Images and Objects

A painting hangs on a wall and does two things at once. It shows you something: a face, a hillside, an arrangement of forms and colours that you can recognize from across the room and identify in a book. And it is something: a physical surface where pigment has been worked into a ground, built up and scraped back, carrying in its texture the evidence of its own making. A Rembrandt portrait shows a man's face, but it is also a few square feet of oil paint on canvas, with ridges where the loaded brush pressed and valleys where a knife pulled. The face is the image. The paint is the object. When you stand before the work you receive both simultaneously, and the two are not easily separated. The depicted face persuades partly because the physical paint persuades.

This essay is concerned with the distinction between these two aspects of a painting: the representational content that can be captured and transmitted, and the material presence that, until very recently, could not.

I. Reproduction as signal loss

It helps to think about reproduction the way a communications engineer thinks about transmitting a signal through a channel. Claude Shannon showed in 1948 that every channel has a finite capacity, and that forcing a signal through a channel too narrow to carry all of it means losing information. The loss can be managed, concentrated in the dimensions the receiver is least likely to miss, but it cannot be eliminated. A JPEG photograph discards spatial detail the eye is slow to notice. An MP3 drops frequencies the ear handles loosely. In both cases the compression is designed around a model of what the receiver will and will not perceive.

Reproductive printmaking has operated on the same logic since the fifteenth century, though nobody involved would have described it that way. An engraving after a Raphael fresco preserved composition, figural arrangement, and a version of the tonal structure, because those were the properties a copper plate and a burin could encode. What it could not encode, it simply discarded: colour, surface texture, the behavior of fresco plaster, the scale of the original wall. Every subsequent technology widened the channel, but always along the same set of axes. Mezzotint recovered continuous tone. Lithography added the direct gesture of the hand. Chromolithography brought colour. Photomechanical halftone replaced the interpretive engraver with the camera. Digital inkjet pushed colour fidelity close to what the unaided eye can resolve. The trend across five centuries is remarkably consistent: each advance captured a new dimension of pictorial information, and the information was always of the same kind. It was always about the image.

What none of these processes attempted to rebuild was the painting's behavior as a physical thing. How its surface catches and redirects light. How thick passages cast shadows that shift when the viewer moves. How gloss varies from area to area, routing the eye differently depending on where you stand and where the light source is. A painting does not have a single correct appearance the way a photograph does. It has a field of appearances that changes with conditions, and that field is generated by the object, not by the image.

The reason for this persistent omission is technical. Image information is two-dimensional and viewpoint-independent. A sensor can capture it in a single frontal exposure with little cost. Object information is fundamentally different. It is three-dimensional, viewpoint-dependent, and relational. A painting's surface is a material topography, rarely more than a few millimeters deep, in which every passage has its own way of interacting with incoming light. That interaction depends on the geometry of the paint deposit, on the optical properties of the pigments and binders, on the roughness of the surface at scales too small for a camera to record. A photograph of a painting samples this interaction once, from one angle, under one set of lighting conditions, and presents the sample as the thing itself. But the thing itself is the entire space of possible interactions, and no single sample can represent it, any more than a single frame can represent a film.

This is why mechanical reproduction has always been better at the picture than at the object. The two kinds of information occupy different dimensions, and the channel through which reproductions have been made was built to carry one of them.

II. Why nobody noticed

The most interesting feature of this centuries-long loss is how invisible it has been. People who care seriously about paintings have not, on the whole, described reproduction as a problem of lost objecthood. They have described it as a problem of insufficient image quality: better colour matching, higher resolution, wider gamut. And by those measures, reproduction has been a success story. The object was not missed because it was not perceived as absent until we've reached at least some acceptable image fidelity. Explaining why requires looking at the tools through which paintings have been studied, discussed, and sold, because those tools shaped what was possible to see.

André Malraux understood this earlier and more clearly than most. In 1947, describing what he called the musée imaginaire, the museum without walls, Malraux argued that photographic reproduction had created an imaginary collection in which artworks from all periods and cultures could be compared side by side, freed from the constraints of geography and physical access. He was right to see this as a liberation. But Malraux also noticed the terms of the liberation, and they were strict. By reproducing everything as photographs, the musée imaginaire equalized scale, suppressed materiality, and reduced three-dimensional objects to flat images. A monumental fresco and a miniature panel painting became the same size on the page. A carved relief and a painted canvas became equally smooth. Photography did not only fail to capture the physical surface. It produced a way of encountering art that made the physical surface seem incidental, by presenting every work as though it were, at bottom, a two-dimensional composition.

