than the traditional wall of text in a dialog .
At the time of our initial design meetings, our characters were designed to only look good at standard game-viewing distances, but were not suitable for cinematic close-ups. This presented an additional quandary for the cinema team. We didn’t have any extra bones allocated for facial animation, and because we’re using a combination of unique faces and customized vertex morphing, any sort of mesh-based animation (i.e. morph targets) was rather impractical, since we’d have to create those animations for each preset head. As such, we settled on a visual compromise, pushing the character back far enough so you could see them in reasonable detail, but not zooming in far enough to see the flaws that would become apparent in a close-up shot.
Painting over the in-game scene gave us a few advantages as well. Since we don’t have to render the world during these scenes, we can ensure that the conversations will run smoothly no matter what the machine. At the same time, it masks background distractions from the player’s field of view—something that’s difficult to control in an MMO unless the player is alone in an instanced region. Finally, we feel that the painted backgrounds mesh seamlessly with our unique cinematic style, presenting a more coherent experience when presented with our full cinematics.
Since those early days, we’ve introduced additional tech that gives us high-resolution heads for specific situations, such as during character creation or cinematics. Chuck Jackman has already discussed the improved animation system that we’ve recently completed as well. What’s interesting is how this process has seen us move from our early ideas of static poses and stylized character rendering to a more lively and realistic portrayal. In game development, there’s nothing too unusual about things ending up a bit differently than you originally planned; and you can rest assure that, as we do with everything else, we’ll continue to look to improve our cinematic conversations even further.
Iterative development means showing off your work early, and continuously working to refine and develop it. In the process, it sometimes leads to interesting and initially unexpected destinations. This is by no means a production methodology exclusive to ArenaNet, but I think we’re fairly unique in how transparent we’ve been throughout the development process, and how we’re allowed (and encouraged) to engage the community as it’s happening.
While it can be a little intimidating to see our work placed under a microscope and scrutinized down to the pixel, I think we’re actually pretty fortunate to have a community that has enough faith in us to expect excellence at every level. After all, it’s no less than what we strive for ourselves.
More...
