“True, Master Doctor, and since I find you so kind, I will make known unto you what my heart desires to have; and were it now summer, as it is January, a dead time of the winter, I would request no better meat than a dish of ripe grapes.”
Christopher Marlowe: The Tragical History of the Life and Death of Doctor Faustus
I mean, have you seen images recently? Can you even look when so many are so distressing? Can you even recognise images given the number and speed that they come at you? Can you make any sense of images when they are so often beyond comprehension?
Fleeting, flowing, streaming like quicksilver. All the frames from a film out of order, or the pages of a library scattered to the winds. The traditional systems of organisation shattered and replaced by feeds. Beginning and end replaced by infinite scroll (invented in 2006 by Aza Raskin). The end of the end.
Like a luxury hotel breakfast buffet serving breakfast, lunch and dinner all at once, chicken feet, black forest gateau, pancakes, spaghetti, burgers …Time and geography have collapsed, the hierarchies of menu, cuisine, sequence have been shredded. Everything all at once, anything-at-any-time.
In what is unquestionably the era of images it has, ironically, become harder to look at them, more impossible to see. Images are now the conduit of everything. They act as the all-powerful organisers of the world from family memories to political power, yet they have become ubiquitous to the point of banality.
Here we are. Haunted by insistent phone-snap memories of things we can’t remember; immersed in a culture so entirely visual that visual culture itself has become ungovernable. How can anything - an image in an age of images especially - mean anything anymore? If images are the water we swim in, has the significance of an image dissolved entirely?
If you've seen everything, what exactly is the point of looking anymore? If every image has equal value, is anything significant anymore? If we’ve become so numb, can anything make us feel again? How are we supposed to both truth and satire are dead?
And what does it mean to be in the business of making images when they are now just an illustration proffered by a prompt. Just another contested truth brought into the world. The image is not something that is made but rather conjured into existence. Images might not even be there to be seen anymore. Perhaps the total ubiquity of the image coincided with its disappearance.
How, in other words, can we resist this culture of images if we, against the odds, still believe in the image? How can we make space or time or focus that an image needs to exist as a thing in the world? As a site itself rather than an atom in a molecule in a drop in a raging ocean. When we are dragged down or drowned out, or washed away or otherwise…
Perhaps our only salvation is irony. That irony’s ability to give us distance and connection simultaneously can offer us the only remaining way to say I love you and really mean it.
In the after/image, in the age of screen burn, suffering from palinopsia: the recurrence of an image after the stimulus has been removed.
The 20th century had its image crisis of course. The rise of processes and means of distribution and consumption: photography, mechanical reproduction, TV and the ‘global village’ etc. And it had its ways of responding and recouping: cubism, collage, pop etc as ways to re-see images as things that could contain concepts and ideas like time and space, category slips, forms of critique.
For us though, deep in the digital mire, without the benefit of hindsight, things are less clear.
Images were once location specific - contextual. They were printed on a specific page of a particular book with a specific caption. These physical and cultural sites, things outside of the image itself, gave meaning to the image. The same image encountered in different environments—the library or a newsstand—might acquire different meanings. Images in different forms of media were also framed by their context - the tabloid, say, or the broadsheet. Where images resided gave images particular meanings. Captions that accompanied images changed how we understood them, how we engaged with them, often preceding the meaning we derived from the image.
With the advent of search engines—Google Images—though, they…the images became disassociated from context. Cut loose from the anchor of place or caption: weightless, free floating in the infosphere like spores in the wind. The organisational structures of culture and canon that once tethered image to context were vaporised by the search algorithm, breaking the html code that tied information together, cutting the bond between image and context. And in doing so, loosening the specific meanings that the images once served.
Now images exist in the multidimensional latent spaces of generative AI models.
At this moment RE-LAION 5B is the largest open image–text pair dataset. It contains 5.5 billion image-text pairs hoovered from the internet. This data is then processed using Contrastive Language-Image Pre-training. CLIP, itself an AI neural network, learns how to connect visual concepts with language which it uses to categorise images without relying on human intervention to label and categorise data. It detects for example the content of images and decides how closely they match their descriptive text. Machine learning and looking allows for the processing of such huge numbers of images.
Using recognition, detection and image processing algorithms each image is ‘tokenised’ or chopped up into fragments of image/text chunks and associating images with concepts even down to the pixel level.
Each of these image-fragments is referenced against an index (objects, colour shape, type etc), often with a single fragment tagged to many categories. The index retains the links between each label and its image-fragment referent to form the fundamental platform that allows future image generation.
The internal logic of an image dismembered, its own self decontextualized. Broken into micro fragments, into inventories of image-ingredients and ready to be reassembled to approximate the inputs of brief text prompts.
Architecture has had its own historical obsessions with logging, measuring, recording, sorting, naming. These acts, from the Renaissance to Charles Jencks, from Stuart and Revett antiquities through Banister Fletcher, in books from language of classical architecture to the language of postmodern architecture. A continuous project that organised orders, styles, windows, regions, periods. And its own history of using these as language models to generate new possibilities. What is John Soane’s house except a (physically small yet incredibly dense) Language Model? All those fragments, casts, and models ready to be reassembled into new architectural propositions?
Digital culture might not have changed all images themselves. Architectural imagery might not all be produced by AI (yet). Much is not a product of collage (though a lot is). Drafting might still assert its truth in relation to construction. Renders as simply high-tech remakes of measured perspectives, Alberti with processing power and Hollywood aesthetics. The structure, intent and logic of the image remains the same. But digital culture does—and has—changed our relationship to images.
A lament for a vanished order that could be disrupted.
All Images Copyright Sam Jacob