Uncanny Valley
January, 2018
In 2017, Computer Generated Imaging has all but conquered the uncanny valley, a place where images are fabricated with technology but look almost real. The general consensus in the field of CGI is that it’s possible to make anything with software. In exhibits such as Bladerunner 2049 and other recent blockbusters, the various races to hyper-realistic computer graphics have been won, so where does CGI technology move forward from here? Now that we can make anything, what do we do?
As artist and filmmaker Alan Warburton points out in the short documentary Goodbye Uncanny Valley, the history of technological advancements in computer animation is short so far. In fact, Warburton suggests the genealogy of the technological advancements in CGI rendering and the production of digital images for massive scales only goes back to the mid 90’s when animation studios like Pixar produced steadily better qualities of CG animation with each year’s blockbuster. Parallel to this note, increasing demand in computer-generated animation in films drives technicians at companies like Autodesk to invent new features and workflows to optimize software for CGI. For example, in ABugs Life(1998), one of Pixar’s earliest releases, animators developed tools to animate and render vegetation. Monster’s Inc., (2001) another Pixar film, featured advanced fur simulation workflows, while releases like Tangled (2010) by Disney Animation Studios and Brave (2012), also a Pixar film, indicated advancements in hair simulation and rendering. Now in 2017 animation studios producing special effects for CGI within live action movies can simulate and realistically render virtually anything, so much so that viewers don't know, or rather forget, they are looking at computer generated images. The phenomenon of a human not knowing an image is fake or computer generated designates the uncanny valley, the place where computer generated images appear real.
The same phenomenon exists in artificial intelligence, except for in this instance it designates a phenomenon where a human thinks they are interacting with another human, but doesn’t know they are interacting with AI, specifically defined in the 1950’s Turing Test. This study refers to the practice of testing an AI with a human to see if the human can tell they are speaking to AI, or inversely, are convinced they are speaking to another human when they are in fact speaking to a computer. A classic example of bug-out uncanny valley is Russian world champion chess master Garry Kasparov’s loss to IBM’s chess computer Deep Blue in a 1997 tournament. At one moment in the match Deep Blue sacrificed a piece to throw off Kasparov’s concentration, and as he became fixated on the possibility that IBM was tampering with the computer to force it to make irrational moves, he suffered a psychological bind that resulted in losing the match. For a moment, Kasparov could not figure out if IBM was, behind the scenes, directing the computer’s moves or if Deep Blue had made a strange, possibly accidental, move on the board. Another example, although fictional, of the Turing Test in action is Alex Garland’s 2014 film, Ex Machina where an unlucky man travels to a remote laboratory to engage in the Turing Test with a beautiful female robot and later dies at the hands of the robot in a Shakespeare-meets-Pygmalion tragedy.
As we look to the uncanny valley in CGI, movies in the first and second decade of the 21st century, like Ex Machina, generate novelty through technical advancements in specific areas of CG animation, such as light simulation, fur, hair, water, explosions, skin, etc. But as animation groups continue to conquer these territories, they must find new territories to conquer; they must advance the capabilities of CGI even further. In Warburton’s words, “what was novel about CGI was seeing it advance with each new movie release,” but as we conquer the uncanny valley, the film industry must find new ways to innovate and keep moviegoers dazzled.
There are three categories, Warburton argues, in which cinematic CGI continues to move forward as it ventures into the uncanny valley; these categories are commerce (the price of the production), construction (referring to the scale studios can achieve while maintaining the resolution of hyper-real CGI), and last is consolidation (referring to how quickly the production can be done by using cutting edge technology to streamline workflows and expedite loads). In 2017, since the question is no longer about the how, e.g. “how can we make this water look as real as possible…” “how can we make this hair simulation appear exactly like Princess Leia’s ear-muff hair from Star Wars: A New Hope…,” new questions revolve around how impressively massive and outrageous a scene can be, how fast it’s produced, and how few resources it can be accomplished with. There is no longer a technological race to the hyper-real; today we contemplate a contested economy of scale, which means more manpower and technological innovations for achieving these incredible scales while maintaining detail and resolution.
Anyone who purposefully misuses and exploits software developed for the film industry should be very interested in all of this. Programs like Maya, Cinema 4D and numerous plugins can be found scattered across the internet like scraps of food for hungry independent artists to make their work with. The architecture student finds himself or herself as an individual among other artists in an emerging field of post-cinematic digital art, which Warburton designates as “The Wilderness.”
