{"id":1317,"date":"2025-07-29T12:06:56","date_gmt":"2025-07-29T10:06:56","guid":{"rendered":"https:\/\/laurentchani.com\/?p=647"},"modified":"2025-09-15T17:02:05","modified_gmt":"2025-09-15T15:02:05","slug":"lia-recommended-procedure-for-image-generation-2","status":"publish","type":"post","link":"https:\/\/laurentchani.fr\/en\/procedure-recommandee-par-lia-pour-la-generation-dimage-2\/","title":{"rendered":"AI Recommended Procedure for Image Generation"},"content":{"rendered":"
\"A<\/a><\/figure>\n\n\n\n

Here is the precise procedure<\/strong> that suggests theGenerative AI<\/strong> to obtain a image creation<\/strong> the most faithful<\/strong> possible to what a describes text<\/strong>. She maximizes<\/strong> the chances of getting a conforming image<\/strong> to what theauthor<\/strong> has in mind, especially when we reach the limits of language<\/strong> to describe some details<\/strong>. L'AI<\/strong> also explains the technical difference<\/strong> between generate a sketch<\/strong> and produce a hyperrealistic image<\/strong>.<\/p>\n\n\n\n

In additional information<\/strong>, this approach<\/strong> provided by theAI<\/strong> that I use refers to the description<\/strong> of theextract<\/strong> demon text<\/strong> concerning the conference room<\/strong>.<\/p>\n\n\n\n

1. Optimized procedure to guide AI<\/strong><\/p>\n\n\n\n

This method<\/strong> works with all Specialized AI<\/strong> in the image generation<\/strong>, but it is particularly adapted<\/strong> when we aim for a maximum loyalty<\/strong> to words<\/strong>.<\/p>\n\n\n\n

Step 1 \u2013 Define the end goal<\/strong>
\u2022 Decide on the image type<\/strong> (conceptual sketch<\/strong>, illustration<\/strong>, 3D rendering<\/strong>, hyperrealistic photo<\/strong>).
\u2022 Specify the purpose<\/strong> : concept art<\/strong> to help visualize a scene<\/strong>, Or almost definitive visual<\/strong> ?
\u2022 List the mandatory elements<\/strong> (eg: three concentric levels<\/strong>, changing celestial vault<\/strong>, central arena<\/strong>).<\/p>\n\n\n\n

Step 2 \u2013 Break it down into sub-steps<\/strong>
Rather than asking for everything at once, proceed by layers<\/strong> :<\/p>\n\n\n\n

    \n
  1. Raw structure<\/strong> (architecture, volumes, arrangement of major elements).<\/li>\n\n\n\n
  2. Bright and colorful atmosphere<\/strong> (changing sky, futuristic or subdued atmosphere).<\/li>\n\n\n\n
  3. Materials and textures<\/strong> (leather, metal, screen, glass).<\/li>\n\n\n\n
  4. Details and characters<\/strong> (postures, clothing, orientation).
    Each stage<\/strong> validates a consistency<\/strong> before moving on to the next one.<\/li>\n<\/ol>\n\n\n\n

    Step 3 \u2013 Use clear, hierarchical wording<\/strong>
    \u2022 Start with themain element<\/strong> (e.g., \u201cmonumental circular conference room with three concentric levels\u201d).
    \u2022 Add the secondary details<\/strong> (leather tables, stairs, central arena).
    \u2022 Finally, specify the visual style<\/strong> (\u201cphotographic hyperrealism, soft light, wide-angle lens\u201d).
    \u2022 Avoid piling up contradictory adjectives<\/strong> ; one word too many can disturb theAI<\/strong>.<\/p>\n\n\n\n

    Step 4 \u2013 Iterate intelligently<\/strong>
    \u2022 After each generation<\/strong>, note what works and what drifts.
    \u2022 Correct by negation<\/strong> (\u201cwithout a fourth row of tables\u201d, \u201cwithout additional characters\u201d).
    \u2022 Repeat the same conversation thread<\/strong> to preserve the memory<\/strong>, but launch a new prompt<\/strong> if theAI<\/strong> gets stuck in repetitive errors<\/strong>.<\/p>\n\n\n\n

    Step 5 \u2013 Lock the composition<\/strong>
    Once thestructural image<\/strong> validated:
    \u2022 Ask for the same one again scene<\/strong> in high resolution<\/strong> (1024\u00d71024 or higher).
    \u2022 Add the finishes<\/strong> (\u201crealistic textures, patinated leather, metallic reflections\u201d).
    \u2022 For an almost photographic<\/strong>, mention \u201cphotorealistic, 8k, ultra-detailed, cinematic lighting\u201d (or French equivalents, but English is often better understood).<\/p>\n\n\n\n

    2. Difference between sketching and hyperrealism<\/strong><\/p>\n\n\n\n

    Sketch (concept art)<\/strong>
    \u2022 Aim<\/strong> : establish the general idea, the volumes, the composition.
    \u2022 Advantage<\/strong> : L'AI<\/strong> can improvise and be creative<\/strong> ; the blur is not bothersome.
    \u2022 Consumption<\/strong> : fast, inexpensive in computation.
    \u2022 Possible to request simple lines<\/strong>, a \u201chand drawing\u201d, \u201csketch\u201d, \u201cminimalist concept art\u201d style.<\/p>\n\n\n\n

