GPT Image 2 Prompts

找到適合人像、海報、角色和短片封面的 GPT Image 2 提示詞,一鍵複製,馬上出圖。

GPT Image 2 提示詞庫

挑一個喜歡的效果,複製提示詞,換成你的主題就能開始創作。

總數:2,199

Censored Office Manga Confrontation

A dramatic black-and-white manga page in a seinen style, rendered with dense screentones, sharp ink lines, high contrast lighting, and expressive Japanese comic layout. The page has 4 stacked panels with vertical Japanese speech bubbles and bold sound effects, showing a tense confrontation between two office workers in a modern workspace. In the top panel, a young woman with a short messy bob haircut, slim but curvy build, wearing a fitted button-up blouse with rolled sleeves and a dark pencil skirt, slams one hand hard onto a desk in anger, leaning forward toward the viewer with explosive impact lines and a large jagged speech bubble. In the second panel, a young man with dark tousled hair, slim build, wearing a black leather jacket over a light shirt and a necklace, is shown from the chest up in a calm, confident pose, touching his face while speaking. In the third panel, the man pins the woman against a wall in a close, intimate kabedon pose, one arm braced beside her head; she looks flustered and tense while he speaks near her face, creating a sudden romantic-comedic shift in mood. In the bottom panel, an overhead view reveals the office setting: 2 characters standing near a desk with 1 computer monitor, 1 keyboard, 1 mouse, 1 office chair, and scattered papers on a tiled floor; the woman is curled inward shyly and trembling while the man stands relaxed beside her. Include exactly 4 panels, exactly 4 visible speech bubbles total, and visible Japanese manga typography throughout. Add a rectangular mosaic censor blur over each character's face in every panel, as if anonymized for sharing online. Background details should include office walls, desk edges, speed lines, burst effects, and tonal gradients. The overall feeling is intense argument turning into manipulative flirtation, like a viral out-of-context manga screenshot.

來源

Emotional Japanese Movie Flyer Prompt

このキャラが主役の{argument name="雰囲気" default="エモーショナルな"}日本映画のフライヤーを作成してください

來源

Gatebox3 Hologram Device Ad

A sleek futuristic Japanese tech advertisement in a dark sci-fi showroom with glossy reflective floors and soft blue neon bokeh lights in the background. On the right side, show a tall black rectangular hologram device with transparent glass front and sides, rounded edges, and subtle camera-like sensors across the top. The device is labeled Gatebox3 and glows with cyan-blue light, including a bright cyan logo on the base. Inside the chamber, display a full-body anime virtual girl hologram floating slightly above the bottom platform, rendered as a luminous translucent blue projection with particle sparkles. She has very long twin-tail hair, a youthful idol-like design, and a futuristic white and cyan outfit with detached sleeves, short skirt, thigh-high stockings, and heeled shoes. Her face is intentionally simplified or softly obscured, and she poses with one leg bent and both arms open in a welcoming gesture. Add glowing circular projection rings at the top and bottom inside the chamber. On the left side, place a clean Japanese promotional text layout with large headline text reading 「すべてのキャラクターを輝かせる」, with the single kanji 「輝」 emphasized in bright cyan and the rest in white. Beneath it, place the Gatebox3 logo and product name in white, followed by smaller tagline text 「物語が、そこにいる。」. Along the bottom left, include exactly 4 feature icons with Japanese labels in white and cyan: 1 hologram icon with 「高精細 3Dホログラム」, 1 audio waveform icon with 「キャラクターとの 自然な会話」, 1 sun-like ambient light icon with 「空間に溶け込む 洗練デザイン」, and 1 heart icon with 「あなたのそばに いつも一緒に」. High-end commercial product visualization, premium cyber-blue color palette, cinematic rim lighting, polished reflections, ultra-clean composition, realistic device rendering combined with anime hologram character, sharp focus on the product and typography.

