“Imaging After Photography”, Brian Droitcour – E-flux

https://www.e-flux.com/criticism/6782397/imaging-after-photography

Moody Center For The Arts, Houston
January 23–May 9, 2026

We still think about images in the terms set by photography, even as the technologies that now produce and interpret images have moved on. Photography fixes the world in discrete, framed moments. But computer systems model dynamics rather than capture them. Machine-learning models do not encounter images as stable things at all, but as arrays of numbers to be continually recombined and re-weighted. And yet we tend to recognize AI through the still images it generates. We grasp it through the remnants of a photographic regime that no longer adequately describes how images function.

“Imaging After Photography” takes this mismatch seriously. Rather than staging AI as a generator of novel pictures, the exhibition asks what images are becoming as photography’s epistemic dominance gives way to a new kind of machine vision, and what it means to see and know the world under these conditions.

The precise focus on modes of vision rather than styles means the exhibition gravitates toward photography’s most fundamental, historically loaded subjects: landscapes, natural forms, and human figures. The seven artists here approach photography not so much as an artistic medium but as an instrument of classification and knowledge production.

Trevor Paglen examines the paranoid logic of the surveillance state, exposing how machine vision systems label bodies through the misapplication of stereotypes and taxonomies. Refik Anadol mobilizes large-scale datasets to produce simulations of nature that possess their own algorithmic sublimity. Lisa Oppenheim pushes photography simultaneously backward and forward, treating it as a malleable, historically situated technology whose meanings shift with its material supports. Grégory Chatonsky works with training sets to probe the unstable relationship between images and their semantic descriptions.

Several artists focus explicitly on photography’s classificatory uses. Sofia Crespo and Joan Fontcuberta revisit its role in biology, while Nouf Aljowaysir turns to ethnographic photography and the Western impulse to make the rest of the world legible through images. Aljowaysir applies contemporary computer vision models to photographs of Saudi Arabia taken by the English explorer Gertrude Bell, alongside Orientalist portrait photography drawn from the Getty Museum’s archives. After using an AI segmentation technique to remove human figures from the archival photographs, she uses these altered images in a training set to generate “Ancestral Seeds” (2025), in which ghostly presences emerge amid rippling grayscale artifacts. The machine’s vision becomes legible by how it treats the scene of photography as a set of zones of probability, with walls dissolving into honeycombed patterns of light and dark.

Alejandro Stein and Frank J. Mondragón’s exhibition design is unusually integral to the exposition of the artists’ concerns. The gallery is divided into six equal sections, each given to a single artist, by frames stretched with nylon that function as low-tech monitors. (Paglen’s work is displayed in the adjacent lobby.) Arranged in triangular configurations and visible from the center of the gallery, the partitions form a room-size panopticon that raises the specter of networked surveillance. More importantly, the design literalizes the condition of artists working with contemporary technologies: each confronts inherited structures and protocols while attempting to articulate a singular vision within them.

The partitions let artists present images not as isolated objects but as ones embedded within processes. Chatonsky uses his to display his training sets as grids of tiny, barely legible images. The center of his section is occupied by an armature of cubes, evoking Sol LeWitt’s modular sculptures that model reason veering into obsession. A slideshow of dataset images plays on a monitor laid horizontally on the floor, emphasizing their rootedness in the world. At eye level, a vertical screen hanging on the cubes presents frozen frames generated by a VQGAN+CLIP model, which was the most advanced computer vision system available before the public release of Dall-E in 2022, and is best known for producing moving images of trippy, glittering scenes that seem to be constantly swallowing themselves. The stills are accompanied by sober, art-historical descriptions generated by a large language model. The effect is absurd: a simulated human voice vainly attempting to stabilize a system defined by flux. Chatonsky’s use of space distills the tension between dynamic modeling and classification.

The exhibition design also encourages viewers to note resonances across works. In the lobby, Paglen’s interactive Faces of ImageNet (2022) applies classification tags to visitors in real time. Standing before its camera, I was labeled—among other things—a vicar, a deist, a jihadist, and, most aptly, a “noticer.” The green bounding boxes and faulty descriptors appear again in Aljowaysir’s presentation of Bell’s photographs, in which Bedouins are mislabeled as militants, and domestic spaces are mistaken for military installations. Across the exhibition, nature recurs as both data and ideal, in Anadol’s simulated landscapes, Crespo’s speculative organisms, and Fontcuberta’s images of real and synthesized corals.

It’s worth noting that the curators of “Imaging After Photography,” Alison Weaver and Noor Alé, have avoided AI spectacle. Visitors won’t find the sort of imagery they’ve become accustomed to seeing in their social media feeds—no slick generative fantasies, no brain rot maximalism. The grotesque appears only sparingly, in Chatonsky’s uncanny figures, Crespo’s biomorphic entanglements, and Aljowaysir’s spectral absences. The exhibition is oriented not toward art as expressive excess but toward photography as a scientific instrument, even as it acknowledges that the byproducts of these processes can be unexpectedly beautiful and strange.

By focusing on perception and classification rather than trying to address the gamut of issues raised by AI—like authorship, misinformation, dystopian misalignment, the electricity consumed by GPUs running machine-learning models and the water required to cool the data centers that host them—the exhibition maintains a clear purpose. “Imaging After Photography” is effective not because it attempts to explain everything, but because it traces a problem across distinct practices, creating a space where divergent artistic viewpoints can come together and mutually enrich each other. In a world increasingly organized around probability, risk calculation, and prediction, “Imaging After Photography” offers a lucid look at how images function within those systems, and how artists might help us see them—not as pictures to be consumed, but as processes shaping how we understand the present.