Hand-in-Hand: Machine-Assisted Media Generation

To start things off, the foundation of machine-generated content lies in a subject area that encompasses a large part of this class: machine-assisted generation of content. While it could be argued that all machine generated content is, in fact, just “machine-assisted generation”, in the context of this work, content that is generated through “machine-assistance” is loosely defined to be content created using a machine such that all “creative decisions” made in the creation of the work are done by humans – machines only serve to expedite or enable the process. For most of human history, this has been the extent to which machines are able to generate media: in the case of Walter Benjamin, think of a machine that merely copies an existing work onto a new canvas; in the case of Copyright Criminals/RIP: A Remix Manifesto, think of the computers and DAWs that allow for the electronic mixing of music into new compositions – and the list goes on. Essentially if you picture what Adobe has built a large part of their business model around (Photoshop, Premiere Pro, After Effects, Audition, Illustrator), you’ve got the main idea (even if non-electronic machines, like the printing press, can fit into this category as well). We spent nearly a quarter or more of the class talking about the potential implications of this style of content generation, but it’s still worth noting here if only because of how foundational it is to the topic at hand. Overall, the main cycle of development in this area has been the back-and-forth battle of enabling more creative functionality while trying to make that functionality as accessible as possible. For example, in the context of photographs, we started simply with the ability to copy photographs (ignoring tricks in the darkroom with regards to masking, burning, and dodging), and then, when we progressed to the digital age, we could modify images at the pixel level – an incredibly granular level that offers a great deal of precision, but it also required a large amount of effort to perform substantial transformations. Thus, we made the tools themselves more accessible by delegating more control to the machine (i.e. Photoshop/Paint.NET filters, patch tools, filler tools, etc). In fact, this engine of progress in the form of “enable more capabilities; make those capabilities more accessible” can serve as both a guiding principle in the development of media generation and as a segue to the development of our many other areas of content generation. Now, this engine, being more cerebral than corporeal, has many ways of being “implemented”, so to speak. One of the key methods that we’ll talk about next is through the “trivial” implementations of machine generation systems.