Photorealism and forgery

Digital Emily specular map. Creators, “ReForm | Hollywood’s Digital Clones,” YouTube, 14:07, May 19, 2015.

In recent years more and more film productions have attempted the final frontier of photorealism: the accurate and undistracting rendering of human faces in motion. Rogue One (2016), Blade Runner 2049 (2017), and The Irishman (2019) all feature more or less believable human faces that exist without 1:1 physically present counterparts. In these films, reference images are used to guide the realization of a face (belonging to deceased or aged actors), but no original photographic imagery is carried over from the real to the virtual. Whereas films like The Social Network (2010) used composite photoscans pasted on digital geometry to seamlessly integrate twinned characters, in this recent wave of productions the depicted actors are entirely rendered, with new thresholds of realism achieved by using AI-animated, procedurally-generated maps of skin pores that wrinkle at microscopic levels. Such advancements break new ground in visual effects technologies and optical sciences, and represent one side of a coin shared with similar advancements in surveillance techniques (where the goal is not to produce distinct human faces but to recognize and categorize extant faces). The novelty of these effects means that they are as important to ad campaigns as the actors themselves—ensuring that the viewing experience is one where technical aspects become as engrossing as narrative conventions. But what happens when they become ubiquitous?

Suspension of disbelief has always been intrinsic to cinematic images. The depicted narrative and all its leaps in logic and physics are counterbalanced by an equally compelling narrative of its creation: what happens behind the scenes? How do real circumstances produce fictitious events? This is a narrative unmarked by plot holes or breaks from reality, and it is the existence of this second minor narrative that allows the first one to operate despite disbelief. But outside of cinema, and without the presence of the minor narrative that explains the process of production, we have no idea what is “really” going on. Because photoreal visual phenomenon are no longer exclusive to camera-recorded images, we lose direct access to the realism of a given image. What is perceivable in both documentary photography and digital rendering have become indistinct, and new paradigms of image circulation mean that context is unreliable for distinguishing the origin of an image. Certainly there are photoreal renderings of previously unobserved real phenomenon—such as dinosaurs in Jurassic Park (1993) or black holes in Interstellar (2014)—and these visual effects can be distinguished from straight-out-of-camera images on that basis. But vfx have also been used to recreate less fantastical realities, too. The digital rendering of coastal mansions in The Wolf of Wall Street (2013) or coffee pots in Panic Room (2002) may not qualify as “real” in that they are not indexical phenomenon captured by some objective camera, but nonetheless they seek to be visually indistinct from such phenomenon. The end result is an image that did not necessarily happen in a specific spatio-temporal location, but it could have.

That which stands in the way of did happen and could happen is arguable: in cinema it is often a question of budget and time. Given enough resources, vfx could be sfx. In other fields, such as architectural visualization or product rendering, the photoreal image serves as a proposition: this could be real, this could be yours. In many instances renderings do provide a precedent for a real image—although here we could say that the product indexes the image rather than the image indexing the product. Writing on the prevalence of product renderings in store catalogues, one photographer asks, “When does this become false advertising, if none of the products showcased are even real?”[1] This is an interesting dilemma for how we tend to conceive of (photo)realism. Without a camera, it is possible that renderings necessitate an entirely different mode of discussing their relationship to the real. Perhaps these images leave the realm of fiction and are best understood as speculations or propositions to affect reality. Of course renderings do not produce false advertisements; the truth of an advertisement is entirely speculative, it lies in the exchange rather than the use of the product. If attention is seized, the ad has been fulfilled. We are already aware that speculative representation is as “real” as any indexical representation of reality: this is increasingly the mode of reality in which we live post-2008 bank bailouts. The relationship of financial markets to really-existing resources is deeply tenuous—infinite growth is physically impossible in a finite ecosystem—but nonetheless our rent, employment, and healthcare are structured by these paradoxes. This cannot be discounted despite the basically fictional representational relationships at play. Rendering is simply the aesthetic correspondent to this financial paradigm.

