Whoever resorts to artificial intelligence to conceive a story, from the scripting to its delivery, is a jackass with no imagination. Can the same be said about those who use image-generating software like Midjourney? If so, send me straight to the guillotine for putting more graphic designers out of a job.

This criminal endeavor began late last June when I set a new life goal: produce my first video story using Midjourney’s AI image generator to fill the visual details of a horror tale I have had in mind for a while. It took me about a week to write the script, which wasn’t that difficult to scribe as it was mostly narration. Sure, OpenAI could have penned it in a few seconds, but it would have read like an overwritten prose typed by an erotically frustrated Harvard academic.

After recording my voice and cleaning up that sound track, the real work began, which basically consisted of yelling at a non-sentient machine when it behaved like a government employee who obnoxiously refused to execute what I requested. To be fair, I was somewhat inexperienced writing prompts that only a soulless program would understand, so I had to suck it up and learn to work around this silicon-based system.

Since 2019, I have published a few audio-only episodes in my Fiction Madness podcast. Although I had great fun producing them, fiction in the podcast world isn’t a coveted category like comedy or true crime. The stats place them way down in the pit, so close to the abyss it almost reaches Hell. I wasn’t discouraged by the data. On the contrary, it compelled me to find new ways to make my stories more available to people who still appreciate storytelling in visual platforms. There was just one major problem.

Relying on existing, copyright-free images like in Pexels, or downloading them for a subscription as in Envato Elements, wasn’t going to provide variety or continuity while telling a visual story. Also, I’m too cheap to hire a graphic designer living in India or Venezuela to design comic book-style illustrations. Fortunately, the years 2022-23 witnessed the rise of artificial intelligence and image prompt writing. Having already experimented with AI-generated voices, it was time to exploit this emerging, epoch-making tool.

I subscribed to Midjourney for the first time back in late March 2024 after feeling completely dissatisfied with OpenAI’s image-generating feature. With so many systems like it exploding in the market, I wasn’t going to jump from one platform to another to see which one fit my creation needs better. Midjourney was, by far, the one with the most positive reviews, and based on samples demonstrated across various YouTube videos and blogs, I spent ten dollars to explore their basic subscription.

This was the very first prompt I wrote using their basic-level subscription:

“In a kitchen, beneath the table, a large mound of disgusting mashed food and liquids, with a small arm poking through the mound.”

The result:

Although the prompt failed to place that mound beneath the table as I requested, the outcome was still mind-blowing. I made a few more tests, and the software produced similar images. Doing so made me realize I had an opportunity to make a short suspense story blending images with music, limited character voices, and sound effects. Between then and June, The Boogeyman of Augusta was born.

I could have chosen any story to produce. In fact, I could have picked an existing podcast story of mine and add images to it, but my gut instinct told me I should get crazy with an original tale that is more genre-oriented. Why suspense with a touch of horror? Not only do I enjoy putting people on the edge (it is an irreparable fantasy of mine), but also because I believed it was the one category that would allow me to discover how far I can go in the use of language, writing prompts without being flagged for inappropriate keywords.

On a few occasions, Midjourney’s surveillance bot cancelled some images before they were generated, as the system rendered the language prohibited. Those words were “little girl,” “vomit,” “blood,” and “naked.” Before you phone the F.B.I., let me remind you that I did not use those words in the same prompt! They were all typed separately, on different days, for different descriptions. If I had, you wouldn’t be reading this article, and my face would have been broadcasted in the crime segment on some news outlet.

Right then, I knew I had to be extremely cautious if I wanted to produce visuals that didn’t give their servers reason to have agents knocking on my door. Example: for “little girl,” I had no issues writing “three-year-old girl,” or any other age. For “blood,” that worked out typing “dark red paint.” For “vomit,” the words “mashed” or any other non-repugnant synonyms. As for “naked,” the words “shirtless man” worked when generating a thumbnail for my podcast story The Idiot Who Lost His Legs. I haven’t tried “shirtless woman” yet. Feel free to explore the consequences and then let me know in the comment section below.

While editing The Boogeyman of Augusta, I worked the story scene by scene, meaning that I would generate images for every number of seconds while also fetching the appropriate music and sound effects. It may not be the most organized way to produce a video, but it works for me, not necessarily for others. As always, the editing process took most of my time for the better part of July and mid-August. Here is a glimpse of the process (click play):

Many content creators are already using AI-generated images to tell whether a real or fictional story, and this craze is just starting. This technology is expected to get better over time, and as someone who can’t draw or design worth a crap, artificial intelligence has given me a few necessary resources to enhance the telling of my stories. Nothing more. The ideas, the characters, the settings, the drama–those things are entirely of my conception. These A.I. tools are merely that: tools, not the primary source of an idea or the entire production of a story.

I guess the days of making audio-only stories are over, and I’m perfectly fine with that!

The Boogeyman of Augusta is available on my YouTube channel and my podcast Fiction Madness on Spotify. Click play and enjoy!