A CG character's facial animation speaks for itself
By Karen Moltenbrey
In an ongoing series of UK television commercials for First Direct Bank, a quirky 3D character doesn't just steal the show-he is the show. Voiced by English comedian Harry Hill, the odd little fellow mills about within a black void, grumbling about every aspect of daily life. His dialog is short and to the point-in one instance, he repeats an emphatic "No!" a dozen times before suddenly changing his mantra to "Oh, alright then; I'll meet you halfway." Getting through the day frustrates him, but his banking does not.
In these minimalist spots, there are no backgrounds or scene elements to carry the piece along, just the black-and-white underlit textured character. While his outward appearance lacks color, he makes up for it with a range of facial and body expressions that run the gamut from ecstatic to downcast, from quizzical to disinterested, all in a single commercial.
|This 3D character is the sole focus in a series of UK television bank commercials. Without background elements to help carry the story line, artists had to make sure the character model was capable of a wide range of expressions. |
"We're trying to attract the audience through a generic character and nothing else, which is unusual for a commercial and extremely bold for a bank commercial," says Stuart Gordon, director and co-owner of Realise Studio in London, which created the peculiar but likable character. "We knew that this spot would live or die depending on how well we could make the little fellow perform. Although simple-looking, the character has a tremendous amount of embedded complexity that enables him to pull off his dramatic acting performances."
Creating the character began with the creation of a clay sculpture of the chap's head, which was then laser-scanned. Next, Gordon squashed and stretched the scanned data inside Side Effects Software's Houdini until he achieved the desired look for the character-akin to the style used for the human characters in the feature film Toy Story 2.
According to Gordon, early in the project he and partner Paul Simpson realized that traditional CG techniques would have restricted their ability to achieve the character's wide capability of facial expressions. So they used a procedural methodology, developed by Gordon christened "shrink-wrap sculpting," which divided the head template data and automatically rebuilt it into individual mesh components. This process enabled the animators to apply complex and sophisticated animation controls easily.
"By separating out the head [data] in a procedural 'layered' system, things like the mouth and eyes, as well as the facial expression, could be animated individually without the overhead of manipulating the highly complex entire head," says Gordon.
|Programmers at Realise spent a great deal of time setting up the lip-sync and facial expression controls in Houdini so the animators could create the mouth movements without having to learn or understand the underlying program technicalities. |
During the final stage, all the separate movements were added back together. Although this process took longer than building a traditional CG model, it enabled the team of animators, which included Gordon, Vanessa Arsen, and Steven Brown, to modify the facial gestures quickly and easily.
In the end, the character had close to 50 individual facial muscles, which were manipulated using high-level expression tools embedded in Houdini, enabling the artists to achieve a complete range of facial expressions. To animate an expression, such as a smile or frown, the artists did not have to move a muscle; they simply chose a look and the individual facial muscles were triggered automatically.
Because the ad agency planned to use the character in cinema as well as television ads, the model had to be created with a high level of detail, down to the tear ducts in his eyes. Also, the character had to be carefully constructed, since he had to be able to perform to just about any given script.
"We spent about two months putting the guy together to where he was performing basic walking and talking motions," Gordon adds. The group created the keyframed animation and lip sync after they received Hill's audio track. In each 40-second spot, the character averaged about 10 lines of dialog, which required about four to six weeks of additional animation, testing, and setting up the particular camera paths. Final rendering required an other week.
The character was rendered using subdivision surfaces, which gives him an especially smooth finish, and the final scenes contain a great deal of light sources, which were achieved using Cinema Graphics' Shade Tree software. The scenes were then rendered using Pixar Animation Studios' Render Man running on SGI and NT workstations, while the other software ran on those platforms as well as Linux machines.
|Because of its versatility, the First Direct Bank character also has been been appearing in a variety of print and cinema ads. |
With three of the ad spots completed and another currently in production, the animators have refined the basic character model. Now they are concentrating on developing his personality through his movements, dialog, and possibly dress. "We're really getting to know this little guy, and so is the audience," says Kate McDonald, animation producer at Realise.
Houdini, Side Effects Software (www.sidefx.com)