Caricature Prompt Pack : 25 modèles de prompts à copier-coller pour des avatars spécifiques à un emploi
Je vois constamment la même frustration : quelqu'un demande une « caricature de moi au travail » et le résultat est générique, l'emploi n'est pas immédiatement reconnaissable, le visage varie, ou l'image invente des détails supplémentaires. Ce pack de prompts résout ce problème en étant précis là où cela compte (identité + indices d'emploi) et concis là où ce n'est pas le cas.
Préparer avant de coller un prompt
Pour de meilleurs résultats, utilisez une photo de référence nette (face avant, bon éclairage) et, si vous en avez une, une photo « indice d'emploi » (uniforme, configuration de l'ordinateur portable, salle de classe, cuisine, outils, etc.). La plupart des outils prennent en charge la conversion de texte en image ou la transformation d'une photo téléchargée, et le flux de travail d'image de ChatGPT est expliqué dans Créer des images dans ChatGPT ainsi que dans le Aperçu des fonctionnalités d'image ChatGPT.
Si vous générez ou modifiez des images via l'API, il vaut la peine de consulter le Guide de génération d'images OpenAI et le Recette de prompts 'gpt-image-1.5', , car ils montrent comment structurer les prompts et itérer proprement.
Les 5 ingrédients qui font fonctionner « caricature de moi + mon emploi »
Chaque résultat performant utilise les mêmes cinq ingrédients : ancrage d'identité, indices d'emploi, style, appareil photo, et contraintes.
Modèle de prompt universel « Caricature de moi + mon emploi »
Utilisez-le une fois, puis modifiez uniquement le bloc de style. Gardez les indices d'emploi réalistes et vous obtiendrez des résultats qui semblent intentionnels plutôt que « personne de dessin animé aléatoire ».
ROLE: Image generator
INPUT: Use my uploaded reference photo as the identity anchor.
GOAL: Create a [STYLE] portrait of me as a [JOB], clearly readable at a glance.
SCENE: [WORK SETTING]. Include 3–5 job-true props: [PROPS].
WARDROBE: [UNIFORM/CLOTHES] with realistic details.
POSE: [POSE]. EXPRESSION: [EXPRESSION].
FRAMING: [HEAD-AND-SHOULDERS | HALF-BODY | FULL-BODY]. CAMERA: [LENS/ANGLE].
LIGHTING: [LIGHTING]. COLOR: [PALETTE].
CONSTRAINTS:
- Keep my face recognizable and consistent with the reference photo.
- No extra people, no extra limbs, no weird hands.
- No logos, no brand names, no readable private data on screens or badges.
- Clean background; the job props should communicate the profession.
OUTPUT: One final image.
Sélectionneur rapide
| Si vous voulez... | Choisissez ce style | Pourquoi ça marche |
|---|---|---|
| Photo de profil qui semble amicale | Dessin animé 2D | Lignes propres + formes simplifiées restent lisibles dans les petites tailles |
| Humour « Grande tête, petit corps » | Caricature | Exagération délibérée + indices d'emploi = reconnaissance instantanée |
| Figurine de collection / « aspect jouet 3D » | 3D de type Pixar | Éclairage doux + matériaux stylisés donnent une impression de qualité et de raffinement |
| En-tête LinkedIn / ambiance papeterie | Affiche minimaliste | Formes audacieuses + palette limitée donnent l'impression d'être « conçu », pas « généré » |
| Authenticité dessinée à la main | Croquis | La texture du crayon dissimule les petits défauts tout en conservant la ressemblance |
Courtes vidéos de référence
Si vous aimez voir un flux de travail complet avant de commencer à itérer, ces démos sont pratiques et faciles à reproduire :

Source: YouTube

Source: YouTube

Source: YouTube
Les 25 prompts à copier-coller
Emplois utilisés : infirmière, enseignant, développeur de logiciels, chef, pompier. Si votre emploi est différent, remplacez la ligne d'emploi et gardez les accessoires fidèles à la réalité.
Style 1 — Avatar de dessin animé 2D
Visez : contour propre, ombrage plat, expression chaleureuse, accessoires d'emploi nets.
Use my uploaded photo as reference.
Create a 2D cartoon avatar of me as a NURSE on an evening shift.
Scene: hospital corridor near a nurse station, calm atmosphere.
Props: stethoscope, clipboard, sealed medication cup, simple wristwatch.
