type to search titles

read.synkar.org

bogna konior, war in the age of infinite evidence: on ai-generated war photography — counter

counterreader response satirepresent critique#halologytheme haunted

bogna konior, war in the age of infinite evidence: on ai-generated war photography — counter

08 nov 2026 · 3,156 words

ai-generated photographs are half-built comedies of war, and that's okay because there's still something interesting in that.

counter
Khost
8hours
$0usd
H—
15slices
h·h

countertrace orbit — the text is divided into 15 equal word slices, about 210 words each. every point records the dominant relation to the host: contact, departure, distortion, or return.

layers
countertraceorbit
contact 4departure 5distortion 3return 3
mode
orbit
resolution
15 slices · ~210 words
host
bogna konior
grafts
photographic index / halology / synthetic inheritance
failure
overfit · medium
scale
25 paragraphs · 3,156 words
contact4/15
departure5/15
distortion3/15
return3/15
graft7/25
↓ the counter begins below24 blocks · 3,156 words · 8h

im still staring at photographs from ernst friedrich's "war against war!" (which is bognas original non ai example) trying to see whether an ai photograph can capture their terror or not, and immediately you can see differences. the photograph called "agricultural worker 36 years of age" shows his entire concern on his face, the protruded swelling of his nose and sloped bruise on the side of his face have details that make you ask how did that particular bruised wave come about, head hit or grass role or explosive emission. why is his eye all so twitchy, how fucked up do you think he thinks what's been done to him is? whereas yurenev’s paintings make me think, well, okay, is there an ai grenade? and this very thought destroys the tragedy inherent in an actual war photograph.

does the ai grenade make someones mouth appear half way through their face, like where their nose is supposed to be? and does it increase their ears three-fold? the comically retarded ai grenade? what is "median, average" about it? are all the dead german soldiers supposed to be captured by a median average of "raised threefold humpty dumpty ears?" do the ears resemble the terror, that is being gathered, subsumed, like, boiled up into the squarehead of a soldier's facial posterior? and the woman with the short weave, a representation of a long-lost lover in the 1940s, a worried home-cook mom? "war actor" to be fair is an actually cool piece, except, it looks like some edgy teen with an illegal skyscraper-climbing hobby printed e.t. the alien on the back of his straight-jacket hoodie and is heading towards the fog in an ambiguous search of some kind. even the image of a battlefield and aftermath, it looks like a really weird fugly alien fly with a protruding neck-head is needlessly heading towards black and white hill-side as horse-men run towards it like a shitty rendition of attack on titan.

it captures literally none of friedrich's terror. the horse corpses there are real, every single one is uniquely distinguishable. it captures three characteristics that yurenev doesn't: 1) the seeming terrifying illusion of infinity, broken by the equally awful realization that, even when it isn't infinite, it's one too many to count or care about 2) desolation not as an explosion or a feeling but an actual landscape that is navigable, searchable, specific in its "thisness" 3) overly specific ruins generalized and ruined by a random arbitrary hanging, a random burial with a decorated cross on it, or a mutilated horse or someone's awfully long legs protruding in the middle of the landscape. if any ai generation model attempts this, it will "exactify" or "generalizify" this depiction, it will over-determine the legs precisely when they need to be under-determined, it will fail in realism by trying too hard to be real.

the specificity of a photo is it's indexability. it's not terrifying because you know an actual person suffered those injuries, but because the generalization of that suffering can extend outwards to capture that exact feeling only because the specificity already "imprints" the aura of that violence outwards due to its essence. now, if in the future an ai can perfectly index a specific photographic feeling, does that mean it's fully original, and therefore a valid ghost-feeling? no, because even if it transfers that feeling, it doesn't hold it. but it does mean it has achieved the halo of that exact feeling. aura is uniquely taggable, it depends on the medium as an inherent experience, artificialization surpasses it. the halo doesn't, but getting to the point where the halo is identical to the index of a photo requires so much more than similarity, than accidents (the inherent accidents (accidentability) of all so-far synthesized intelligence).

