AI and Filmmaking

(This post will cover a bit more than just screenwriting, but AI is a quickly growing technology and its integration with filmmaking is not a matter of *if* but rather *when*. Thus, we discuss.)

While it's still too soon to tell exactly how AI will shape the future of movies, the increasing integration of this technology in the film industry raises important questions about the role of human creativity in filmmaking. As AI becomes more advanced, will it eventually replace human writers, directors, and editors? Or will it simply provide new tools and insights for filmmakers to enhance their craft? Exploring these questions will be crucial in understanding the potential impact of AI on the art of cinema.

For those unaware of what AI options are out there, or how they're being used in films, here's a quick course. There are basically two big AI elements out there right now: Text generation and Image generation. Text generation comes from sources such as ChatGPT, which is a program that can interpret what you type in and output a response in what it considers to be a logical string of words. It's pretty good at this. There are plenty of natural-enough-sounding conversations you can have with the program.

The image generation side of things is a whole other beast. There are programs that can interpret and learn what certain objects are in images, there are programs that can take different textures or images and essentially paste it over footage, and there are programs that can produce "original" works of art. More later on why original is in quotes there.

While AI like this is still fairly fresh in the grand scheme of things (Some of these programs require a bit of computer and coding knowledge for the best results) that hasn't stopped people from attempting to push the boundaries of what it can do for film. A recent example of this comes from the channel Corridor Digital. These are independent filmmakers who recently sparked a big wave of discussion with a video they released. In this video, they took footage filmed on a green screen and used AI image generators to essentially redraw the footage in the style of a reference anime. It's a fairly long process, but I'll post their series of videos where they do a good job of explaining:

The original video:

The same video with their original footage:

Their explanation video:

If you watch the video(s), you can see that what they've done is objectively fairly impressive. It's akin to rotoscoping, but done with computers rather than people. They obviously have done a lot of other work on top of just what the computer did. All it did was make them look hand-drawn, but they added the other effects and such. While I think this was pretty cool, not everyone, understandably, felt the same way.

The problem with AI as it relates to more artistic endeavors comes with its ties to originality and intent. Text generators have no intent. They can write a poem, but that is just strings of code telling the program what a poem looks like. Image generators face issues with copyright, as many of them simply feed off of other art that they find and have been trained on. There are some major lawsuits being put in motion. One of these is being led by Getty Images, a major image hosting site on the internet. In that particular case, people noticed that the Getty Images watermark started to appear on AI-generated images. The program had been trained on enough watermarked images that it started to believe this was a regular piece of art and thus could be included.

This last point of where these images come from is where most of the contention lies with Corridor's video. They used an anime that they love the style of and trained their software to emulate that style when creating their anime versions. This has sparked debates on what the differences are between the computer doing this, versus say a person just trying to emulate that style. Corridor has come out in defense of their video essentially saying that they were just excited that this idea worked, as it cobbled together several separate programs over months of hard work. They say that, in the future, they would use what they learned from this experience to train their software off of their own, original artwork.

And that makes sense. Once this is a solid and understandable workflow, the AI can just be turned into a tool used to make films. This is undeniably the future of the industry, as artists and major corporations both see the new possibilities at hand. This has, of course, riled up many animators and artists for reasons beyond the copyright issues. They're afraid of losing their jobs to automation. It's the same idea as a factory worker losing their job to a machine. What's unfortunate, is this fear isn't going to stop the march towards AI integration. History is full of jobs being replaced by newer technologies. This being a creative field, there's an argument for there always needing to be a human touch to make something into true art.

Like it or not, this is where things are going. For me, I'm going to be doing my best to figure out how to get ahead of this curve. Not with screenwriting, as I like doing that work myself, but with short films I'm hoping to make. I'm already working to figure out how to use these AI tools. In fact, the opening paragraph of this post was written using ChatGPT.

---