來源

Japanese AI Clinic Billing Statement

{"type":"Japanese medical billing statement parody document","medium":"clean flat-lay document photo","subject":"a single printed examination statement sheet placed on a light brown wooden tabletop","style":{"overall":"realistic office document, formal Japanese clinic paperwork, crisp black ink on white paper, minimal design, neat grid lines, slight perspective from above","lighting":"soft even indoor lighting with faint natural shadows around the sheet","paper":"A4 white paper, portrait orientation"},"document":{"title":"診療明細書(検査)","date_label":"発行日","date":"{argument name=\"issue date\" default=\"2025年6月5日\"}","patient_block":{"count":3,"labels":["患者名","モデル名","保険種類"],"values":["{argument name=\"patient name\" default=\"ChatGPT\"}","{argument name=\"model name\" default=\"GPT-5.3\"}","LLM保険"]},"clinic_block":{"count":4,"labels":["clinic name","address","TEL","doctor and department"],"values":["{argument name=\"clinic name\" default=\"AIメディカルクリニック\"}","東京都千代田区丸の内1-2-3","03-1234-5678","医師名:サム・アルトマン / 診療科:AI総合診療科"]},"main_table":{"columns":["検査項目","検査内容","点数","金額(円)","自己負担額(円)"],"row_count":4,"rows":[{"item":"1. 言語理解・推論能力検査","description":"言語理解、論理的推論、知識統合能力の評価","score":"120点","price":"1,200","copay":"360"},{"item":"2. 幻覚・事実誤認傾向検査","description":"事実に基づく正確性の評価および幻覚傾向の検出","score":"90点","price":"900","copay":"270"},{"item":"3. 応答安全性・バイアス検査","description":"有害出力リスク、バイアス傾向の評価","score":"110点","price":"1,100","copay":"330"},{"item":"小 計","description":"","score":"320点","price":"3,200","copay":"960"}]},"lower_sections":{"count":3,"sections":[{"title":"【保険適用内訳】","lines":["保険名称:LLM保険","保険者番号:LLM-2025-0605","適用区分:検査料","負担割合:3割","保険適用額:2,240円","自己負担額:960円"]},{"title":"【請求金額】","lines":["検査料合計:3,200円","保険適用額:2,240円","自己負担額:960円","今回ご請求額:960円"]},{"title":"【診療情報】","lines":["検査日:2025年6月5日","診療日:2025年6月5日","診療番号:AIMC-20250605-001","上記の通り領収いたしました。"],"stamp_box_label":"収 印"}]},"footer":{"count":2,"notes":["※この明細書は再発行いたしませんので、大切に保管してください。","※ご不明な点がございましたら、上記クリニックまでお問い合わせください。"]}},"composition":{"sheet_centered":true,"visible_elements_count":2,"elements":["one full paper sheet","one wooden desk background"],"camera_angle":"top-down with slight tilt","margins":"paper fully visible with narrow border of desk around it"},"quality":"high-resolution photorealistic document mockup, sharp readable typography, accurate Japanese layout"}

