{"id":941,"date":"2024-04-22T07:33:43","date_gmt":"2024-04-22T06:33:43","guid":{"rendered":"http:\/\/databaz.org\/xtrm-art\/?p=941"},"modified":"2024-04-23T18:26:19","modified_gmt":"2024-04-23T17:26:19","slug":"completion-le-reel-sera-son-double","status":"publish","type":"post","link":"https:\/\/databaz.org\/xtrm-art\/?p=941","title":{"rendered":"COMPLETION (le r\u00c3\u00a9el sera son double)"},"content":{"rendered":"<p>\/\/ French &#8211; English below<\/p>\n<p>&nbsp;<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" src=\"http:\/\/databaz.org\/xtrm-art\/wp-content\/uploads\/image\/premiere-photo-homme-paris_completionJPG.png\" width=\"600\" height=\"431\" align=\"left\" alt=\"\" \/>Notre m&eacute;moire, parfois nous le regrettons, est fragmentaire. Pendant des si&egrave;cles, il n&rsquo;y eut que la r&eacute;tention parla conscience et ses potentialit&eacute;s mn&eacute;siques pour se souvenir en image d&rsquo;un pass&eacute;. Que l&rsquo;on avait v&eacute;cu, ou bien que l&rsquo;on avait entendu. La photographie a chang&eacute; cela. Alors que la m&eacute;moire eid&eacute;tique est tr&egrave;s rare chez l&rsquo;&ecirc;tre humain, avec la photiographie, on a pu faire des ponctions dans le r&eacute;el. Regarder une photo c&rsquo;est revoir un fragment (temprorel et spatial du pass&eacute;) de ce qui a eu lieu. <br \/>\nMais il y a toujours du hors-champ. Le hors-champ c&rsquo;est ce qui a &eacute;chapp&eacute; &agrave; la captation et la m&eacute;moire technique. On peut ainsi penser la photographie comme une proth&egrave;se technique de notre m&eacute;moire. Une extension. Cependant cette m&eacute;moire est limit&eacute;e, elle est aussi fragmentaire. Elle est un segment infime de l&rsquo;espace et du temps. Une portion.<\/p>\n<p>&nbsp;<\/p>\n<p>Avec Compl&eacute;tion, il m&rsquo;est apparu n&eacute;cessaire d&rsquo;interroger la question du hors-champ &agrave; l&rsquo;&egrave;re des IA : si le hors champ de ce qui a &eacute;t&eacute; m&eacute;moris&eacute; est ce qui a disparu, que je ne peux qu&rsquo;imaginer, comment peut-il &ecirc;tre abord&eacute; et qu&rsquo;est-ce qu&rsquo;il ouvre &agrave; partir des inf&eacute;rences des IA. Comment une IA peut-elle imaginer un hors-champ ? La compl&eacute;tion, au niveau des logiciels, c&rsquo;est la possibilit&eacute; automatis&eacute;e de compl&eacute;ter des donn&eacute;es partielles. Ici, la donn&eacute;e fournie est une photographie, cadr&eacute;e, il s&rsquo;agit alors de compl&eacute;ter son hors-champ, de produire son contexte.<\/p>\n<p>&nbsp;<\/p>\n<p>Pour la s&eacute;rie Compl&eacute;tion, j&rsquo;ai pris des images historiques. Dont je n&rsquo;ai pas &eacute;t&eacute; le t&eacute;moin. Dont peut-&ecirc;tre plus personne ici, vivant, ne pourra de m&ecirc;me t&eacute;moigner maintenant ou plus tard. Et j&rsquo;ai utiliser une IA qui fonctionne sans prompt, pour qu&rsquo;elle imagine &agrave; partir de la sc&egrave;ne les hors-champs. Le fait que l&rsquo;IA fonctionne sans aucun prompt est important, car cela met en avant un processus d&rsquo;imagination artificielle sans vecteur d&eacute;terminant linguistique de la part de l&rsquo;utilisateur. L&rsquo;IA n&rsquo;a que le contexte photographique. Ici, l&rsquo;IA fonctionne en deux temps : tout d&rsquo;abord il y a une perception machine du contexte et une analyse, ensuite, elle produit et compl&egrave;te l&rsquo;image d&rsquo;un point de vue g&eacute;n&eacute;ratif. Sans le prompt, elle inf&egrave;re la suite (le hors-champ) de l&rsquo;image sans autre r&eacute;f&eacute;rence que ce qu&rsquo;elle en comprend. Compl&eacute;tion ouvre ainsi &agrave; une r&eacute;flexion  &agrave; une autonomie de l&rsquo;imaginaire. Alors qu&rsquo;avec le prompt, on dirige (on induit une vectorialit&eacute; s&eacute;miotique et s&eacute;mantique dans le mod&egrave;le), ici on demande seulement &agrave; l&rsquo;IA d&rsquo;imaginer &agrave; partir de sa perception ce que pourrait &ecirc;tre le hors champ. Avec Midjourney, ou bien Dall-E 2, l&rsquo;utilisateur est comme un ma&icirc;tre qui donne le sujet que doit r&eacute;aliser un &eacute;l&egrave;ve. Compl&eacute;tion propose &agrave; une imagination artificielle de g&eacute;n&eacute;rer, sans autre processus que sa propre compr&eacute;hension non induite, le contexte d&rsquo;une image.