Thermal image of emergency diesel generator at data centre

We can begin thinking of images in two senses. On one hand, we have the historical precedent of additive images. These images are constructed: they are planned (drawn) and then executed (painted). This process can be used to describe all manner of frescoes, engravings, animation cels, and so on. It is necessary to position rendering within this trajectory, as does media theorist Lev Manovich when he writes that the “manual construction and animation of images gave birth to cinema and slipped into the margins… only to reappear as the foundation of digital cinema.”[2] On the other hand, there is the relatively new phenomenon of reductive images. These images are edited, manipulated, and otherwise selected out of an infinite number of other potential images. Reductive images are taken rather than constructed. They retain indexical links to that which they depict: in film photography this index is physically instantiated by chemical processes that react to the light of a specific time and place. It’s important to note that reductive and additive images do not just correspond to realistic and caricaturistic effects. Reductive images, for instance, can be indexical without being photoreal—I am thinking of x-rays and thermal imaging that reflect reality outside of the visible light spectrum. It should be well-understood at this point that images are never direct representations of reality, even if confined to an exclusively visual sense. They are processed and edited, they are produced for specific purposes, and even documentary devices like security cameras make concessions to the “real” field of vision in order to optimize storage and enhance low-light conditions. Similarly, additive images may forego any claim to photoreality while still maintaining accurate representational linkages to other aspects of reality. Cartography, computer aided-design, and 3D modelling all strive for a realism of geometry rather than light. Furthermore, we can consider pictorial writing systems like hieroglyphics as imagistic languages that exist at the edge of indexical representation in order to assert real connections between signs. What is common to both rendering and writing is their effect on perception: they may not accurately reflect reality, but they follow a set of laws (like light physics or established syntax) that enable one to experience an unobtrusively mediated form of reality. Given the right conditions, one can forget that their experience of reality is mediated at all.

Photoreal digital rendering therefore represents not only an evolution of additive image-making; it also enables a kind of additive realism that can be confused for a reductive realism. This capacity has occasionally existed within photorealistic painting, but is extended even further with advancements in rendering. Additive images have always possessed more plasticity than reductive images due to the fact that they do not have to be recorded, only rendered. Conflating the two produces a new kind of realism that is not simply about mimicking photographic representations of reality, but is speculatively or imaginatively real. In order to confront this confusion, the way we ascertain the realism of a given image should not be limited to criteria like “how well does the index mimic its source?” Instead, given only the existence of an additive image without any 1:1 physical correspondent, we ought to pursue a line of questioning that begins by asking “is the image believable?” That is, does the image obey our existing cultural and aesthetic experience of reality? Or is it so well camouflaged that we don’t even think to ask? Additive images succeed not by representing reality, but by recreating familiar mediations of reality: photoreal images corroborate a culturally-specific worldview that is informed by photography. They achieve a high degree of realism precisely because they represent reality in a familiar way. In fact, photoreal images not only replicate lighting conditions and physics simulations, but also camera mechanics like lens distortion and chromatic aberration. This form of imagery closes the loop between our expectations and our perceptions—if nothing is amiss our worldview remains intact. As additive realism becomes more prominent, so too will the forms of realism it proposes become more entrenched in our aesthetic unconscious.

Although Hollywood films currently offer the most prominent and advanced instances of additive realism, we inevitably view these films through the suspension of disbelief. What remains most interesting to me is the possibility of circulating additive images in contexts where photography still functions as an index of the real. It is in these instances where an additive image can be truly confused for a reductive one, despite the great disparity in their modes of production. This confusion is enabled by the lack of disbelief that still clings to some zones of aesthetic reality. Such zones may be found on eBay and Craigslist, where product photography is typically produced by individuals with smartphones rather than studios with rendering workstations, or on Tinder and Grinder, where images may be used for deceitful purposes but are still reductive even if they don’t correspond with one’s real identity. There are certainly disruptive possibilities here, especially as rendering technology becomes increasingly accessible even to individuals of modest means. (Blender, for instance, is an open source software that can run on basic laptops, and innumerable tutorials are freely posted on YouTube.)

But, in my view, the most interesting disruptions lie in the field of visual art. This is a zone in which the reliance on images cannot be overstated. Consider not only that the majority of artwork is seen (and contemplated) through the proxy of its documentation—appearing widely in press releases, reviews, art blogs, newsletters, catalogues, archives, and Instagram—but also that many auction houses and commercial galleries now derive notable percentages of their business exclusively through the transmission of digital images. Websites like Paddle8 and, as well as proprietary online viewing rooms at Zwirner, Gagosian, et al., are increasingly common sites of commerce in the art world. It is no longer a prerequisite that an artwork be seen in person. (Advancements in the financialization of artwork also mean that it is no longer necessary to ever see one’s collection in person, if at all: artworks can be stored in remote duty-free freeports until they have accrued sufficient value.) The centralized model of art-as-object—wherein an artwork is physically unique, and therefore confined to a single place at a single time—has given way to a decentralized model of art-as-image. In many instances, the image does not function as an index of the artwork so much as the artwork indexes the image. This was certainly one of the propositions advanced by postinternet artists. As Hito Steyerl puts it, “Reality now widely consists of images; or rather, of things, constellations, and processes formerly evident as images.”[3] Images function not so much as documents or indexes, but as end-products. Post-processing is the mode of production.