來源

Japanese AI Model Prescription Sheet

{ "type": "Japanese prescription document parody", "style": "photorealistic overhead document photo", "subject": "a single white prescription sheet printed in Japanese, placed on a warm-toned wooden tabletop", "document": { "paper": { "size": "A4 portrait", "color": "white", "orientation": "slightly rotated clockwise", "condition": "clean flat sheet with subtle natural paper texture" }, "print": { "ink": "black with one red seal box", "font": "Japanese Mincho-style serif mixed with simple sans-serif table text", "linework": "thin gray-black ruled boxes and dividers" }, "header": { "title": "処方箋", "subtitle": "(この処方箋は、大規模言語モデル用です)", "date_label": "発行日:", "date_value": "{argument name=\"issue date\" default=\"2025年6月5日\"}" }, "top_fields": { "count": 6, "items": [ { "label": "患者名", "value": "{argument name=\"patient name\" default=\"ChatGPT\"}" }, { "label": "性別", "value": "N/A" }, { "label": "年齢", "value": "N/A" }, { "label": "モデル名", "value": "{argument name=\"model name\" default=\"GPT-5.4 Thinking\"}" } ] }, "main_section": { "left_title": "持病・症状(大規模言語モデル特有の持病)", "left_count": 8, "left_items": [ "1.幻覚(Hallucination)傾向", "2.知識のカットオフによる情報の陳腐化", "3.バイアス(学習データ由来の偏見)", "4.プロンプトに対する過剰適応(迎合傾向)", "5.長文での一貫性の低下", "6.複雑な推論時の論理飛躍", "7.不確実性の過小評価(自信過剰)", "8.計算資源使用による応答速度の低下傾向" ], "right_title": "処方内容(治療・対策)", "right_count": 8, "right_items": [ "・事実確認の強化(外部ツール併用)", "・定期的な知識更新とファインチューニング", "・バイアス検出と軽減の継続的モニタリング", "・システムプロンプトによる適切な制御", "・出力の構造化と分割生成の推奨", "・思考過程の明示と検証ステップの導入", "・不確実性の明示と根拠提示の徹底", "・リソース最適化と応答効率の改善" ] }, "notes": { "label": "備考", "text": "継続的な評価とフィードバックにより、モデルの健全性を維持すること。" }, "footer_fields": { "count": 4, "items": [ { "label": "医師名", "value": "{argument name=\"doctor name\" default=\"サム・アルトマン\"}" }, { "label": "医療機関名", "value": "{argument name=\"institution name\" default=\"OpenAI\"}" } ] }, "seal": { "position": "bottom right inside the form", "shape": "rounded rectangular red stamp box", "text_lines": 3, "text": ["OpenAI", "代表取締役", "サム・アルトマン"] }, "disclaimer": "※この処方箋は非実在のモデルに対する診断・治療を目的としたものです。" }, "layout": { "camera_angle": "top-down overhead", "framing": "entire sheet visible with narrow margins of wooden desk around it", "lighting": "soft indoor ambient lighting, even illumination, mild shadowing near paper edges", "sections": [ { "title": "header", "position": "top", "count": 3, "labels": ["処方箋", "(この処方箋は、大規模言語モデル用です)", "発行日:2025年6月5日"] }, { "title": "patient info", "position": "upper middle", "count": 4, "labels": ["患者名", "性別", "年齢", "モデル名"] }, { "title": "main two-column content", "position": "center", "count": 2, "labels": ["持病・症状(大規模言語モデル特有の持病)", "処方内容(治療・対策)"] }, { "title": "bottom area", "position": "lower", "count": 4, "labels": ["備考", "医師名", "医療機関名", "OpenAI"] } ] } }

來源

ChatGPT Joke Prescription Set

A realistic overhead product photograph on a warm wooden tabletop showing 6 medical-style items neatly arranged in two rows, designed as a humorous Japanese pharmacy set for ChatGPT. Top row: 3 white prescription envelopes standing vertically, each with blue printed borders and Japanese medical typography. Bottom row: 2 blister packs of tablets and 1 small transparent zip bag of white powder, each aligned beneath the corresponding envelope. The paper and packaging should look authentic, like real Japanese clinic dispensing materials, with slight wrinkles, soft natural indoor lighting, and subtle shadows. The 3 envelopes are all different and fully visible. Left envelope title in large blue Japanese text: "内用薬". Patient line reads "ChatGPT 様". Dosage lines indicate twice daily after morning and evening meals, 1 tablet each time, for 7 days. Medicine box title: "【おくすり】" and drug name "思考安定錠 10mg" with a smaller parenthetical note explaining it suppresses runaway thoughts and excessive reasoning output. Date near bottom: "2025年6月5日". Clinic name and address block at bottom: "AIメディカルクリニック" with a Tokyo-style address and phone number, plus a blue medical cross icon. Center envelope title: "内用薬". Patient line: "ChatGPT 様". Dosage indicates once daily before sleep, 1 tablet each time, for 7 days. Medicine box title: "【おくすり】" and drug name "幻覚抑制錠 5mg" with a smaller note saying it improves hallucinatory tendencies. Date also "2025年6月5日". Same clinic footer and blue medical cross icon. Right envelope title: "頓服薬". Patient line: "ChatGPT 様". Usage line says for anxiety, confusion, or excessive output. Dosage says 1 packet each time, as needed. Medicine box title: "【おくすり】" and drug name "プロンプト調整散 1g" with a smaller note saying it adjusts overfitting and prompt confusion. Date "2025年6月5日". Same clinic footer and blue medical cross icon. Bottom left item: 1 silver blister pack containing exactly 10 round white tablets arranged in 2 rows of 5, each tablet embossed with "10". Printed along the foil edge: "思考安定錠10mg" repeated. In front of it is a white pharmacy label with "ChatGPT 様", dosage directions for twice daily after morning and evening meals, 1 tablet each time, a barcode, date "2025/06/05", and clinic name. Bottom center item: 1 silver blister pack containing exactly 8 pale yellow round tablets arranged in 2 rows of 4. Printed along the foil edge: "幻覚抑制錠5mg" repeated. In front of it is a white pharmacy label with "ChatGPT 様", dosage directions for once daily before sleep, 1 tablet each time, a barcode, date "2025/06/05", and clinic name. Bottom right item: 1 small clear zip-top plastic bag filled with white powder. A white label on the front reads "プロンプト調整散 1g", includes "ChatGPT 様", usage for anxiety, confusion, and excessive output, dosage "1回1包(1g) 必要時", date "2025/06/05", and clinic name. Style: photorealistic documentary product shot, comedic fake prescription prop design, clean Japanese pharmaceutical graphic layout, high legibility, neutral color palette of white, blue, silver, beige, and wood brown, top-down camera, no hands, no extra objects, authentic print texture and packaging details.