<\/p>\n<p><strong><em>&nbsp;Boulevard du Temple<\/em><\/strong> photographie en daguerr&eacute;otype r&eacute;alis&eacute;e &agrave; Paris en 1838 par l&#8217;inventeur du proc&eacute;d&eacute;, Louis Daguerre. Ce serait la premi&egrave;re ou l&#8217;une des premi&egrave;res photographie de l&#8217;histoire de l&#8217;humanit&eacute;. Et surtout la premi&egrave;re photographie o&ugrave; il. aurait un &ecirc;tre humain (un cireur de chaussure). La deuxi&egrave;me, est une des photographies de Amstrong sur la lune. Le premier pas de l&#8217;homme sur notre satellite. La troisi&egrave;me est celle du G&eacute;n&eacute;ral Degaulle au Qu&eacute;bec. La quatri&egrave;me est celle du jour de l&#8217;assassinat de JFK.<br \/>\n1838 &#8211; Louis Daguerre r&eacute;alise la premi&egrave;re photographie : Boulevard du Temple. Premi&egrave;re captation m&eacute;canique, premi&egrave;re m&eacute;moire technique de l&#8217;image.  Plus de t&eacute;moins de ce qui apparaissait hors-champ.  La s&eacute;rie Compl&eacute;tion se pr&eacute;sente comme une compl&eacute;tion du hors-champ par une IA (et ceci sans la d&eacute;termination du moindre prompt). L&#8217;IA, &agrave; partir seulement de ce qu&#8217;elle peut inf&eacute;rer selon sa perception algorithmique du donn&eacute; du pixel, vient compl&eacute;ter la photographie, vient constituer ce hors-champ impossible.  La photographie g&eacute;n&eacute;r&eacute;e fait 10000 \/ 7841 pixels en 300ppp<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" src=\"http:\/\/databaz.org\/xtrm-art\/wp-content\/uploads\/image\/Degaule_Montreal2.png\" width=\"800\" height=\"523\" border=\"500\" align=\"left\" alt=\"\" \/>Degaulle le 24 &#8211; juillet 1967 &agrave; Montr&eacute;al &#8211; La photographie est serr&eacute;e. Le g&eacute;n&eacute;ral Degaulle est au premier plan. Dans le paroxysme de son &eacute;loquence, il &eacute;nonce avec force : &quot;Vive le Qu&eacute;bec libre&quot;. 15000 personnes &agrave; le regarder.  Les IA g&eacute;n&eacute;ratives sont entrain&eacute;es avec des sets d&#8217;images et des classements linguistiques qui permettent une g&eacute;n&eacute;ration selon une inf&eacute;rence statistique. <br \/>\nEn ce sens, l&#8217;IA g&eacute;n&eacute;rative ne recherche pas ce qui fait rupture, mais ce qui correspond plus ou moins avec des r&eacute;currences, des r&eacute;p&eacute;titions, analys&eacute;es lors des analyses sets d&#8217;images. De sorte que les mod&egrave;les courants qui sont con&ccedil;us (Midjourney, Dall E 2, Gemini) sont normalis&eacute;s. S&#8217;il y a surprise dans l&#8217;image (ou une sorte de r&eacute;alit&eacute; qui rompt l&#8217;ordinaire) cela provient de l&#8217;input de l&#8217;utilisateur. Hormis cela, l&#8217;IA produit ce qui lui para&icirc;t statistiquement le plus probable. <br \/>\nCOMPLETION utilise une IA g&eacute;n&eacute;rative sans prompt. Donc l&#8217;IA ne se fonde que sur le contexte des donn&eacute;es pixels pour d&eacute;terminer la compl&eacute;tion de l&#8217;image. Il n&#8217;y a aucune forme d&#8217;influence de la part de l&#8217;utilisateur. C&#8217;est en ce sens que l&#8217;IA ne peut envisager d&#8217;&eacute;v&eacute;nement. L&#8217;&eacute;v&eacute;nement pour reprendre Alain Badiou : &ldquo;ouvre la possibilit&eacute; qu&rsquo;une v&eacute;rit&eacute; in&eacute;dite surgisse dans un monde donn&eacute;&rdquo;. C&#8217;est ce qui vient faire effraction dans l&#8217;&ecirc;tre (qui est gouvern&eacute; par Badiou par le math&egrave;me, donc l&#8217;ordre logis-math&eacute;matique). L&#8217;&eacute;v&eacute;nement c&#8217;est ce qui fait histoire. Or, si l&#8217;IA g&eacute;n&egrave;re &agrave; partir d&#8217;une induction statistique, sans autre influence elle va produire du monde selon l&#8217;&ecirc;tre au sens de Badiou et non pas l&#8217;&eacute;v&eacute;nement. <br la=\"\" photographie=\"\" de=\"\" degaulle=\"\" est=\"\" serr\u00c3\u00a9e.=\"\" on=\"\" ne=\"\" per\u00c3\u00a7oit=\"\" pas=\"\" les=\"\" personnes.=\"\" et=\"\" pour=\"\" son=\"\" imagination=\"\" peut=\"\" imaginer.=\"\" comme=\"\" elle=\"\" imaginer=\"\" ce=\"\" le=\"\" quebec=\"\" ou=\"\" bien=\"\" libert\u00c3\u00a9.