“Goodbye Uncanny Valley,” Still, Alan Warburton, 2017
Work in The Wilderness processes graphics in alternative ways far inferior to the realistic CGI appearances of major blockbusters. These artists expose raw CGI flaws, less than accurate textures and materials, and constrained scale. Artists like Pussykrew, Cool 3d World, Fvckrender and Andrew Thomas Huang expose raw CGI and the imperfections that come with it. Such artists embrace these flaws as part of their artistic language; less than perfect CGI becomes part of their cinematic universe, and like the major animation houses with their proprietary render engines, material libraries, and endless stock assets, these fringe artists mass their own language through reuse and abuse of plug-and-play models. One primary reason such artists struggle to produce such realistic CGI could be because life-like computer simulation isn’t probable without the assets of a major animation house, and a massive taskforce of faceless CGI technicians.
However, we could argue artists in The Wilderness expose CGI flaws on purpose to reveal some truths about digital imagery as a discipline. For instance, work by these artists tells viewers where labor gets invested in the economy of labor in the animation industry. Because of advancements in software, as Warburton points out, some visual effects require only a small dedicated team to execute, while other effects like hair simluation in Pixar’s Moana (2016) need larger teams of people. This is because software has been optimized to perform well in certain tasks requiring few man hours, while other tasks like roto-scoping, still require man power and time. For Wilderness artists, lack of labor force can mean jarring and imperfect camera rigging; the finer details of texturing and material shadings are overlooked. Objects intersect and normal orientations flip, revealing sloppy, erroneous geometries. These flaws demonstrate the complexity of software and give some evidence that many tasks require dozens of animators to work through to erase the artifacts of technology used; this is precisely where fringe artists speak to the discipline. They embrace a sympathy for the finitude of their capabilities and push what they can do into new speculative grounds; they reveal to viewers exactly how the technology relates between singular fringe artists and huge studios, namely that the expensive resources are spent on man power and outsourcing work to Canada, China, India or elsewhere to cushion expenses.
My argument is that The Wilderness is compelling. It’s extremely intelligent, self-referential, and funny. It’s rich in appropriation; fringe artists reference pop culture and art in their work. One could argue that Dada, the turn of the century art movement, for all its randomness and absurdity, is one art form that fringe artists parallel. They attempt to undermine the seamlessness of the contemporary computer-generated images by exploiting the potential for errors in technology in the form of careless bending of geometries in unfamiliar, funny, or disturbing ways. Another is to disrupt logical continuity from scene to scene in editing, in which one scene transitions randomly into another one with no shared content, much like a dream’s tendency to transition from one unrelated topic to the next, jumbling imagery in a random psychic clutter that makes no sense in retrospect but by some undisclosed invisible thread of psychic, still somehow stays connected. This is how many fringe artists connect with Surrealism, the child of Dada. With these methods, fringe artists make new Dada via random disconnected relations of impactful silliness or new digital surrealisms via dreamlike content.
Dada is in a way a reaction against the madness of World War I; could we look at the work of fringe artists as critiques to the seamlessly connected hyper-reality of big studio CGI? Both create cinematic universes, but where movies create temporal connections of events and emotions, which manifest as tropes like “plot” and “story,” artists in The Wilderness seek to unravel and jeopardize that continuity by distorting the same strategies used in storytelling.
“Poposal for a Public Park” Still, Jackson Lukas, 2017
We find this tendency in Architecture too. My own work attempts to envelope the viewer in its own cinematic world; it’s a world of fantasy and dream, one dominated by dismembered nude figures and strange growths. The work proposes a public park in Rosario, Argentina. The large inhabitable sculptures are secluded in a field of grass the user must walk through to reach the project. The project proposes a material that has the surface-scatter properties of skin yet seems to be melting in a constant stage of mutation and evolution and in this way, creates a material we can only visualize through rendering.
The project surveys the nude, or the culture of nudity; it tries to appropriate the tradition of nude painting and sculpture as the quintessential study for these two respective disciplines, as first memorably proposed by Michelangelo who said that the greatest subject for painting was the male nude. The park in particular takes several classical sculptures of Venus, a subject of many paintings from the Renaissance and the Baroque, to our current era, and mixes her with other nude figures for formal exploitation.
The project demonstrates that despite attitudes in accademic architecture that fascination with technology will wane, the individual’s creative power will continue to increase as sophisticated software becomes more and more available to artists. Furthermore, as new tools become available people will seek out ways of exercising whatever they want through these powerful mediums, as they always have. This is very important because the relationship of the creator with their software will move in favor of the artist’s imaginative will as technology continues on its entropic course.