this is the problem of method, and we will re-discover it later in this review. but, take any passage from laruelle's non-photography. literally any single passage, they all confirm this exact idea, it is not the subject being represented, but the identity of the subject as the one (the one that remains itself), check it out: "the power-of-semblance of the photo - its power to (re)semble - is power-of-presentation (of) identity, but a power that lets it be as identity, without hybridizing itself with it or degrading it in an image. doubtless, photo-being presems, or even is the very presentation of the one;"

hostfrançois laruelle, non-photography.

this political tangent that is about to come up is necessary, but only two skippable paragraphs. it's necessary because without understanding how this type of art is supposed to interact with us, there's no point in talking about war at all. bogna could have stuck to researching how ai depicts kitchen environments and utensils, but there's a reason she didn't. war is profitable, even for artists trying to sentimentalize it. it's okay to make profit out of it, but it's about how you want to position what you're doing.

lets say bogna really authentically decided to talk about this to an actual small niche public of real readers of theory and not anyone who has stumbled on her name accidentally. lets say this wasnt political at all, and yurenev was not selling prints of the ai-generated imagery book for up to 5000 dollars a copy on stripe, directly deliverable to your house, which includes a copy of his great granduncle who was lost during ww2. lets say that this is okay, that thats a valid price. that maybe purchasing the work is an act of support for war refugeeism or something like that.

its funny, but tragically so. in the "materials" section it says shit like -102 sheets of vellum, -original polaroid, and then, yeah, actually: "reproduction of postcard from naum lipkin, my granduncle who disappeared during ww2."

i don't understand, what makes that a separate part of the material. does that drive the cost up or down? like oh okay, i put my dead grandpa in here, i gotta make it worth a few extra hundred books just cause of that. or the other way around, like, since im sharing something so valuable, i may as well make it reachable? (genuinely, it's being put up as "material", it doesnt come as its own separate category called "super valuable priceless additions to the art" or something).

screw it, the cost is the cost, but why is there a copy with some special type of plate that inflates the cost five times? obviously the artist is being genuine, its a mix of prestige economy (inflating so that rich people think its important to buy), natural artist inflation (only few people will buy it anyways) and cost of materials probably being way higher per book than normal publisher print. also hes selling and promoting in nyc so theres local market inflation too.

i mean i understand artistically why. look, this really isn't about the cost. it's about, the pretentions that involve bogna making a post like this, the work of art being shown in this way. because truthfully, there's nothing deep or sentimental about shit like this. bogna knows this, it's not like shes fetishizing it. she travels to kyiv to see the war ruins, she's deeply embedded in the culture that makes this interpretable. she's no eyel weizmann, but her connections between ai imagery and photography are genuinely decent.

hostbogna konior, war in the age of infinite evidence.

the funniest thing is, these generative images are described by bogna as discriminating inwards, look at the generation process: "in this method, two neural networks—one that generates images and another that judges them—compete with each other. the generator learns to create images that seem real, while the discriminator tries to spot the fakes. typically, this process helps the generator strike a balance between realism and originality." so, you're caught in a bubble where one software has one set of ideas about how something is generated, and another that's just going off on its bullshit. i'm not here to argue about actual ai software constraints, i don't care if it really generated e.t. the alien or if its an accurate depiction of a thousand bruised soldiers. but the method of thinking already constrains the idea behind the generation. when less artificial (dare i say human, but i do dare, because it isn't just human of course) methods of "generation" come about with something, there isn't a democratic voting process on what feels artificial.

of course, photography has a mechanical constraint, even what is faked and evidentialized looks overly naturalized. at least ai-generated imagery returns us to the world of painting, and bogna is right to point this out. "photography and ai might seem diametrically opposed. where one is considered the gold standard of evidence, the other is accused of “hallucinations” and falsifying reality. yet photography, too, was once dismissed as unnatural, blasphemous, and mechanical." but, the method of discrimination that cultivates originality, even if our brain actually does discriminate, override, force decisions and reason linearly to an extent, still is not linear. more importantly, it isnt divided across a false binary competitive assumption, attempting to synthesize a result. an original and realistic depiction of something is booming with fruition. fruition is a kind of halo. it isn't an aura, an aura is the mystified post-effect of a real "labour-event" so to say.

