Model-generated images / From photograph through model

Hermit

Beginning with a photograph made in rural Andhra Pradesh more than two decades ago, Hermit follows one figure through a generative model. The body persists as ornament, machinery, and cultural reference are recomposed. Authorship moves among the photographed encounter, the model, and the act of selection.

An ongoing study / the body over time

From Lived Encounter to Simulacrum

I keep coming back to the 2024 images. They feel closer to the man I photographed in Andhra Pradesh, and somehow more authentic to me. I can say that with some confidence and still have trouble saying why. With these models, so much of what happens between the prompt and the image is opaque. I know when something strikes me as tasteful, or when it just feels right. It’s much harder to say what the model did to get me there.

I almost want to call some of these images endearing, though I’d use that word carefully. A few go straight into body horror. Hands turn into things I can’t quite name. I wouldn’t have asked for that. But they came out of a particular moment in the life of these models, and I can’t go back to that moment. The oddness is part of what the images hold onto. They’re pictures of the Hermit, and in a way they’re pictures of the medium itself.

Earlier Hermit cyborg interpretation with ambiguous fingers in the hands resting on the floor — Look at the hands resting on the floor: finger shapes and joints are difficult to separate.

2026 interpretation of the original seated hermit photograph, with brass mechanisms and clearly separated fingers — Distinct fingers, knuckles, and wrists give the hands a more readable structure.

2026 Hermit with the same eyepiece, arm mechanisms, and layered jewelry in dimmer, softer light — Lower exposure and softer contrast give the brass and skin quieter highlights.

I notice the hands first. In the 2024 image I can’t quite work out where one finger ends and the next begins. In the 2026 image I can follow the fingers, the wrists, and the bits of hardware that sit over them. In the second version I took the exposure and contrast down. The brass catches less light, and the room feels a little more like somewhere you could sit.

And still, when I put them next to each other, the older image feels closer to the dark, austere atmosphere I remember from the original photograph. That’s what makes me think of Baudrillard’s simulacrum: I can make the image more convincing and still feel myself getting further from the encounter. I haven’t tried this with today’s diffusion models, and the prompts changed too. This is what I see in these particular images.

2024 series / Transformations / 7 images

All collections