Deeper into AI

I recently read about Darren Aronofsky, who is now also building an AI company, Primordial Soup, and using artificial intelligence in his film projects. Generative AI has clearly come a long way when such a respected director is investing in it and even founding a company around it.

I’ve also started creating my own short films partly AI-assisted. I’ve been making shorts for a long time using traditional methods and I really enjoy developing stories. My background in VFX, motion graphics, and music composition is a combination that now allows me to create films on my own — and maybe even stand out a little because of it. I’ve been using AI tools for years, sometimes more visibly, sometimes simply as part of the workflow integrated into other tools, but also independently for concept development and building specific elements. A couple of years ago I was involved in producing the first commercial AI project I know of, and it was one of the most labor-intensive projects I’ve worked on. Since then, things have advanced significantly.

To support new productions of my own, I started Rändöm. Rändöm is a production company that releases 1–2 short films a year, or other shorter series. Its main purpose is to act as a platform for my own creative expression, separate from my work at Stringular. The stories of Rändöm often hover on the edge of the uncanny—worlds that resemble our reality, more or less. In a way, it’s like my personal Black Mirror development lab.

At the moment, I’m working on an animated short and a three-part series. I’m also writing a new project. The series combines AI and visual storytelling, partly with live actors. Right now, I’m exploring the best workflow for this. The goal is to use AI in such a way that it doesn’t look like AI. The toughest challenge is, of course, AI’s weakness: consistency. It shouldn’t distract from the story. In the process I use a range of tools, like ChatGPT, Midjourney, Runway, Seedance, and Topaz. Some environments are built in 3D, and then I apply style transfer on top. And just when I find a process that works, a newer and more efficient way to do the same thing appears. But as I said, the story is always at the center. Whether AI draws attention away from the story or somehow prevents deeper storytelling remains to be seen. For those interested in writing stories and building worlds, I can highly recommend a great book on the subject: Into the Woods by John Yorke.

I believe AI-generated content is here to stay in video as well, but it won’t destroy traditional filmmaking. Streaming-like services dedicated to AI content are already starting to appear. Right now, AI video is still a kind of fast-food entertainment — you can usually spot it quickly. And it’s worth pointing out that this kind of content is not something you can just whip up in a few hours. It still involves a lot of writing, VFX, and other technical design on top of prompting. And it’s not exactly cheap to make either. Still, it’s fascinating enough that I think it’s worth exploring.