Wardrobe: clean scrubs, name badge with NO readable text.
Pose: half-body, slight three-quarter turn, friendly confident smile.
Style: clean line art, flat colors, subtle shadow, modern avatar look.
Constraints: no extra people, no logos, no readable data, hands normal.
Use my uploaded photo as reference.
Create a 2D cartoon avatar of me as a TEACHER in a classroom.
Scene: whiteboard background with simple shapes (no readable text).
Props: marker, stack of notebooks, small desk bell.
Wardrobe: smart-casual.
Pose: half-body, one hand gesturing as if explaining.
Style: clean outlines, flat color blocks, slight cheek blush.
Constraints: no students, no brand names, no readable board text.
Use my uploaded photo as reference.
Create a 2D cartoon avatar of me as a SOFTWARE DEVELOPER at my desk.
Scene: cozy workstation, two monitors with abstract UI shapes (no readable code).
Props: mechanical keyboard, coffee mug, sticky notes with blank lines.
Wardrobe: hoodie or simple t-shirt.
Pose: seated, slight lean forward, focused but approachable expression.
Style: clean vector look, flat shading, crisp edges.
Constraints: no company logos, no readable screens, realistic hands.
Use my uploaded photo as reference.
Create a 2D cartoon avatar of me as a CHEF in a modern kitchen.
Scene: stainless counter, warm kitchen lighting.
Props: chef knife (safe pose, blade down), wooden cutting board, herbs, plated dish.
Wardrobe: chef jacket + apron (no logos).
Pose: half-body, holding the plate proudly.
Style: flat colors, clean lines, slightly exaggerated eyebrows for character.
Constraints: no brand names, no readable labels, no extra people.
Use my uploaded photo as reference.
Create a 2D cartoon avatar of me as a FIREFIGHTER.
Scene: fire station interior, equipment racks blurred.
Props: helmet under one arm, radio on shoulder strap, gloves clipped to belt.
Wardrobe: turnout jacket (generic, no department text).
Pose: full-body, confident stance, calm heroic expression.
Style: bold outlines, flat shading, poster-like clarity.
Constraints: no logos, no readable patches, no flames close to face.
Style 2 — Caricature classique
Visez : grande tête/petit corps, exagération amicale, uniforme + outils lisibles.
Use my uploaded photo as reference.
Create a classic caricature of me as a NURSE: big head, small body, warm humor.
Scene: nurse station desk with tidy supplies (no readable patient data).
Props: stethoscope, penlight, folded gloves, small hand sanitizer bottle.
Wardrobe: scrubs + badge without readable text.
Pose: standing with a clipboard, eyebrows slightly exaggerated, friendly grin.
Style: caricature ink lines + soft color wash.
Constraints: respectful tone, no mockery, no extra people.
Use my uploaded photo as reference.
Create a classic caricature of me as a TEACHER.
Scene: classroom corner with books and a globe.
Props: chalk/marker, lesson plan papers, simple pointer.
Pose: big head, small body; one hand raised mid-explanation.
Style: exaggerated facial expression, ink outline, light watercolor shading.
Constraints: no readable text on board, no students, no logos.
Use my uploaded photo as reference.
Create a classic caricature of me as a SOFTWARE DEVELOPER.
Scene: desk with two monitors showing abstract blocks (no readable code).
Props: keyboard, mouse, cable organizer, coffee cup.
Pose: sitting on an oversized chair, big head, playful smirk.
Style: caricature sketch lines + clean colors, slightly oversized glasses if present.
Constraints: no company branding, no readable screen content.
Use my uploaded photo as reference.
Create a classic caricature of me as a CHEF.
Scene: kitchen pass with warm light.
Props: sauté pan, spoon, plated dish, towel over shoulder.
Pose: big head, small body; proud expression like presenting a signature dish.
Style: bold ink outline, gentle shading, slightly exaggerated nose/cheeks (subtle).
Constraints: no brand labels, no readable menu text.
Use my uploaded photo as reference.
Create a classic caricature of me as a FIREFIGHTER.
Scene: station bay with a fire truck silhouette blurred in the back.
Props: helmet, gloves, radio, coiled hose in background.
Pose: big head, small body; confident stance, kind expression.
Style: strong outline, comic-like shading, vibrant but controlled colors.
Constraints: no department identifiers, no readable patches, no gore.