but a halo is a virality, a hype, a fruition into a charged environment where " the world-interior of capital cultivates the alien as estranged but distant" (my own definition). ai produces this same process of alienated relations, where fruition is seen as the ability for something to be itself enough to count. a war photograph trying to be a war photograph is already doing enough, but ai trying to be a war photograph is never doing enough, because it isn't charged with any of the openings or alienations that charge up its fruitive possibility. whereas, an ai photograph trying to coincidentally replicate a form of differential divinity, that is already possible in its form. the problem of the ai grenade becomes real in the art depicting it, but it isn't a valid problem outside of itself. the alienation is always global in such a halology. the war photograph cannot participate in the kind of fruition that makes the alienated condition of an ai-photo necessary, the same way the ai-photo cannot enfruit what a war-photograph already exists within the context of. you can feel that the halo is weakening when the subject is misplaced within the hierarchy of alienation.

grafthalo / halology.

its supposed to mimic a bunch of disfigured faces moulding/transforming under war-worn clothes like hats and helmets and uniforms, but there isnt a language of terror there because we dont know what giant ears would even mean medically for someone, but we know what a half destroyed face is and how annoying and fucked up that person will be to talk to and be around after that, and war photographs are scary because of that reason. the "median" of terror doesnt describe real events, it doesnt seem to be able to capture the terror of apocalyptic emptiness, the seeming infinity of scattered corpses, what it can do is bastardize silhouettes, but thats a type of positive perversion that is nowhere near the type of terror photographs capture, and as a medium its not even supposed to.

the "realism" of photographs isnt mitigated simply because bogna says that some guy was conscripted by the government to make propaganda images of uplifted soldiers, just like how when you look at a painting depicting war, you're not looking for war porn the same way you'd look for it in a photograph, cause you'd be looking for a different thing. in yurenev, i wouldn't look for an authentic portrayal of a soldiers fear in any single moment, but i would look for an abstract feeling of bureaucratic uneasiness that a soldiers figure could have, cause thats what the machine is built to stick together and process, it isnt built on identifying or building hypercomplex emotion or particular details of wounds. and even if it was it couldnt do the rest of what makes a photograph worthwhile. and all of them share an aspect of non-artistry, the photo captures whats already there, the painting illuminates or highlights what doesnt even exist, and the ai photograph, in attempting to capture the median, turns it into a stupid cartoon that is trying to capture something real but creating an entirely new dimension of reality.

a painting has no such constraints, yet, if we could exist in a world where painting is the true evidentiality of reality, then, bogna's dream becomes closer to a reality we once actually did live in. look at how she speaks about this issue:

the issue is, the way she talks about it just isn't the reality of what it captures, and the reality of how it's being presented isn't the reality of war. bogna and yurenev are war fetishists, the likes of which céline and jünger would have thought of as evil genius, and they're doing it practically effortlessly. hell, i am one too, we all are under modernity. even the supposed war veterans that saw something supposedly deep in yurenev's depictions. "“this is terrifying,” one says of the ai-generated portraits. “this is a disfigured face, ears all popped out, [wearing] a soldier’s hat.” “a burned face!” another shouts out." and then they all run out of tokens and dissapear or something along those lines. and the real veterans are somewhere dead stashed under a graveyard and there's a giant machine called the ai-tractor that tries to create energy for ai by snipping grassyards near graveyards. i don't care that i'm dismissing veterans reactions. this isnt about how someone can react or feel about something.

