From Digital Emily’s realistic facial animation to advances in stereoscopic technology, the buzz word this year at Siggraph 2009 was “believe”.
The goal of animation has always been to make the audience believe your characters are real; to empathize with them, care for them, or fear them depending on their role in the story. Through a willing suspension of disbelief an audience can have a deeply emotional response to completely fabricated characters.
Now, more than ever, this suspension of disbelief is becoming unnecessary, and CG animation and technology, as a whole, is about to fool its audience. From Digital Emily’s realistic facial animation to advances in stereoscopic technology, the buzz word this year at Siggraph 2009 was “believe.”
Believe that we are only a few years away, if even that, from creating CG characters that move and react like real people. We are reaching a pinnacle in which actors will virtually live forever.
Much of this year’s Computer Animation Festival discussed the stereoscopic boom seen this past year and how to maximize this technology’s functionality and aesthetic. Almost every CG-laden film this year used the appeal of 3D to draw audiences to the theatre and pay more, even during the current economic struggles. There is no doubt that stereoscopic is here to stay, like it or not, and Siggraph confirmed that this technology will also soon be widely available in the home.
Keynote speaker, Bob Whitehill of Pixar Animation, discussed using stereoscopic technology fort ultimate visual and emotional impact, and not falling into a cheap thrills trap. Obviously, 3D has been used to “wow” the audience and surprise them with the unexpected, however he stressed the importance of consciously using every aspect of a film to help tell the story first and foremost. “Why are we doing it?” is Pixar’s main concern with every element they add. For example, there is a contrast of characters inherent in Up and various aspects were manipulated to continually reinforce this contrast for the audience, such as highly saturated vs. desaturated scenes, or square character features and props vs. round and circular features. For 3D this is no different, and Pixar cleverly used varying degrees of depth to enhance the emotion of a scene and connection for the audience. In “square” sequences, as Whitehill referred to them, or scenes in which Carl was the dominant character, Pixar used less stereo to enhance the trapped, depressed nature of the scene. In later scenes, stereo was used more to raise the audience’s excitement and give a sense of space and adventure. Also, every shot in Up was created with inward 3D, rather than 3D that extends out into the viewers real-world space. If the stereo had protruded into the audience it would “break the fourth wall” and disturb the viewer’s suspension of disbelief, which ultimately weakens the emotional impact of the story.
“Know your movie. Is it a 3D extravaganza or more subtle?” Whitehill emphasized that 3D has a danger of getting out of hand, and it should be used carefully to define a scene’s purpose and create believable worlds by not misusing or contradicting the parameters set by designers of that world. So “is 3D the enemy?” asked Whitehill. He would say that 3D is not “the answer” to drawing an audience, as the novelty of stereo will likely wear off, and when it does, he looks forward to films having compelling characters and stories again. “It will be used gracefully.” Whitehill did also mention that the short preceding the upcoming Toy Story 3 demonstrates the most unique use of stereoscopic technology he has seen to date. The movie is not due to release until mid-June of next year, so we’ll see how unique it is when that time comes.
Stereoscopic adds one element of realism, but even more important is the actual performance of the characters and their ability to emote. Chris Landreth, Academy Award winning director of Ryan, and another featured keynote speaker at this year’s event, has been working to study and develop the subtleties of realistic facial expression in his animation. He began by discussing the “Uncanny Valley,” a graph that illustrates the point at which emphatic response from the viewer quickly drops as the realism of animated characters increase. He cited a few examples of this phenomenon, such as Polar Express, and his own short film Bingo. What is it that makes the characters in these films so creepy? Landreth gave several reasons:
A time-honored, classic technique used in animation is pose-to-pose. Landreth argued that humans don’t actually move “pose-to-pose” so already the animator is setting up the characters in unrealistic circumstances. In The Spine, Landreth’s featured short at this year’s Siggraph, he was trying to push the boundaries of realism in animation even further. To do this, the animators were actually told not to use a dope sheet when planning their animation, which prevented the use of “posing.” Instead, the animators were encouraged to take an “inside-out” approach when defining their character’s movements. He suggested that they “inhabit the souls of the characters” and “do the motions to get it into the body.”
Landreth then commented on the importance of gestures in animation. “Doing nothing excellently” is important to keeping your characters real. In the mid-60s, Andy Warhol did a series of “screen-tests” in which he sat a person in a chair and told them to “do nothing.” What emerged from these tests were incredibly engaging scenes that seem to have such weight and action in very subtle movements. Bob Dylan’s screen-test was particularly inspiring to Landreth’s study of realism. “Allow your characters to just simply exist.”
“Twitching meat” is the third key to realism, according to Landreth. There is a fluidity to motion in most “creepy” animation that is extremely distracting. For example, the characters in Final Fantasy exhibited a lack of impulsiveness in their actions, when what they should have been doing is a “fast ease in” to show the brain firing neurons into the body causing it to act. This “twitching” is one element Landreth explored while making The Spine.
One often overlooked element to animation is saccade, which means “shaking eyes.” Humans, by nature, have poor peripheral vision, and therefore our eyes must continually, and involuntarily, “twitch” to create a “big picture” of what we are looking at one piece at a time. When saccade is not present in eye movement, such as in Polar Express, the eyes can look dead, which of course leads to “dead” characters. Eyes are the windows to the soul after all, and if the audience can not see life in them, then the animated character easily seems lifeless, regardless of how realistically the rest of their body is moving.