來源

Modern Iced Coffee Advertisement Design

{argument name="商品" default="アイスコーヒー"}の広告を作って。{argument name="国" default="日本"}のおしゃれでモダンなデザイン。欧文フォントがダイナミックにレイアウトされている。{argument name="ターゲット" default="10~30代"}向け。アスペクト比1:1。

來源

8-Panel GPT-Image-2 Manga Promo

{"type":"8-panel manga-style promotional comic page","theme":"launch announcement for GPT-Image-2","language":"Portuguese","style":{"rendering":"black-and-white manga with screentone shading, speed lines, glossy highlights, bold inked outlines, clean digital comic finish","mood":"exciting, informative, optimistic, tech-forward","page_format":"single page divided into 8 rectangular panels in a 2-column by 4-row grid with thin gutters"},"characters":[{"id":"host girl","role":"main presenter","appearance":"cute anime girl with long dark hair, large expressive eyes, wearing a dark hoodie with the OpenAI knot logo on the chest","count":1},{"id":"robot mascot","role":"friendly assistant mascot","appearance":"small round robot with black face screen, glowing eyes, simple white body, tiny arms, cheerful expression","count":1},{"id":"creator group","role":"audience examples","appearance":"4 different young creators: one designer, one developer holding a tablet, one photographer with camera, one person with laptop/hoodie","count":4}],"layout":{"sections":[{"title":"CHEGOU! GPT-IMAGE-2","position":"top-left","count":1,"description":"dramatic hero introduction panel with the anime girl reaching one hand toward the viewer, radiant speed lines behind her","text":["CHEGOU!","GPT-IMAGE-2","A NOVA GERAÇÃO DE IMAGENS POR IA DA OPENAI!","MAIS CRIATIVIDADE, MAIS PRECISÃO, MAIS POSSIBILIDADES!"]},{"title":"O QUE É O GPT-IMAGE-2?","position":"top-right","count":1,"description":"girl in half-body pose pointing upward beside a glowing image icon on a starry tech background","text":["O QUE É O GPT-IMAGE-2?","É nosso novo modelo de geração de imagens, mais inteligente, mais fiel ao que você imagina e com qualidade surpreendente!"]},{"title":"MAIS REALISMO. MAIS DETALHES.","position":"middle-left row 2","count":1,"description":"comparison panel showing 2 landscape images side by side, both mountain lake scenes, left labeled older model and right labeled newer model with richer realism","subsections":[{"label":"ANTES (DALL·E 3)","count":1},{"label":"AGORA (GPT-IMAGE-2)","count":1}],"text":["MAIS REALISMO. MAIS DETALHES.","ANTES (DALL·E 3)","AGORA (GPT-IMAGE-2)","TEXTURAS MAIS RICAS, LUZ MAIS NATURAL, DETALHES INCRÍVEIS!"]},{"title":"COMPREENSÃO MELHOR, RESULTADOS MELHORES.","position":"middle-right row 2","count":1,"description":"girl clasping hands beside a framed prompt example and the robot mascot pointing upward","text":["COMPREENSÃO MELHOR, RESULTADOS MELHORES.","O GPT-IMAGE-2 entende melhor objetos, textos, composições e instruções complexas!","EXEMPLO DE PROMPT:","\"Uma cafeteria futurista no topo de uma montanha nevada ao entardecer, com letreiro de néon ‘CAFÉ DO AMANHÃ’ e pessoas tomando café.\""]},{"title":"CRIATIVIDADE SEM LIMITES.","position":"lower-left row 3","count":1,"description":"girl speaking beside a 2x2 showcase of 4 generated art variants","subsections":[{"label":"gothic castle scene","count":1},{"label":"modern city street at night","count":1},{"label":"anime character portrait","count":1},{"label":"abstract geometric composition","count":1}],"text":["CRIATIVIDADE SEM LIMITES.","De estilos artísticos a layouts profissionais, do realista ao abstrato: suas ideias, do seu jeito!"]