Style 3 — 3D de type Pixar / « aspect jouet 3D »
Visez : matériaux lisses, éclairage de studio doux, profondeur de champ subtile, ambiance de collectionneur.
Use my uploaded photo as reference.
Create a Pixar-ish 3D toy figure of me as a NURSE (family-friendly 3D animation look).
Scene: clean minimal hospital backdrop, soft gradients.
Props: miniature stethoscope, tiny clipboard, simple badge (no text).
Wardrobe: scrubs with realistic fabric folds.
Lighting: soft studio key light + rim light, subtle depth of field.
Pose: full-body, slight wave, warm smile.
Constraints: no logos, no readable text, no extra people, hands correct.
Use my uploaded photo as reference.
Create a Pixar-ish 3D toy figure of me as a TEACHER.
Scene: simplified classroom set like a miniature diorama.
Props: tiny whiteboard with shapes only, small book stack, marker.
Wardrobe: smart-casual, soft materials.
Lighting: warm, soft shadows, gentle highlights.
Pose: full-body, pointing at the board, friendly expression.
Constraints: no readable text anywhere, no brand names.
Use my uploaded photo as reference.
Create a Pixar-ish 3D toy figure of me as a SOFTWARE DEVELOPER.
Scene: miniature desk diorama with two monitors showing abstract UI blocks.
Props: tiny keyboard, mouse, coffee mug, headphones.
Wardrobe: hoodie, clean stylized fabric.
Lighting: soft studio lighting, subtle reflections on screens (no readable code).
Pose: seated on a small chair, focused but smiling.
Constraints: no logos, no readable screens, no extra people.
Use my uploaded photo as reference.
Create a Pixar-ish 3D toy figure of me as a CHEF.
Scene: miniature kitchen diorama with warm bokeh lights.
Props: tiny pan, spoon, plated dish, apron.
Wardrobe: chef jacket with soft fabric folds (no branding).
Lighting: warm key light, soft rim light, glossy highlights on cookware.
Pose: full-body, presenting the dish with pride.
Constraints: no readable text, no labels, hands normal.
Use my uploaded photo as reference.
Create a Pixar-ish 3D toy figure of me as a FIREFIGHTER.
Scene: miniature fire station bay with simple shapes in the background.
Props: helmet, gloves, radio, small hose coil.
Wardrobe: turnout gear with simplified reflective strips (no department text).
Lighting: dramatic but clean studio lighting, soft shadows.
Pose: full-body, heroic stance, calm friendly face.
Constraints: no logos, no readable patches, no flames touching the character.
Style 4 — Croquis (crayon/encre)
Visez : lignes dessinées à la main, détails sélectifs (visage + accessoires clés), texture de papier.
Use my uploaded photo as reference.
Create a pencil-and-ink sketch portrait of me as a NURSE.
Scene: minimal nurse station outline.
Props: stethoscope + clipboard (no readable notes).
Style: graphite shading, ink accents, paper texture, cross-hatching.
Framing: head-and-shoulders with a hint of uniform neckline.
Constraints: no extra people, no readable text, natural hands if visible.
Use my uploaded photo as reference.
Create a pencil-and-ink sketch of me as a TEACHER.
Scene: classroom suggested with a few lines (bookshelf, board edge).
Props: marker in hand, notebook under arm.
Style: loose sketch lines, cross-hatching on shadows, subtle paper grain.
Framing: half-body, slight three-quarter view.
Constraints: no readable board text, no students.
Use my uploaded photo as reference.
Create a pencil-and-ink sketch of me as a SOFTWARE DEVELOPER.
Scene: desk outline with two monitors (blank/abstract).
Props: keyboard, mug, headphones.
Style: clean portrait lines, heavier ink on facial features, light graphite shading.
Framing: half-body, seated.
Constraints: no readable screens, no logos, hands realistic.
Use my uploaded photo as reference.
Create a pencil-and-ink sketch of me as a CHEF.
Scene: kitchen suggested with minimal line work.
Props: spoon + plated dish; apron straps visible.
Style: expressive line work, cross-hatching, paper texture.
Framing: half-body, proud expression.
Constraints: no labels, no readable text, no extra people.
Use my uploaded photo as reference.
Create a pencil-and-ink sketch of me as a FIREFIGHTER.
Scene: station bay suggested with a few structural lines.
Props: helmet and radio.
Style: strong ink outlines on gear, graphite shading on face, paper grain.
