Exiting the Photographic Universe — An Essay by Fred Ritchin
Written for ReVue — Fred Ritchin – February 2022
Recently, on a wintry day at New York’s Metropolitan Museum of Art, surrounded by large numbers of masked people wandering about, I entered a gallery devoted to English paintings from the 18th century. There I came upon two young women, their coats stacked on a nearby bench, who were taking turns posing in front of individual paintings, imitating the gestures of those depicted within the frame.
Moving about the room, each would photograph the other while murmuring instructions about how best to stand. Unlike the rest of us, trudging throughout intent on seeing as much art as possible, they appeared quite happy and self-contained performing their intricate ballet with cellphone cameras in hand.
Like Lewis Carroll’s character Alice, they had gone through a looking-glass that opened into a world beyond, bonding within the aura of these celebrated paintings. And, in turn, they had entered yet another world of image, that of social media, within which they could both celebrate and share their experiences, a parallel universe in which imagery from two centuries would merge.
They had created a virtual world of their own choosing, melding time and space, resonating with appearances, where actions have no physical consequences and immortality is assumed. The photographs produced, while made by a camera, were not intended to record the visible and interrogate the real, as photography had been employed before, but rather to make it malleable and amorphous. They had used their cameras to emigrate.
Previously, the dialectical relationship between the photographer, the camera, and that which had existed was central to the medium’s importance as well as critical to its reading.
While a photographer selected a point of view and the moment at which to release the shutter, the camera and its lens served as both ballast and counterpoint. Viewers could interpret the photograph in a variety of ways, certain it was a recording of the visible, overriding or circumventing the photographer’s initial intent.
This dependence upon the camera as well as the subjectivity of the photographer are acknowledged in Henri Cartier-Bresson’s preface to his highly influential 1952 book, The Decisive Moment: «These photographs taken at random by a wandering camera do not in any way attempt to give a general picture of any of the countries in which that camera has been at large.»
Almost a century before, in 1859, the critic Charles Baudelaire disparaged this new medium soon after its invention, calling it no more than a recording device, suggesting that it be utilized by «the secretary and clerk of whoever needs an absolute factual exactitude in his profession… But if it be allowed to encroach upon the domain of the impalpable and the imaginary, upon anything whose value depends solely upon the addition of something of a man’s soul, then it will be so much the worse for us!»
But what Baudelaire overlooked was photography’s stenographic importance to society as a credible witness, due largely to that very same «factual exactitude» that he found so diminished. The old bromide, «the camera does not lie,» tellingly did not include the photographer, nor did it include a paint brush or pencil.
But this fidelity to the visible, however incomplete, was deeply ruptured by the advent of the digital. Made of continuous tones rather than discrete pixels, the analog photograph did not invite extensive modifications beyond changes in contrast and some cropping of the frame. The difficulties of manual retouching made the process extremely exacting and time consuming as well as too expensive for most, and the alterations made were often relatively easy to detect.
The analog photograph’s architecture did not invite such wholescale reinvention, and its strength lay in leaving it largely alone. «A photograph is not only an image (as a painting is an image), an interpretation of the real; it is also a trace, something directly stenciled off the real, like a footprint or a death mask,» Susan Sontag wrote in the 1970s.
«If 'the medium is the message', as Marshall McLuhan put it in the 1960s, then the message of photography has become the malleability of that which is considered real rather than its recording and contemplation.»
Now the millions of discrete pixels which make up the photograph can be easily and nearly undetectably modified by software, transforming the photograph into a mosaic to be reshuffled quickly according to one’s preference. The medium’s transformation into more of a synthetic medium, like painting, has prioritized its modification while lessening the importance of its initial recording. «The camera does not lie» has little resonance today.
A photograph that primarily functions as a rendering of a scene without significant modifications may well need to be differentiated, possibly by a special frame or symbol, from the mass of similar imagery that has been extensively manipulated or, as we shall see, created by algorithms to simulate photographs without the involvement of a camera. A very large percentage of more conventional photography is now being made not by humans but by machines so that the digital code of the image can be read by other machines, such as happens in surveillance. The paradigm has shifted.