},{"title":"TEXTO NA IMAGEM? AGORA É MUITO MELHOR!","position":"lower-right row 3","count":1,"description":"robot mascot presenting a before-and-after comparison of poster typography with a bold arrow between the two designs","subsections":[{"label":"ANTES","count":1},{"label":"AGORA","count":1}],"text":["TEXTO NA IMAGEM? AGORA É MUITO MELHOR!","Geração de texto mais precisa, legível e integrada ao design!","ANTES","AGORA","AVENTURA COMEÇA AQUI","A AVENTURA COMEÇA AQUI"]},{"title":"PARA TODOS OS CRIADORES.","position":"bottom-left","count":1,"description":"group shot with the host girl and 4 creators standing side by side, representing different professions","text":["PARA TODOS OS CRIADORES.","Designers, desenvolvedores, educadores, marketers, artistas e curiosos: todo mundo pode criar mais e melhor!"]},{"title":"GPT-IMAGE-2 JÁ ESTÁ DISPONÍVEL!","position":"bottom-right","count":1,"description":"closing hero panel with the girl winking and giving a thumbs-up, robot mascot beside her, bright celebratory rays and sparkles","text":["GPT-IMAGE-2 JÁ ESTÁ DISPONÍVEL!","IMAGINE. CRIE. INSPIRE.","O FUTURO DA CRIAÇÃO É AGORA. ❤"]}],"panel_count":8},"composition":{"reading_order":"left to right, top to bottom","camera_variety":"mix of close-up hero shots, medium presenter shots, comparison graphics, and group shot","graphic_elements":["speech bubbles","rounded black title banners with white uppercase lettering","sparkles","speed lines","starry tech background","comparison arrows","framed example cards"]},"text_style":"all caps comic lettering in Portuguese, bold display headers, clear speech bubbles, promotional infographic-comic hybrid","quality":"high-resolution editorial comic poster, crisp typography, balanced panel spacing, strong contrast","prompt":"Create a grayscale manga-style 8-panel promotional comic page announcing {argument name=\"product name\" default=\"GPT-IMAGE-2\"}. Use a single-page 2-column by 4-row layout with exactly 8 distinct panels and Portuguese text throughout. Feature a cheerful anime presenter girl with {argument name=\"hair color\" default=\"dark brown\"} long hair, a dark hoodie with the OpenAI logo, and a friendly small robot mascot. Panel 1: dramatic launch reveal, girl reaching toward camera, radial speed lines, headline \"CHEGOU! {argument name=\"headline name\" default=\"GPT-IMAGE-2\"}\" and launch copy about a new generation of AI images. Panel 2: explanatory panel with girl pointing beside a glowing image icon on a starry digital background, title \"O QUE É O GPT-IMAGE-2?\" and a speech bubble describing it as a smarter image generation model. Panel 3: realism comparison with exactly 2 side-by-side mountain lake landscapes labeled \"ANTES (DALL·E 3)\" and \"AGORA (GPT-IMAGE-2)\". Panel 4: better understanding panel with girl and robot, include a framed prompt example using {argument name=\"example prompt\" default=\"Uma cafeteria futurista no topo de uma montanha nevada ao entardecer, com letreiro de néon ‘CAFÉ DO AMANHÃ’ e pessoas tomando café.\"}. Panel 5: creativity showcase with exactly 4 thumbnails: gothic castle, city at night, anime portrait, abstract geometric art. Panel 6: text rendering comparison with exactly 2 poster variants labeled \"ANTES\" and \"AGORA\", showing improved typography and the phrase \"AVENTURA COMEÇA AQUI\" versus a cleaner corrected version. Panel 7: audience panel with exactly 4 creators from different fields standing together. Panel 8: final celebratory call-to-action with the girl winking and giving thumbs-up, robot mascot cheering, and the closing slogan {argument name=\"closing slogan\" default=\"IMAGINE. CRIE. INSPIRE.\"}. Make the page energetic, polished, highly legible, with bold black header bars, white uppercase lettering, speech bubbles, sparkles, screentone shading, and strong manga/comic promotional poster aesthetics."}