Framing: head-and-shoulders, confident calm expression.
Constraints: no department identifiers, no readable patches, no gore.
Style 5 — Style d'affiche minimaliste
Visez : formes audacieuses, palette limitée, beaucoup d'espace négatif, 1 à 2 accessoires emblématiques.
Use my uploaded photo as reference.
Create a minimalist poster portrait of me as a NURSE.
Style: flat geometric shapes, limited palette (2–4 colors), lots of negative space.
Props: one iconic stethoscope silhouette + small clipboard shape.
Framing: head-and-shoulders, centered, calm confident expression.
Background: clean blocks, no text.
Constraints: no logos, no readable badge text, no extra people.
Use my uploaded photo as reference.
Create a minimalist poster portrait of me as a TEACHER.
Style: flat shapes, limited palette, modern poster composition.
Props: chalk/marker silhouette + simple book stack shape.
Framing: head-and-shoulders, slight three-quarter angle.
Background: abstract classroom shapes, no readable text.
Constraints: no students, no logos.
Use my uploaded photo as reference.
Create a minimalist poster portrait of me as a SOFTWARE DEVELOPER.
Style: flat vector shapes, limited palette, strong contrast.
Props: monitor silhouette + keyboard shape + tiny mug icon.
Framing: head-and-shoulders, clean edges, subtle facial simplification.
Background: abstract UI blocks (no text).
Constraints: no readable screens, no logos, no extra people.
Use my uploaded photo as reference.
Create a minimalist poster portrait of me as a CHEF.
Style: flat poster design with 2–4 colors, clean negative space.
Props: chef hat silhouette OR spoon silhouette (pick one) + plate circle.
Framing: head-and-shoulders, proud expression.
Background: abstract kitchen shapes, no labels.
Constraints: no brand names, no readable text.
Use my uploaded photo as reference.
Create a minimalist poster portrait of me as a FIREFIGHTER.
Style: bold flat shapes, limited palette, strong silhouette.
Props: helmet silhouette + radio icon.
Framing: head-and-shoulders, centered, steady confident look.
Background: abstract station shapes, no text.
Constraints: no department identifiers, no logos, no gore.
Même pack, regroupé par profession
Si vous pensez que « l'emploi d'abord, le style ensuite », utilisez cette correspondance :
| Profession | Dessin animé 2D | Caricature | 3D de type Pixar | Croquis | Affiche minimaliste |
|---|---|---|---|---|---|
| Infirmière | 01 | 06 | 11 | 16 | 21 |
| Enseignant | 02 | 07 | 12 | 17 | 22 |
| Développeur de logiciels | 03 | 08 | 13 | 18 | 23 |
| Chef | 04 | 09 | 14 | 19 | 24 |
| Pompier | 05 | 10 | 15 | 20 | 25 |
Corrections qui battent « réessayez »
Si le visage cesse de vous ressembler
Réduisez la complexité : utilisez moins d'accessoires, simplifiez l'arrière-plan et resserrez le cadrage (la tête et les épaules fonctionnent mieux). Lorsqu'un outil prend en charge les images de référence, appelez-la explicitement votre « ancrage d'identité ». Pour Midjourney, voir Prompts d'image; ; pour ChatGPT, voir Créer des images dans ChatGPT.
Si l'emploi ne correspond pas
Remplacez les accessoires avant de remplacer le style. Votre décor + deux outils emblématiques communiquent généralement l'emploi plus rapidement que « des vibrations ». Si vous utilisez Midjourney, leur Bases des prompts sont une référence claire pour construire « sujet + décor + détails ».
Si l'image ajoute du texte, des logos ou des écrans lisibles
Définissez « aucun texte lisible / aucun logo / aucune donnée privée » comme contrainte stricte, et répétez-la une fois dans le prompt. Si vous itérez via l'API, le Guide de génération d'images montre comment les flux et les modifications en plusieurs étapes sont généralement gérés.
Références
Documents officiels et démos utilisés dans cet article : Créer des images dans ChatGPT, Fonctionnalités d'image ChatGPT, Guide de génération d'images, Guide de prompts 'gpt-image-1.5', Bases des prompts, Prompts d'image Midjourney, , ainsi que les trois visites guidées YouTube intégrées ci-dessus : tUIrW3ROpzo, Ez-QzhW5x0k, leN4Vhs90RE.