Whenever I start playing around with a new AI tool, I always start with this image of myself at age 6. What began as a fun way to troll my child self became a clear baseline - the single constant in a sea of generative variables - allowing me to easily compare tools.

Midjourney has an image retexturing mode that lets you reskin an image while keeping its general shape intact. I fed it my pic and just started typing prompt after prompt. I used really simple ones. “Blue”, “Wet Spaghetti”, “Screaming” … these all work well. I blacked out for an hour and noticed I had 984 images, so I made 16 more and figured that was enough.

With 1,000 variations on my soft lil face, now I wanted to see them in motion. I made some selects and used Runway to capture short performances using my own face. I chose an old favorite meme-track (Shooting Stars) for its emotionality and immaculate vibes, and timed the sequences to the beat in AE.

This exercise clarified my hunch that a deep understanding of the capabilities of AI tools is essential in not just understanding HOW to use them but WHY to use them in the first place. And, it will always take skill, experience and a point-of-view to make anything of quality no matter how smart any models get.

Also, adding in a touch of humanity (in this case, some simple facial capture) really matters if you want to create AI art that connects with people. As a redditor commented on my video: “It made me feel like my time isn’t over yet, that I can still change, be something else, unbound by the image of oneself.”

———
This video took me under a day to make. I spent zero time refining prompts or color correcting. Subtracting the time spent waiting for generation, actual working time was around 5 hours. I’m usually quick, but this was shocking to me. To create something equivalent would have taken weeks if not months pre-AI.