Finally, the animator must always work with honesty to create living characters. Again, citing Polar Express, Landreth explained that the audience knew they were being tricked because of various elements in the animation. There was something not quite right about the characters’ movements and it was easy to see that they were not human, even though the goal was to make them appear as living, breathing humans. Of course, the audience was not fooled, so being honest in the fabrication of the animated characters allows the audience to enjoy the performance. Exaggerating any element of the characters motion would have helped the performance aspect in Polar Express and created a much stronger emphatic response from its audience. It’s important that a character does not try to “hide anything” from its audience regarding its authenticity. In The Spine, Landreth displays many fantastical elements in the design of his characters such as a growing, revealed spine and mutated anatomy which immediately says to the audience that these characters are not “real” but act in realistic ways that allow us to relate to them. The audience knows right away in The Spine that this film is not trying to depict reality.
When asked where he thinks CG animation is headed, Landreth said he wasn’t sure, but he did speculate that CG is riding a crest of human animation and also noted how excited he was to see the work being done on Avatar. “You will begin to see people actually being fooled,” continued Landreth. He also commented briefly on the future of stereoscopic technology in film; “Each time, I just want to rip off those glasses and see the film.” In his own work, Landreth has experienced a tremendous amount of pressure to produce stereoscopic work. He did say he is interested in 3D work, but only to help in the telling of the story.
The Spine was an exploration of realistic animation done completely by hand. Digital Emily is another exploration of realistic animation, although this time created using hi-res 3D scanning and blendshape techniques. In fact, Image Metrics, the company responsible for these developments in realistic facial animation, touted that only two animators working for one week were necessary to complete the required animation. It is quite remarkable how lifelike the character appears. Image Metrics gave a course at Siggraph this year called Performance-Driven Facial Animation that explained their process in creating Digital Emily.
To begin, Image Metrics first scanned an actual actress’ facial expressions to capture realistic “key poses,” paying particular attention to details like skin wrinkling. The initial scans were then used as both reference and transformed into a 3D model consisting of about 4,000 polygons after cleaning up the mesh to allow for partial blendshapes. The project then went to rigging. Oleg Alexander of Image Metrics, described the process of creating the rig and giving the animators simple curves to manipulate. Details such as the “sticky-lips” effect were very important in creating realistic facial animation and needed to be included in the rig. Peter Busch, head of production on the Digital Emily project, described the animation process, which mostly consisted of moving between blendshapes and relied on the actress’ performance more than hand-animation.
Another production relying heavily on performance-based animation for an entire character, not just the face, is the upcoming Disney’s A Christmas Carol. Imagemovers Digital has been hosting an exclusive “Train Tour” that began in Los Angeles and is making its way across the country, stopping in various cities to allow a behind-the-scenes look at the studio’s work on the film, due to release this December. The train tour made a special appearance in New Orleans, just in time for Siggraph 2009. After a shuttle transported the guests to a waiting tent with carolers, hors d’oeuvres, and drinks, they were then invited into a room to watch the behind-the-scenes footage. A trailer for the CG film was shown following the breakdown of the production process. One striking difference between this animated film, and other films, is that the performance of the character was allowed to exist, silently, for a considerable length of time. This often is not seen in animated films, and although motion-capture technology was being used primarily to create the animation data for this particular film, the fact that an animated character is simply existing is certainly a step in the right direction for animation and CG as a whole. The film looks beautiful and engaging and is sure to be a hit with audiences this holiday season.
After the preview, the audience was invited to board the train that has been touring the country. Displayed amongst a number of train cars were various concept and costume designs, as well as actual historical reference used to inspire and assist the production team. Moving on, were screens with key players from Imagemovers Digital describing their part in the process of creating Disney’s A Christmas Carol and the particular accomplishments and challenges faced. One car focused on the motion-capture technology used and another car led the guests to a room allowing them to morph their own face into one of the film’s characters. Overall, it was a very impressive promotion for what will be a film to push the expectations of filmgoers and the CG industry alike.
Siggraph 2009 was certainly smaller than the Los Angeles events have been but the Computer Animation Festival was no less beneficial and entertaining than last year’s. The winners of this years Evening Theatre presentations were French Roast, by Fabrice O. Joubert of The Pumpkin Factory in France, which took home the Best in Show award. The Jury Award winner was Dix by BIF Production and The Mill of the United States. Alma, by Rodrigo Blass and Cecile Hokes of Spain, received the Jury’s Honorable Mention award. The Student Prize winner was Alpha created by Matthias Bjarnason, Christian Munk Sorensen, and Nicolai Slothus of The Animation Workshop in Denmark. Finally the Well Told Fable Prize went to Unbelievable Four, by Sukwon Sin and In Pyo Hong of the United States. These films as well as many included in the festival exemplified narrative animation in new ways that made the judging process very difficult for this year’s jury.
Stereoscopic technology, as well as advances in realistic animation in both films and games, has made considerable progress and impact in the past years and was absolutely the major draw at this year’s Siggraph. From games like the new Fight Night 4 to films like the upcoming Avatar, audiences can expect to find themselves unable to distinguish from CG and reality. This is an exciting time for visual effects in entertainment and will certainly continue to evolve as the technology and inspiration to share new stories grows.