From artist Hito Steyerl, I learned to call AI-generated images “statistical renderings”—images that, by nature, are VERY average. To my eye most of them feel like frozen, lifeless stock images. And yet, there’s still something magical about conjuring an image from language, like writing or speaking a dream into existence.

But for me, that’s rarely the end of the process. It’s more often a starting point.

My Generative Portraits begin with Stable Diffusion, but they come alive through custom tools I build in Processing. These programs let me distort, animate, and repaint each image with what I think of as autonomous paintbrushes—like the broomsticks in Fantasia, they take on a life of their own. As those paintbrushes do their magic they generate thousands of images that can become animations as seen in the first four entries below. That wealth of material also offers so many choices to pick definitive still frames as seen after those videos.

This isn’t about perfection or realism. It’s about restoring gesture, play, and presence to digital portraiture—bringing my hand, eye, and art history into the loop. There’s a lineage here, too: the luminous textures of Klimt, the methodical layering of Chuck Close, and a deep curiosity about how artists can collaborate with their tools, even when those tools are made of code.