Images generated by artificial intelligence can sometimes approximate photographs while leading to unexpected results. Fred Ritchin used a text-to-image generator that allows one to input a phrase and the software responds by producing an image within two minutes. Above (top): «The perfect family», «Children with balloons». Below: «People in a concentration camp», «Racist people».
If «the medium is the message,» as Marshall McLuhan put it in the 1960s, then the message of photography has become the malleability of that which is considered real rather than its recording and contemplation. (Concerned by this eventuality, I titled a book that I published in 1990, In Our Own Image: The Coming Revolution in Photography.)
This tendency is the apotheosis of consumerism, in which the consumer (or the prosumer capable of both producing and consuming imagery) gets what he or she desires, or more accurately what they are persuaded by various forms of media to desire. While the products offered for consumption are often substandard, even deleterious to one’s well-being, this is rendered largely irrelevant by the ways in which they are rendered— the sleekly good-looking but tasteless tomato is made to prevail over the wrinkled one that is tastier and more nutritious.
Essentially, individuals are being encouraged to reconstitute the world to make it appear differently while remaining for the most part powerless, destined to be minor deities in their own cosmologies.
«A capitalist society requires a culture based on images. It needs to furnish vast amounts of entertainment in order to stimulate buying and anesthetize the injuries of class, race, and sex,» Sontag wrote nearly half-a-century ago in On Photography. She followed this with a thought that is more pertinent today: «Social change is replaced by a change in images. The freedom to consume a plurality of images and goods is equated with freedom itself. The narrowing of free political choice to free economic consumption requires the unlimited production and consumption of images.»
The «change in images» that she describes has metastasized beyond any reference to the actual. Now, while compressing the size of the digital image file, software routinely extrapolates the rendering of large numbers of pixels to produce a certain effect. No longer are they actual recordings of that which was visible but renderings that are made according to how the software determines that a scene should look.
And in a further disconnect, there are not only various digital filters that can be used to make one appear to be thinner or more alluring, or to modify a scene to make it appear retro as if it was photographed decades before or vice-versa, but algorithms that completely dispense with the need for actual observation by producing photographically realistic images of people and situations that have never existed.
Such synthetic imagery will soon begin to extensively infiltrate the media sphere, used in a variety of ways.
One will be able to make a workplace appear more diverse or one’s social circle more exciting. Produced quickly and in enormous quantities, the existence of such imagery will make the credibility of both lens-based photographs and videos more suspect, a destabilizing effect called «the liar’s dividend» in which actual recordings can be more easily dismissed as possibly fake. As a result, all that is inconvenient or upsetting, or requires a response as an individual and a citizen such as to a natural disaster or a massacre, can be denied as not having happened, and the same skepticism can then be applied to the historical record.
For example, was nine-year-old Kim Phúc actually running down the road in 1972 while her skin burned from napalm as she appeared to be in the famous Nick Ut photograph from the Vietnam War, an image that turned many against the conflict?
Was George Floyd really brutally suffocated by the policeman Derek Chauvin in Minneapolis in 2020, Chauvin’s knee on his neck, while other officers merely watched, or was this a figment of someone’s imagination?
Once again, the world can be thought of as flat.
Already, drawing from the remaining verisimilitude of the lens-based images that this synthetic imagery resembles, heads of countries are made to give fabricated speeches, and celebrities, almost exclusively female, are placed into pornographic films. Given the availability of enough imagery online, the same simulations can be made of anyone, including one’s classmate or professional colleague, a stranger, or even a child.
The damage to people’s reputations and sense of self-worth will be profound, as the current use of «revenge porn» against former partners has already demonstrated.
Moreover, such customized imagery can be weaponized to encourage racism and misogyny, as well as to fracture communities and provoke violent conflicts, undermining democratic processes. And while there are researchers working to be able to quickly identify such synthetic imagery, it is unlikely that by themselves technical processes, contemporary laws, or crash courses in media literacy will be sufficient to deal with the challenges that emerge as this imagery is more widely distributed.