Art is becoming a more and more stable asset category. More and more resources are funnelled into the production of art, meaning that the modes by which art is circulated become of the utmost importance; as in the “attention economy” more generally, value is directly related to provenance and ubiquity. Artist Artie Vierkant writes that “conceptual art assured its own legacy by the overwhelming volume of language produced within and around it at a time when summary-through-language was the easiest means of disseminating an object.”[4] This paradigm is long obsolete. Images now circulate incomparably faster and further than writing, especially as publications’ budgets for criticism dissolve and art history departments are increasingly staffed by short-term sessional faculty. A few years ago, art blogs like VVORK represented novel extra-institutional ways of introducing artworks and artists into the discourse, even without supplementing the images with writing. Now, image dispersal platforms like Contemporary Art Daily and Art Viewer have moved from the margins to the centre of artistic discourse. Although they use the same format as the earlier wave of art blogs, their operation is less about freely distributing images and more akin to the ads-as-content paradigm of Vogue. Advertisers pay to “support” the platform, and in turn their exhibition documentation is distributed through the site’s channels.

Given that images are increasingly detached from the artworks they depict, and given that images do not always circulate with appropriate contextualizing information, it has become necessary for photographic documentation to index not only the work but also its context. Artwork–images are authenticated in subtle ways—minimalist architectures in the background of installation shots reveal the identity of where the artwork was documented, and offer a legitimizing institutional context for the image’s contents. That an artwork was curated into an art institution is precisely what allows the work to be art, and therefore to be valuable. This legitimacy is secured by the fact that no two white cubes are exactly alike. Consider SculptureCentre’s warm and grungy basement, the curved stairs in Tanya Leighton’s Berlin space, or Preteen Gallery’s infamously overblown fluorescent lighting. These traits behave similarly to watermarks—their presence in the image is at once negligible as well as crucial to declaring the image as art. A sculpture on a seamless backdrop simply cannot be counted as an artwork–image in the sense of being legitimated by institutional authority.

David Diao, Slanted MoMA, 1995, acrylic and vinyl on canvas, 46 x 75 in. (117 x 190 cm)

So: the current moment is one where additive realism is increasingly accessible, and where the art economy is increasingly image-dependent. Artist Alan Warburton is correct that the territory of software and digital production has gone largely unexplored despite structuring an incredible portion of contemporary society (watch his excellent video essay “Goodbye Uncanny Valley” (2017) here). The reasons are many: the learning curve is steep, the field is deeply male-dominated, and the art market has found little use for digital art. But whereas Warburton advocates for absurd or surreal video, I am interested in the possibilities that may inevitably open for art forgery. If we incorporate some Hollywood thinking, then perhaps we can speculate a kind of forgery that is not so much about replicating already existing assets, but about replicating the authenticating signs bestowed on work by art institutions. In this way, new assets could be simulated, having already been marked by a curatorial stamp of approval—and in fact it would be easy to simply drop the artwork into multiple virtual gallery spaces as evidence of the work’s circulation. Of course, such a form of forgery would still require the use of other (already readily available) digital forgery techniques to bypass the two-factor authentication of confirming a work was curated in a given place: hacking institutional websites, falsifying CVs, generating certificates of authenticity, and so on. But these are details. The possibility exists.

The end game of these new possibilities for art forgery is an epistemic distrust in image economies. The loss of faith that allows one to accept an image as real or indexical may, in some instances, produce a compulsion to look harder. By inhabiting the forensic gaze, the art viewer not only grapples with whether or not the image is believable, but also ends up contemplating previously overlooked aspects of the work’s appearance, ultimately begetting new possibilities for meaning and critique. The idea of legitimation through curatorial affirmation, or the spectacle of big budget studio fabrication, may give way to the kind of objectless preoccupations that once marked conceptual art. Would this forthcoming epistemic shift represent the loss of a certain visual language? Or does the relinquishment of indexical representation open up the visual field for more complicated forms of realism that do not subordinate the digital to the analogue? We may move from reading images to truly seeing them.

–February 2020


  1. Justin Heyes, “IKEA’s Hyper-Renders Could Be the End of Product Photography,” SLR Lounge (2015)
  2. Lev Manovich, “What Is Digital Cinema?” (1995)
  3. Hito Steyerl, “Too Much World: Is the Internet Dead?” e-flux 49 (November 2013)
  4. Artie Vierkant, “The Image Object Post-Internet” (2010)