Haller

Description

What happens when a body is evaluated against a space that was never designed to accommodate it? Confined within its limits, the figure compresses, adapts, and disappears where it can no longer contain it. At first glance, this work presents a woman compressed inside a narrow, concrete enclosure. Her body is constrained, her movement limited. The scene appears plausible, almost photographic, yet something in it does not fully settle: the figure does not break the space, nor does the space yield to her presence. Instead, the body is forced into adaptation under constraint. What should be impossible becomes normalized. Her pose reads as endurance without overt reaction. The work points to how AI resolves contradiction: when space cannot expand and the body cannot fit, the body is altered. The result is a “complete spatial” generated task on account of an incomplete body. The image resembles a photograph that cannot exist in conventional photography. The story here is: “How long can this be maintained? And what will keep on happening?”

Process

The work began as a visual study of scale: a large female body lying inside a narrow room, seen from a low angle using foreshortening. I wanted to exaggerate the mismatch between body and space, and to place the figure in a position where fitting inside was physically impossible. As I developed the prompts, I noticed a recurring pattern. When I used the phrase “enormous woman,” the generated figures repeatedly appeared as heavy or full-bodied women. This was not what I meant. The size I was referring to was spatial, not a specific body type. Yet the images kept returning to the same biased interpretation: “enormous” was translated into weight, rather than scale. Refining the prompts, another behavior became clear. When the body could not fit the space, the image did not hold the misfit. Instead, it resolved it. The leg shortened, stretched out of the ceiling, disappeared, or was cut—so the figure could remain inside the room. At first, I tried to correct these results. But the more I resisted, the more I hit the wall: the algorithm suggested adaptation at any cost, even at the cost of true human form. Something different happened in the last rendition: The diagonal layout and volume were powerful, yet the quiet counterbalance of the woman’s stillness was unexpected. It was as if her compression no longer registered as a problem. This outcome was far more compelling and thought-provoking than my original intention, and I decided to follow it.

Tools

I worked primarily with text-based image generation models, developing prompts that described a monumental female body confined within a narrow architectural space. Early outputs functioned as sketches: they established pose, perspective, and spatial tension, but lacked precision in proportion, lighting, and depth. As the process continued, I refined the prompts to emphasize scale, foreshortening, and spatial compression. I also introduced reference images to guide the model toward the specific physical configuration I was aiming for. Despite these controls, the algorithm repeatedly altered the body to resolve the spatial conflict—shortening limbs, cutting forms, or reshaping anatomy to fit the enclosure. At this stage, the process shifted from directing the model to observing its behavior. Rather than forcing corrections, I began to treat the outputs as responses to a problem: how the machine negotiates constraint. The creative process itself, including AI’s insistence on correction and completion, became inseparable from the meaning of the work. Since only a single image could be submitted, I am including a link below that provides additional context and allows the full series to be reviewed: https://www.instagram.com/p/DT3_H2gD8qc/?img_index=1