The problem of realism in animated characters: has the Uncanny Valley been crossed?

When computer generated animation (CG) emerged in the early eighties, and because of technical limitations, the creation of animated characters that fell into the realm of what is known as the Uncanny Valley was not an issue. Early examples such as Lucasfilms The Adventures of André and Wally B. (1984) or Pixar’s Tin Toy (1988), the first CG animated short to win an Oscar, portrayed characters based on simple geometric shapes, which inherently lent them a cartoon-like appearance. At the beginning of the 21st century, however, CG technology had evolved enough in is multiple technical aspects (rendering ability by ever-faster computers, textures, lighting, and even features such as the simulation of hair or cloth) as to allow for animated characters to become closer representations of humans. A frequently quoted milestone in this respect is Final Fantasy - The Spirits Within (2001), a film in which the characters, albeit not based on specific human actors, look eerie and move a little too smoothly, placing them within the constrains of what the Japanese roboticist Masahiro Mori defined as the Uncanny Valley. We argue that in the past decade, animated feature film directors have adopted a different stance, learning from the relative box-office failures of films such as Polar Express (2004) or Beowulf (2007), and readapting CG technology to create characters more in keeping with the visual language and movement syntax of animation. As such, we expect that this drift away from the Uncanny Valley will continue over the next years, leading to the appropriation of cutting edge CG technology by animators and directors by acknowledging its possibilities and limitations, rather than by falling victim to the problems it can cause in character development for animation.