The consequences of this reach further than Malraux himself explored. Art history as an academic discipline consolidated in the late nineteenth and early twentieth centuries, and it consolidated around photographic reproductions. Heinrich Wölfflin, whose Principles of Art History (1915) established the analytical method that subsequent generations inherited, constructed his argument through side-by-side comparisons of projected slides. His categories are powerful: linear versus painterly, plane versus recession, closed versus open form. But they are, every one of them, properties visible in a photograph. Surface texture, material presence, the way paint behaves differently under changing light: these were outside Wölfflin's analytical framework, and they were outside it because they were outside the medium through which he received his evidence. The discipline's conceptual vocabulary was built for, and built from, images. The channel shaped the scholarship, and the scholarship shaped how people learned to look at paintings.

This created a self-reinforcing cycle. Photography transmitted images. Scholars working from photographs developed categories suited to images. Students trained in those categories evaluated paintings by image criteria. When they became curators and critics they perpetuated the framework, and the framework made objecthood peripheral by making it analytically invisible. Over time the omission started to become the natural order of things: of course reproductions were flat, of course the surface was something you could only get in person, of course what mattered for purposes of study and appreciation was the image. These came to feel like observations about the nature of painting when they were really observations about the nature of photography.

The commercial world reinforced the same structure from another angle. Because objecthood could not be reproduced, the market organized into two poles. At one end, the singular original: maximal physical presence, maximal perceived value, accessible through galleries and auction houses to a small number of collectors. At the other, mass reproductions that preserved the image and discarded the rest, priced as the commodity products they were. Between these two poles there was almost nothing, and the emptiness was not the result of indifference. It was a consequence of the technology. No process could produce a wall object that fell between the singular painting and the flat poster, so no such product category existed, and because the category did not exist it could not be named, and because it could not be named no demand could organize around it. The thinness of the middle looked like a fact about the market when it was really a fact about the channel.

III. Benjamin's aura as bandwidth

Walter Benjamin's 1935 essay on mechanical reproduction gave the word "aura" to this conversation, and the word has been so widely circulated that it has become difficult to use precisely. But the underlying observation is more exact than its afterlife suggests, and the compression framework offers a way to recover that exactness.

Benjamin wrote that the original work of art possesses an aura, a "unique apparition of a distance, however near it may be," and that mechanical reproduction destroys it. He located this quality in the work's authenticity, its rootedness in a particular place and a particular history, its embeddedness in tradition. When the work is reproduced, he argued, these are stripped away. What remains is the detached image, mobile and contextless.

The standard reading treats aura as irreducibly mystical, a halo that vanishes the moment a copy is struck. But the compression framework suggests a more useful account. What Benjamin called aura is, at least in substantial part, the information surplus of the object over any available reproduction. When people stand before a painting and sense something that no photograph prepared them for, they are registering signals the photographic channel could not carry: the surface relief, the way gloss varies from passage to passage, the micro-shadows cast by impasto, the way the work shifts as you move around it, the sheer physical scale of it in the room. These are not mystical properties. They are measurable characteristics of a physical object interacting with light and space, and they went unrecorded because the recording medium could not encode them. But it must also be said that this aura takes from the preciousness of this singular object, whose precise arrangement grew out of a specific time and place in history, by the hand of a specific figure.

This reading does not diminish Benjamin. It gives him something he lacked, which is a mechanism. If aura is constituted, at least partly, by information that the available channels cannot transmit, then the "destruction" of aura under reproduction is not a metaphysical catastrophe. It is in part an engineering limitation. And if the limitation is engineering, then it is in principle addressable. Not fully, perhaps. Not in every dimension. But progressively, in the way that colour fidelity was progressively addressed over the course of the twentieth century. Benjamin wrote at a moment when the available reproductive channels were photographic and photomechanical. He could not have imagined a channel that would begin to encode painted surface, because in 1935 no such channel existed or was foreseeable. His conclusion that reproduction shatters the original's authority was correct for the channels available to him. Whether it holds for channels he could not anticipate is a separate question.

IV. Berger and the stillness of things

John Berger arrived at a related insight by a different path. In the opening chapter of Ways of Seeing (1972), he observed that "original paintings are silent and still in a sense that information never is," and that "the silence and stillness permeate the actual material, the paint, in which one follows the traces of the painter's immediate gestures." This has the sound of a poetic observation, and Berger was a poetic writer, but it is also analytically precise.