    Hyperrealistic image<\/strong>
    \u2022 Aim<\/strong> : rendered almost photographic<\/strong>, fine textures<\/strong>, precise shadows<\/strong>.
    \u2022 Inconvenience<\/strong> : every poorly described detail can be misinterpreted (anatomical errors, added or missing objects).
    \u2022 Request more precision<\/strong> and moreiterations<\/strong> to correct the inconsistencies<\/strong>.
    \u2022 There light<\/strong> and the point of view<\/strong> (high angle, low angle, wide angle) become crucial to avoid flat results<\/strong>.<\/p>\n\n\n\n

    3. My strategic advice<\/strong><\/p>\n\n\n\n

    Always start with a conceptual sketch<\/strong>, validate the composition<\/strong>, then go up in loyalty<\/strong> towards thehyperrealism<\/strong>. Wanting a direct perfect photo<\/strong> often leads to frustrations<\/strong>, because theuser<\/strong> has not yet locked the basic elements<\/strong>.<\/p>","protected":false},"excerpt":{"rendered":"

    Voici la proc\u00e9dure pr\u00e9cise que sugg\u00e8re l\u2019IA g\u00e9n\u00e9rative pour obtenir une cr\u00e9ation d\u2019image la plus fid\u00e8le possible \u00e0 ce que d\u00e9crit un texte. Elle maximise les chances d\u2019obtenir une image conforme \u00e0 ce que l\u2019auteur a en t\u00eate, surtout quand on atteint les limites du langage pour d\u00e9crire certains d\u00e9tails. L\u2019IA explique aussi la diff\u00e9rence […]<\/p>\n","protected":false},"author":1,"featured_media":836,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_uag_custom_page_level_css":"","footnotes":""},"categories":[10],"tags":[],"class_list":["post-1317","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ia"],"blocksy_meta":[],"uagb_featured_image_src":{"full":["https:\/\/laurentchani.fr\/wp-content\/uploads\/2025\/07\/big-daddy-17-1.jpg",512,512,false],"thumbnail":["https:\/\/laurentchani.fr\/wp-content\/uploads\/2025\/07\/big-daddy-17-1-150x150.jpg",150,150,true],"medium":["https:\/\/laurentchani.fr\/wp-content\/uploads\/2025\/07\/big-daddy-17-1-300x300.jpg",300,300,true],"medium_large":["https:\/\/laurentchani.fr\/wp-content\/uploads\/2025\/07\/big-daddy-17-1.jpg",512,512,false],"large":["https:\/\/laurentchani.fr\/wp-content\/uploads\/2025\/07\/big-daddy-17-1.jpg",512,512,false],"1536x1536":["https:\/\/laurentchani.fr\/wp-content\/uploads\/2025\/07\/big-daddy-17-1.jpg",512,512,false],"2048x2048":["https:\/\/laurentchani.fr\/wp-content\/uploads\/2025\/07\/big-daddy-17-1.jpg",512,512,false],"trp-custom-language-flag":["https:\/\/laurentchani.fr\/wp-content\/uploads\/2025\/07\/big-daddy-17-1-12x12.jpg",12,12,true],"mailpoet_newsletter_max":["https:\/\/laurentchani.fr\/wp-content\/uploads\/2025\/07\/big-daddy-17-1.jpg",512,512,false]},"uagb_author_info":{"display_name":"admin5261","author_link":"https:\/\/laurentchani.fr\/en\/author\/admin5261\/"},"uagb_comment_info":0,"uagb_excerpt":"Voici la proc\u00e9dure pr\u00e9cise que sugg\u00e8re l\u2019IA g\u00e9n\u00e9rative pour obtenir une cr\u00e9ation d\u2019image la plus fid\u00e8le possible \u00e0 ce que d\u00e9crit un texte. Elle maximise les chances d\u2019obtenir une image conforme \u00e0 ce que l\u2019auteur a en t\u00eate, surtout quand on atteint les limites du langage pour d\u00e9crire certains d\u00e9tails. L\u2019IA explique aussi la diff\u00e9rence…","_links":{"self":[{"href":"https:\/\/laurentchani.fr\/en\/wp-json\/wp\/v2\/posts\/1317","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/laurentchani.fr\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/laurentchani.fr\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/laurentchani.fr\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/laurentchani.fr\/en\/wp-json\/wp\/v2\/comments?post=1317"}],"version-history":[{"count":4,"href":"https:\/\/laurentchani.fr\/en\/wp-json\/wp\/v2\/posts\/1317\/revisions"}],"predecessor-version":[{"id":4186,"href":"https:\/\/laurentchani.fr\/en\/wp-json\/wp\/v2\/posts\/1317\/revisions\/4186"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/laurentchani.fr\/en\/wp-json\/wp\/v2\/media\/836"}],"wp:attachment":[{"href":"https:\/\/laurentchani.fr\/en\/wp-json\/wp\/v2\/media?parent=1317"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/laurentchani.fr\/en\/wp-json\/wp\/v2\/categories?post=1317"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/laurentchani.fr\/en\/wp-json\/wp\/v2\/tags?post=1317"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}