Undoubtedly it will be autocrats and extremists of one sort or another who benefit the most from the breakdown of credible lens-based media, as they encounter diminished resistance by a disoriented public. Those in need of outside interventions, suffering from famines, conflicts, epidemics, environmental disasters, and other catastrophes will suffer the most. But others will be negatively affected as well, as an increased disconnect from contemporary issues and events exacerbates anxiety and passivity, making it increasingly difficult to grasp what is going on. The current attacks on «fake news» are an ugly aspect of this transition.
The virtual world, as a result, will enlarge to compensate for a lack of agency in the physical one. It most likely will be populated by many more synthetic beings—some will be created to serve as avatars, made to look like the individuals they represent, while others will be made to look and act completely differently, making social relations online even more uncertain.
Unable to trust others, people will quite possibly turn increasingly inward and become clannish, the global online communications infrastructure having made them feel paradoxically more isolated and alone.
Imagine, for example, a Zoom meeting with many of the rectangles populated by avatars, there to achieve the goals of those they represent or perhaps those of the software itself. One website already recommends that the viewer «get generated photos that will remind you of your skin color, age, gender, hair length, etc.,» and use them as a way to «give people an idea of your appearance, while still protecting your true identity.»
Imagine a Zoom meeting with, in this case, all of the rectangles populated by synthetic images that simulate photographs which can then be made to function as avatars. These images were fabricated by a «GAN (generated adversarial network) StyleGAN2» on the website thispersondoesnotexist.com on February 19, 2022.
Such synthetic imagery, often mistakenly referred to as photographs, will further transform both films and video games, as one is allowed to substitute images of others, including people one knows, as various kinds of characters.
Nor is this behavior only confined to staring at screens: while on the street some may choose or be forced to wear contact lenses or glasses programmed to substitute alternative images to conceal certain aspects of the environment—a homeless person sitting on the sidewalk replaced by an image of a fire hydrant, or people of one racial group exchanged for another. People with certain religious or other kinds of scruples could utilize software trained to transform a poster of a semi-nude woman in a bikini advertising an island vacation, for example, into an image of a palm tree.
Steve Mann, a self-described cyborg, has called this a strategy of «diminished reality». One might go further and call it the materialization of a parallel universe according to certain ideological preferences.
The publication of Magnum photographer Jonas Bendiksen’s recent book, Veles, purporting to be a documentation of a town in North Macedonia known as a hub for disseminating misinformation, was intended to start a discussion as to the role of synthetic images and the artificial intelligence behind them. The book appears to be a conventional photographic reportage, but all the imagery of people, although not identified as such, were created by software that Bendiksen trained for this purpose; the 5,000-word text that accompanies these images was itself produced by artificial intelligence.
Tellingly, no one caught on to the subterfuge, even his sophisticated colleagues, and months later the photographer finally found himself having to confess what he had done, an admission for which he was then criticized by those who viewed his project as a betrayal.
More recently Smarterpix, a German stock photo agency operated by PantherMedia, has partnered with the technology company VAIsual to offer the first set of licensable stock images consisting of photographically realistic portraits of people who do not exist.
As was reported in PetaPixel, all of these «can be generated with a green-screen background which allows them to be easily merged or inserted with other synthetic elements or real-life photography backgrounds to create entirely new content». Michael Osterreider, CEO of VAIsual, is said to view synthetic images as «the future of the stock photo industry» and «estimates that within the next five to 10 years, 95% of all images will be AI-generated media.»
As recounted by Jaron Schneider, PantherMedia makes the argument that these images «represent the next generation of ‘frictionless content’» since they «require no model release and can be used in any context, and those who download the images no longer have to double-check the terms of usage or fear expensive litigation.»