=\"\" imagine=\"\" seulement=\"\" ville=\"\" selon=\"\" plus=\"\" forte=\"\" probabilit\u00c3\u00a9=\"\" au=\"\" vue=\"\" sa=\"\" perception=\"\" seul=\"\" face=\"\" silence=\"\" \/><br \/>\n<br \/>\nCOMPLETION &#8211; 3 &#8211; John F. Kennedyen voiture lieu le i 22 novembre 1963 &#8211; L&#8217;&eacute;v&eacute;nement historique ici n&#8217;aura pas lieu, n&#8217;aura pas eu lieu. Car l&#8217;espace compl&eacute;t&eacute; du hors champs n&#8217;est pas Dallas. JFK est dans une autre ville, d&#8217;un autre temps. La voiture roule, mais personne ne regarde. Le monde est vide. Il n&#8217;est pas vid&eacute;, car personne n&#8217;est jamais venu. Car cette voiture implique un contexte sans histoire, car l&#8217;imagination qui imagine ce r&eacute;el-l&agrave;, par sa logique inductive, par sa logique statistique, ne peut imaginer ce qui viendrait rompre la cha&icirc;ne logique d&#8217;une ville statistiquement r&eacute;guli&egrave;re. Seul, un tueur, peut imaginer de rompre la ville r&eacute;guli&egrave;re et le chemin r&eacute;gulier d&#8217;un pr&eacute;sident qui parade dans une ville. <br \/>\n.   \/\/english<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" src=\"http:\/\/databaz.org\/xtrm-art\/wp-content\/uploads\/image\/premier_homme_luneJPG.jpg\" width=\"600\" height=\"337\" align=\"left\" alt=\"\" \/>Our memory, sometimes we regret it, is fragmentary. For centuries, there was only retention by consciousness and its memory potential to remember the past in images. That we had experienced, or that we had heard. Photography changed that. While eidetic memory is very rare in humans, with photography, we have been able to make punctures from reality. Looking at a photo is seeing a fragment (temporal and spatial of the past) of what took place. But there is always something off-camera. The off-camera is what has escaped capture and technical memory. We can thus think of photography as a technical prosthesis of our memory. An extension. However, this memory is limited, it is also fragmentary. It is a tiny segment of space and time. A portion.<\/p>\n<p>&nbsp;<\/p>\n<p>With Completion, it seemed necessary to me to question the question of off-camera in the age of AI: if the off-camera of what was memorized is what has disappeared, which I can only imagine, how can it be addressed and what does it open up from AI inferences. How can an AI imagine something off-camera? Completion, at the software level, is the automated possibility of completing partial data. Here, the data provided is a photograph, framed, it is then a question of completing its off-camera, of producing its context.<\/p>\n<p>&nbsp;<\/p>\n<p>For the Completion series, I took historical images. Which I did not witness. To which perhaps no one here, alive, will be able to bear witness now or later. And I used an AI that works without prompts, so that it can imagine things off-camera from the scene. The fact that the AI \u00e2\u20ac\u2039\u00e2\u20ac\u2039operates without any prompt is important, because it highlights a process of artificial imagination without a linguistic determining vector on the part of the user. The AI \u00e2\u20ac\u2039\u00e2\u20ac\u2039only has photographic context. Here, the AI \u00e2\u20ac\u2039\u00e2\u20ac\u2039works in two stages: first there is a machine perception of the context and an analysis, then it produces and completes the image from a generative point of view. Without the prompt, it infers the rest (off-camera) of the image without any other reference than what it understands. Completion thus opens up a reflection on the autonomy of the imagination. Whereas with the prompt, we direct (we induce a semiotic and semantic vectorality in the model), here we only ask the AI \u00e2\u20ac\u2039\u00e2\u20ac\u2039to imagine from its perception what the off-camera could be. With Midjourney, or Dall-E 2, the user is like a teacher who gives the subject that a student must complete. Completion allows an artificial imagination to generate, without any process other than its own uninduced understanding, the context of an image.<\/p>\n<p>The two photographs proposed in this article: Boulevard du Temple daguerreotype photograph taken in Paris in 1838 by the inventor of the process, Louis Daguerre. It would be the first or one of the first photographs in the history of humanity. The second is one of Amsrong&#8217;s photographs on the moon. Man&#8217;s first step on our satellite.