ABSTRACT
VFX is a competitive industry where studios and developers are compelled to tout their business as “cutting edge,” “leading,” “revolutionary,” and “state of the art,” yet most in fact rely on the same existing and widely shared technologies. The real innovations often happen in research labs and are published in obscure scientific papers, far away from the glitter of showbiz. While the term artificial intelligence comes bundled with the inevitable mix of hype, fantasy, and popular misconceptions, machine learning feels like a somewhat tamer (and more accurate) way to describe a recent transformation that is all but tame. Machine learning algorithms and neural networks emulate (to a certain extent) the way the authors acquire knowledge and experience. A baby learns what a cow is by looking at a drawing or a photo and subsequently identifies cows in other drawings or photos by association, eventually forming a comprehensive visual concept of a cow.