Similarly, Mark Milstein, CEO of Microstock Solutions, had suggested last year Digital Asset Managements (DAMs) «that offer synthetic media options will give users the ability to generate endless volumes of unique, custom-made photography with little to no licensing fees, legal restrictions or worries about competitors having the same image…. Synthetic media, photos generated from powerful algorithms, will be the heart and soul of tomorrow’s DAMs. Searching for content on a stock media site will be a thing of the past.»
What is considerably more innovative and illuminating than these simulations are the images generated by artificial intelligence that are meant to resemble photographs but do not, instead communicating other kinds of perceptions.
Both a word-to-image generator that I have used in which one inputs a phrase and the software responds by producing an image within one or two minutes, and an innovative artificial intelligence project by Alexey Yurenev exploring the mostly unknown World War II exploits of his somewhat uncommunicative grandfather when he was a soldier in the Russian army, «Silent Hero», produce a variety of images that are eerie, haunting, and weirdly ambiguous, the stuff of dreams and nightmares.
These images are unsettling in ways that photographs are not, as if the subconscious and subliminal can be delineated, concretizing a largely ephemeral psychological landscape that had previously been known only to the individual experiencing it.
Using the text-to-image generator, one can write, for example, «a scary man followed me down a dark street,» and the software may produce an image that features a man-like figure that does not resemble a photograph of the «real» but something more frightening, as if it had come from the depths of someone’s unconscious. Undoubtedly these will soon start to disappear as artificial intelligence becomes capable of simulating photographs with more precision; one would then need «dream» and «nightmare» buttons, as well as a «memory» one, that induce the software to continue to produce these forays into the nearly unimaginable.
Undoubtedly these images will modify our understanding and relationship to our own psyches, as well as to other domains. Ultimately, the impact of synthetic imagery may be even more profound than the revolutions in consciousness previously caused by photography which has been responsible for foregrounding seeing over the other senses, compressing space, highlighting the fractional second while appearing to arrest time, making both the far away and exceedingly small visible, and concretizing the present as the past, superseding human memory.
These paradigmatic shifts should challenge us with urgent questions as to photography’s role, as well as that of synthetic imaging. Can the credible witnessing function of lens-based media be sustained in the future? And if not, what will we replace it with? Are there more holistic and healthful ways in which both lens-based and synthetic imagery can be employed?
Certainly, those photographing in the future will have to consider other approaches to the medium if greater authenticity and credibility are concerns. They may need to reflect upon how to better contextualize their imagery while becoming more transparent in their practice, how to utilize other media such as video and sound synergistically with the photograph while creating narrative strategies that better engage their audiences, and how to collaborate more closely with those whom they depict to represent them with greater accuracy, among other considerations.
Each photographer will need to acknowledge more fully his or her role as author, responsible not only for their images but for how they are presented and understood as well.
Finally, as we exit the photographic universe that we have known, might we begin to discuss how to transition into a world that is less predicated upon its depiction as image?
Der ReVue Newsletter erscheint einmal im Monat. Immer dann, wenn ein neuer Artikel online geht. Hier en passant abonnieren.
Sie möchten unsere Arbeit
mit einer Spende unterstützen?
Hier en passant spenden!
Fotografie ist allgegenwärtig, wird aber in journalistischen Medien immer noch wenig hinterfragt oder erklärt. Bei «Journalismus» denken die meisten an Texte. Das digitale Magazin ReVue verfolgt einen neuen Ansatz, es betrachtet die Themen vom Bild ausgehend. Wir untersuchen in unseren Beiträgen die Rolle und Funktion der Bilder in Bezug auf Text, Wahrheit, politischen oder historischem Kontext. Wie nehmen wir Bilder wahr? Gibt es eine Geschichte dahinter? Unsere Artikel erscheinen auf Deutsch, wir übersetzen aber auch fremdsprachige Texte und erleichtern so den Wissenstransfer zu einer deutschsprachigen Leserschaft. ReVue ist unabhängig. Die Redaktion arbeitet ehrenamtlich. ReVue ist ein Projekt der gemeinnützigen DEJAVU Gesellschaft für Fotografie und Wahrnehmung e.V. in Berlin.
ReVue wird unterstützt von