<br \/>\n1838 &#8211; Louis Daguerre made the first photograph: Boulevard du Temple. First mechanical capture, first technical memory of the image. No more witnesses to what appeared off-camera. The Compl&eacute;tion series presents itself as an off-screen completion by an AI (and this without the slightest determination). The AI, based only on what it can infer according to its algorithmic perception of the given pixel, completes the photography, constitutes this impossible off-camera.<\/p>\n<p>Degaulle on July 24, 1967 in Montreal &#8211; The photography is tight. General Degaulle is in the foreground. In the paroxysm of his eloquence, he forcefully stated: &ldquo;Long live free Quebec.&rdquo; 15,000 people watching it.<br \/>\nGenerative AIs are trained with sets of images and linguistic classifications which allow generation according to statistical inference. In this sense, generative AI does not look for what breaks, but what corresponds more or less with recurrences, repetitions, analyzed during image set analyses. So the common models that are designed (Midjourney, Dall E 2, Gemini) are standardized. If there is surprise in the image (or some sort of reality that breaks the ordinary) it comes from user input. Aside from that, the AI \u00e2\u20ac\u2039\u00e2\u20ac\u2039produces what appears statistically most likely to it. COMPLETION uses generative AI without prompting. So the AI \u00e2\u20ac\u2039\u00e2\u20ac\u2039only relies on the context of the pixel data to determine image completion. There is no form of influence from the user.<br \/>\nIt is in this sense that the AI \u00e2\u20ac\u2039\u00e2\u20ac\u2039cannot envisage an event. The event, to quote Alain Badiou: &ldquo;opens the possibility of an unprecedented truth emerging in a given world&rdquo;. This is what breaks into being (which is governed by Badiou by the matheme, therefore the logis-mathematical order). The event is what makes history. However, if AI generates from statistical induction, without any other influence it will produce the world according to being in the sense of Badiou and not the event.<br \/>\nDegaulle&#8217;s photography is tight. We don&#8217;t see the 15,000 people. And for the AI, its imagination cannot imagine them. As she cannot imagine what Quebec or freedom is. She only imagines the city according to the highest probability in view of her perception of the image. And Degaulle is alone facing the silence of space.<\/p>\n<p>&nbsp;<\/p>\n","protected":false},"excerpt":{"rendered":"<p>\/\/ French &#8211; English below &nbsp; Notre m&eacute;moire, parfois nous le regrettons, est fragmentaire. Pendant des si&egrave;cles, il n&rsquo;y eut que la r&eacute;tention parla conscience et ses potentialit&eacute;s mn&eacute;siques pour se souvenir en image d&rsquo;un pass&eacute;. Que l&rsquo;on avait v&eacute;cu, ou bien que l&rsquo;on avait entendu. La photographie a chang&eacute; cela. Alors que la m&eacute;moire [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[11,184,135,163,13,6],"tags":[137,204,208,209,210,182,235],"class_list":["post-941","post","type-post","status-publish","format-standard","hentry","category-article","category-imagination-artificielle","category-intelligence-artificielle","category-photographie","category-theorie","category-une","tag-ai","tag-artificial-imagination","tag-completion","tag-daguerre","tag-histoire-de-la-photographie","tag-imaginairemachine","tag-imagination-artificielle"],"_links":{"self":[{"href":"https:\/\/databaz.org\/xtrm-art\/index.php?rest_route=\/wp\/v2\/posts\/941","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/databaz.org\/xtrm-art\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/databaz.org\/xtrm-art\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/databaz.org\/xtrm-art\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/databaz.org\/xtrm-art\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=941"}],"version-history":[{"count":3,"href":"https:\/\/databaz.org\/xtrm-art\/index.php?rest_route=\/wp\/v2\/posts\/941\/revisions"}],"predecessor-version":[{"id":945,"href":"https:\/\/databaz.org\/xtrm-art\/index.php?rest_route=\/wp\/v2\/posts\/941\/revisions\/945"}],"wp:attachment":[{"href":"https:\/\/databaz.org\/xtrm-art\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=941"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/databaz.org\/xtrm-art\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=941"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/databaz.org\/xtrm-art\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=941"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}