they'd probably get ptsd from war if their light bulb broke in a way that resembles a gun firing, it's self obsessive. this doesn't validate or invalidate ai-as-terror-able. what it actually means is that, it's possible to discover a memory of trauma in them, the same way it's possible to discover the same taste in a different meal. tell you what, we'll torture a bunch of random people whilst showing them ai photos, then call them ai-veterans. and in the future, where everything is ai, we'll place them on the last ai-free place in the world, but have a bunch of people create real drawings of figures mutating. their world will be so perfectly still (analog: traumaless) that they'll associate any disfiguration at all as artifical fuckery, and therefore as violent to the nature of their being. war is not violent because it's realistic, only because it's happening. the realism is an accidental property, like how when god decided to give specific colors to specific objects, he only mechanized the functional "why" of the color, but likely not its teleological counter-point (obfuscating example, but an important one).

i'm not saying this to say that they are artificial. they're worse than artificial - they're an entirely different art form that has nothing to do with representing the terror of war. and that's a great thing, but it isn't the terror of war. the terror of war cannot be captured in the idea of a captain's face, only in the memory that the captain's face is somehow the face of your mother and bosses commands to you, and the real human nagging that comes with that. and the recognition in the eyes of a real human in a photograph that that type of terror is immanent, not too far away from you at any point, that those human experienced are shared, no, dispersed!

look, i know she's not saying it's supposed to capture the particular horror of photographic evidence, but that it can be a mode that reveals something new about it. but what i'm saying is that its faux presentation reveals, in fact, less, about memory, trauma and impossibility than the photos do. it actually crushes whatever they do reveal about them, somewhat! it does other interesting fetishizations, but not these.

yes, they are mechanized (mechanized meaning, the general problem in all of art where some form of artistic legitimacy or agency is automatically removed by the form of the medium, or rather, the historical epistemological modus that made approaching it differently temporarily impossible under its own operational constraints, and bogna is right to point out how photography was seen the same way. wait, was alchemy seen the same way? or the art forms that preceded painting, when painting eventually came out?

"in the synthetic version, the portraits do not seem to depict any specific individual." and this is because the photographs are machinically cultivated fetish objects. bogna ends up calling them monstrous, terrifying, reality-depicting, mediocre in their gruesome reality. these are terrible labels, because the photographs are, really, jokes. they are half-built comedies of war, and that's okay because there's still something interesting in that.

hostbogna konior, war in the age of infinite evidence.

so how do we escape the mechanical assumption that makes ai photography stuck in being unable to capture the vibe of a war photograph, or even its median (it clearly is not its median in any real way, its its derived median, not its successful median, not even painting could succeed in that. a real median would be a type of hyper-photo that captures many events at once yet somehow still comes up with one photo, one photo that doesnt look like the events are stitched together but somehow actually compromised or captured). we escape it, and i proposed this to my friend after reading him bogna's essay, that we need to create a subjective reality where evidence is actually taken seriously by conventional standards. someone's slightly mythical depiction of an event becomes a reality. actually, this already a real thing probably to an extent, like maybe the recent trump shooting, or mangione scandal, or the recent mass propagandized denial of the holocaust on instagram, or whatever else. but these are light politicizations, or mass media hallucinations, not heavy subjective disfigurations that are taken as above actual visible evidentiality.

grafthyper-photo / evidentiality.

it's not just evidentiality-as-visibility that would be questioned, but the rhetoric and expected value of the medium itself. photographs would be forced to look like awfully assembled 1450's depictions of cathedrals, that would be the only allowed photo you can take. whereas paintings, artists would be forced to design post-hyper-realistic paintings where, essentially, the disfigured face of a soldier is actually heightened, like, the wound is made to be a portal where his parents are also killed inside of it, and some type of mass hallucination sponsors this as the real depiction. it wouldnt necessarily be fake, because analogically maybe the war would actually lead to them being killed in this way, and the wound would be seen as the real source? psychotic accelerationism could make the mechanization of art actually unlock art's real possibility. but that won't happen because some guy is cross-generating depictions of people wearing hazmat suits backwards, we're gonna need more of that.

colophon
register
counter
host
bogna konior, war in the age of infinite evidence: on ai-generated war photography
mode
orbit
trace
15 equal word slices · ~210 words each
contact
4
departure
5
distortion
3
return
3
layers
host / deformation / graft
labour
3h active · 5h passive · 8h total
paid
$0
full schematic & video →the counter register →