Photorealism: Painting, Photography and the AI Image

A closer look

The archive collage assembles a diner, ice creams, a shop window, a crushed can, vehicles, apples and a seated figure. Repeated highlights and carefully described surfaces unite the scenes, though visible brushlike texture remains. These are generated illustrations of possible subjects, not paintings by the artists discussed below.

What the term identifies

The exhibition account for Picturing America: Photorealism in the 1970s describes painters working directly from photographs, translating ordinary urban and suburban subjects onto canvas or paper. That relationship to a source image is more specific than a general ambition to make something lifelike. Printing technology is not a defining requirement for a painting to be photorealist. [1]

Keep three questions separate: what does the finished image look like, what material object are we seeing, and how was it made? A photograph of a painting, a painted copy of a photograph and a generated image of an imagined painting may look similar on a small screen. Their processes and evidential value are different. Zooming in can reveal details, but a caption and production record still do work that pixels cannot.

Richard Estes: reflections are a structure

The Smithsonian’s Richard Estes’ Realism exhibition explains that Estes combines multiple viewpoints to build complex scenes. Its examples include Diner (1971), an oil painting on canvas. This makes the idea of passive copying inadequate: photographic information can be selected and organised into a composition whose relationships demand sustained attention. [2]

When looking at a documented Estes reproduction, separate the window surface from the interior and from what is reflected outside. Follow a vertical edge until it meets a second plane. Which apparent continuation is real within the depicted space, and which belongs to a reflection? Such questions explain why a familiar street can become difficult to read. The subject may be ordinary; the act of looking need not be.

Chuck Close: a face assembled through a system

SFMOMA describes Close’s use of a source photograph and a grid to transpose an image onto another surface. MoMA’s record of Self-Portrait (1991) explains how a gridded Polaroid became a painting assembled from small, irregular coloured forms. His later work makes the units conspicuous rather than concealing all traces of making. [3][4]

Compare the scale at which individual marks become visible with the distance at which a face coheres. This is a productive alternative to grading realism as a contest in microscopic detail. The picture can move between abstraction and recognition depending on how it is encountered. A useful account of an artist allows a practice to change; it does not turn an early classification into a lifelong technical rule.

A source photograph is not automatically a painting

The Whitney catalogues Audrey Flack’s Queen, dated 1975 and printed in 1984, as a dye-transfer photograph mounted on paperboard. Its objects include a rose, a locket, a chess piece and a playing card. That specific record is useful precisely because it is a photograph: identifying an artist associated with Photorealism does not tell us the medium of every work by that artist. [5]

The same care applies to broad labels such as realism and hyperrealism. Exhibition terminology has overlapped, and a fixed rule that one is truthful while another is fantastical is too crude. Start with the work’s catalogue record and the writer’s stated usage. Our guide focuses on photographic reference, painting and the choices involved in translating between them. [1]

Using AI without confusing appearance and evidence

In our collage, a reflective window and a crumpled can invite close looking, but they cannot establish an actual shop, product or event. A plausible highlight may coexist with inconsistent geometry or unreadable lettering. Inspect the relation between light source, shadow and reflection, rather than assuming that a detailed surface must be coherent.

For a practical study, photograph an ordinary object you own under one light. Draw its outline, the largest shadow shape and three highlight locations before adding texture. If you make a separate AI interpretation, record the prompt and compare the same relationships. Keep the two results labelled. Precision becomes useful when it clarifies a decision; endless detail without a decision is simply a very busy surface.