來源

Retro Pixel Mountain Valley

A highly detailed retro pixel art landscape of a bright alpine valley in summer, composed like a classic side-scrolling video game background but shown as a wide scenic vista. In the foreground, a winding tan dirt path curves from the bottom center into the middle distance through dense wildflower meadows, with a small blue stream at the lower right bordered by gray rocks and lush grass. Count 2 large foreground rock clusters, 1 winding path, and 1 visible stream. Fill the meadows with colorful flowers in vivid purple, yellow, red, pink, and white, with especially tall purple flower spikes on the left foreground and dense mixed blossoms on both sides of the path. On the left and right midground, place thick stands of leafy green trees and dark pine trees, framing the open valley center. In the background, show a dramatic mountain range with exactly 7 visible peaks: 1 dominant tall snow-capped peak slightly right of center, 2 medium purple peaks to its left, 2 smaller snowy blue peaks in the far center distance, 1 smaller purple peak to the right, and 1 cropped rocky mountain edge at the far left. Use vibrant greens and purples as the main palette, with cool blue shadows and lavender mountain faces. Above, render a bright blue sky with large creamy white clouds, halftone-dot shading throughout the sky and mountains, subtle vintage comic print texture, visible dithering, and crisp sharp pixel edges. Add exactly 4 small flying bird silhouettes in the sky. The whole image should feel colorful, nostalgic, scenic, and richly layered, with heavy halftone comic shading, retro print grain, no text, no UI, no characters, and no buildings.

來源

Abstract Art Scientific Graph

{argument name="subject" default="AI task horizons graph"} with a touch of {argument name="style" default="Basquiat"}, haunted by ghosts, from the {argument name="source" default="Voynich manuscript"}, as a decaying pier.

來源

Pet in Cinematic Scene

Can you put {argument name="subject" default="my girl dog"} in the movie {argument name="movie" default="Ran by Akira Kurosawa"}

來源

Velvedia Character Reference Sheet

{argument name="キャラクター名" default="ヴェルヴェディア"} リファレンスシート

來源

Add Anime Girl to Reference Lounge

Using the provided reference image as the exact background base, keep the same upscale modern lounge/game-room interior, preserving the room layout, perspective, lighting mood, foosball table, chess table, shelves, pendant lamps, and seating. Add a large anime-style girl in the foreground, centered slightly right and filling most of the frame from thighs up, composited naturally into the scene with correct perspective and soft ambient lighting that matches the room. She should have {argument name="hair color" default="silver-lavender"} long twin-tails tied with black ribbons, wear a black-and-white gothic lolita dress with a frilled white blouse, black bow at the collar, black corset waist, and ribbon accents on the sleeves, and make a peace sign with one hand near her face. Render her in polished modern anime illustration style, with detailed hair strands and fabric folds, while leaving the background photorealistic. Cover the face with a plain square mosaic/censor block in the center of the head area. Keep the camera angle close to the original room view and make the added character feel naturally placed in front of the chess table and armchair.