Consider what "silent" and "still" actually mean here. A painting does not narrate itself. It does not unfold in time, does not direct attention in sequence, does not argue. All of its information is present at once, available to be received in whatever order the viewer chooses, for as long as the viewer cares to look. When Berger showed, in the television version of Ways of Seeing, how a camera panning across a Breugel converts spatial simultaneity into temporal sequence, he was demonstrating a concrete form of information loss. The viewer's freedom to look at the painting in any direction, at any pace, is replaced by the filmmaker's decision about where to look next. The painting's capacity to hold its own authority, to exist as a complete thing rather than as raw material for someone else's argument, is compromised because the medium of transmission has imposed a structure that the original did not have.

But Berger's "stillness" points at something deeper than the absence of temporal sequence. A painting on a wall is still because it is an object that does not change between viewings unless the conditions change. It sits in the room and responds to light, to proximity, to the angle from which it is approached. This responsiveness is a property of the object, not of the image. A reproduction on a screen presents its image simultaneously too, but it does not possess Berger's stillness, because it lacks the physical presence that generates the experience. The screen is always a screen, regardless of where you stand or how the light falls. The painting is always, quietly, itself.

V. Goodman and the logic of forgery

The philosopher Nelson Goodman gave the image-object distinction its most rigorous formulation in Languages of Art (1968), though he framed it in terms that appear, at first glance, to be about a different problem. Goodman proposed a division of the arts into two classes. An art is autographic, he wrote, if the distinction between original and forgery is significant: if even the most exact duplication does not count as genuine. An art is allographic if a correct copy simply is the work. Painting is autographic. Literature is allographic. You can forge a Vermeer, but you cannot, in any meaningful sense, forge a novel. A copy of Ulysses that reproduces every word in the correct order is not a forgery. It is Ulysses.

The difference, Goodman argued, lies in whether the art possesses a notational system, a way of specifying the work's identity through compliance with a set of instructions, independent of history of production. Music has a score. Literature has a text. Painting has nothing of the kind. Its identity is bound to its specific physical existence: the particular pigments, the particular surface, the particular hand. Two paintings that look identical in a photograph remain two different paintings, and the difference between them is not administrative. It is physical. It lives in the surface, in the material, in the traces of making.

This is where Goodman's framework connects to the question of reproduction. What much of reproductive history has done, without stating it explicitly, is treat painting as though it were allographic. Every poster, every giclée, every framed print implicitly assumes that the painting's "score" is its image: the composition, the colour relationships, the depicted scene. A rendering of this score is offered as a satisfactory surrogate. But Goodman's entire point is that painting does not work this way. The information that distinguishes the original from a copy, even a perceptually excellent copy, is the information that resides in the object: the history of production inscribed into the physical surface. This is what the present essay has been calling objecthood. Goodman's analysis gives it some philosophical rigor: objecthood is the collection of properties that make painting autographic, the properties for which no notation exists and which therefore cannot be transmitted by any medium that operates like a notation. Photography operates like a notation. It extracts a description and transmits the description. What it leaves behind is everything that makes the painting unreducible to a description.

This is why seeing a painting you know well from reproductions so often produces surprise. The reproductions transmitted the notational content with increasing fidelity. What they could not transmit was the non-notational remainder: the physical fact of the thing. The surprise you feel in the museum is the registration of information that no channel you previously encountered could carry.

VI. The channel widens

Against this background, the emergence of a process that begins to encode surface information is significant in a way that has nothing to do with printing technology per se and everything to do with the shape of the channel. For the first time, the channel is being widened in the direction of objecthood rather than along the familiar axis of image fidelity. The reproduction is being asked to carry some portion of the non-notational content, to behave, partially and imperfectly, as a physical thing that interacts with light and space rather than only as a carrier of pictorial information.

A companion piece of mine takes up the technology, the market implications, and the institutional questions that follow from this. What the present essay has tried to do is something prior: to explain what reproduction has always lost, why the loss was systematically invisible, and why recovering even part of it constitutes a different kind of advance from the improvements that came before.

If a painting's perceptual force depends partly on its objecthood, and if reproduction has discarded objecthood for its entire history because no channel could transmit it, and if the tools we built to study and sell and discuss paintings were shaped by the limitations of that channel and therefore treated objecthood as peripheral, then a process that begins to transmit objecthood is not simply a better printer. It is addressing the dimension that the entire apparatus of reproduction, scholarship, and commerce learned to ignore because ignoring it was the only available option set forth by technology.

An image is what a painting looks like. An object is what a painting is. The vocabulary needed to describe what was lost has been available for decades, distributed across Shannon and Benjamin and Berger and Goodman, but it has never been assembled in one place, because until recently there was no practical way to assemble it. Now there is.