Every now and then, I get on a new kick with AI image creation. It usually starts with a single spark—a seed image—and turns into a multi-step journey through a handful of prompts before I land on the finale. Sometimes that seed image is an old family photograph from over a century ago; other times, it’s a snapshot from just last week. It might be a quick screen capture off my computer or phone, an abstract painting brought to life with a realism prompt, or simply a word-salad string of instructions that evolves as I watch the generator respond.
Early on in my AI adventures, I leaned heavily into the classic fantasy realm—barbarians, orcs, elves, gnomes, dwarves, satyrs, and imps. I’ve always found my personal aesthetic favors the female form, but to each their own. These days, I’ll usually take one of those earlier creations and steer it in a fresh direction using a new set of prompts.
Of course, one of the quirks you learn to live with is when the AI suddenly puts its foot down and tells you it can’t do something:
“I can create images of real people, but not one like that. Can I help with a different image of this person?”
The amusing part about that message is that the person in the picture isn’t a real person at all—it’s a completely artificial creation that the AI generated itself just minutes prior! The exact same prompt worked fine on a different image, and if I hit regenerate once or twice, it usually goes through without a hitch. It leaves you wondering: is the AI-generated subject so uncannily lifelike that it fools the system’s internal sensors into thinking it’s looking at a real person, or is the censor just overzealous?
I even had a video generator flag a newly created clip for community guidelines because of something in the audio track—even though I hadn’t requested any specific audio to begin with. Go figure.
Anyway, long story not so short, I spent a good chunk of the past few days having fun with these workflows and put together the following video:
Each sequence in the video was born out of a layered, iterative process. First, I dug through my archive for an image where I liked the main subject. From there, I’d tweak the subject (or leave it as-is), then completely rework the environment and lighting—all grounded in a hyper-realistic style. Then came a prompt to transform the aesthetic into an Art Nouveau style, just to shift the feel of the image. After that, I’d run it back through a hyper-realistic, highly detailed render with natural lighting before feeding the image into the video generator to bring it to life.
All in all, it’s quite a process—full of fine-tuning prompts, regenerations, downloads, and careful inspections before I’m finally satisfied with the result. But as always, there’s a real joy in watching a simple seed of an idea evolve into something entirely new. Pour yourself a fresh cup and take a look.