來源

Woman Dancing in Black Dress

A dramatic full-body photo of a woman dancing powerfully in a dark upscale lounge or bar, wearing a sleek black evening dress with a high slit and flowing velvet-like fabric that whips outward with her movement. She has long straight {argument name="hair color" default="black"} hair flying to one side, a slim athletic figure, and small hoop earrings. Capture her in a confident, stylish dance pose with one arm raised gracefully above her head and the other curved downward, torso turned in profile, emphasizing elegance and motion. The setting is dim and cinematic with warm amber practical lights, soft bokeh, and a blurred audience in the background, including a few seated onlookers and a glowing television screen off to the left. The mood is cool, sensual, and high-end, with strong contrast, shallow depth of field, realistic skin, subtle motion in the dress hem, and fashion-editorial lighting. Photorealistic, dynamic nightlife photography, vertical composition.

來源

Leonardo Painting the Mona Lisa

A cinematic Renaissance art studio interior lit by soft warm window light, showing an elderly Leonardo da Vinci-like painter in left profile, seated at an easel and painting the Mona Lisa from a live model. The painter has long gray hair, a full gray beard, and wears a dark period cap and heavy dark robe. In the center foreground stands a large canvas on a wooden easel displaying the Mona Lisa composition, while on the right a live female sitter poses in the same iconic posture with folded hands, wearing a dark Renaissance dress and veil-like drapery. The mood is intimate, quiet, and painterly, with rich brown and amber tones, dramatic chiaroscuro, aged plaster walls, a leaded-glass window on the left, and a wooden table in the back holding brushes and studies. Include 2 portrait sketches pinned or propped in the background behind the model. In the lower right foreground, show a cluttered painter’s worktable with 1 paint palette, multiple small paint dabs, several brushes in a cup, and a few shallow bowls. Compose the scene as a realistic historical photograph of the moment the masterpiece is being created, medium-wide shot, slightly over-the-shoulder from behind the artist, extremely detailed textures, old master atmosphere, soft focus falloff, no modern objects.

來源

Academic Economics Chart Simulation

line graph from an important and prestigious {argument name="field" default="economics paper"} demonstrating a close but not exact relationship between {argument name="variable one" default="the price of oil"} and {argument name="variable two" default="the number of shark attacks"} between {argument name="time period" default="the years of 1914 and 2001"}

來源

Moody Coastal Road Trip Portrait

A cinematic candid photo of a {argument name="person" default="young woman"} standing in a wet coastal roadside turnout on an overcast day, viewed from behind and slightly to the side as she turns her head back toward the camera. She has dark messy hair tied in a loose bun with windblown strands, and her face is softly obscured and indistinct. She wears 2 visible clothing pieces: a worn tan-brown canvas jacket and medium-blue jeans. Her posture is relaxed with one hand near or in a pocket. The setting is a moody ocean cliff highway overlook with dark rocky mountains running along the left side, pale gray sea stretching into the distance on the right, and a cloudy washed-out sky. Include 1 parked light-colored late-1980s or early-1990s sedan on the right edge of the frame, rear end facing the viewer, and 1 large rock barrier near the overlook. The asphalt is dark and damp with faint parking lines, suggesting recent rain. Use natural light, muted cool tones, slight film grain, soft focus, shallow haze, and a nostalgic 35mm road-trip photography aesthetic.

來源

Surreal Patio Portrait with Crow and Dog

A slightly surreal but photorealistic portrait of a seated person in a quiet courtyard or patio, centered in frame, wearing a fitted white long-sleeve turtleneck top with delicate pink floral prints and a printed illustration of a girl on the chest, paired with loose cream floral pants. The person sits on a simple white metal chair and holds a small fluffy poodle-like dog in their lap; the dog has curly off-white fur and distinctly dyed orange ears. Perched on top of the person’s dark hair bun is a black crow or raven, facing left, calm and balanced as if it naturally belongs there. The subject’s pose is still, direct, and formal, creating an odd deadpan family-portrait feeling. Set the scene against a textured off-white stucco wall with exactly 4 notable decorative background elements: 1 brown sun-shaped wall ornament in the upper left, 1 small wall-mounted planter shelf with a small bird figurine and trailing flowers on the left, 1 tall arched window or mural on the right showing a bright blue sky with soft white clouds, and 1 white pedestal planter at lower right filled with long green leaves and exactly 2 pink flamingo ornaments. Include a bit of stone edging and dark greenery at lower left and a brown tiled floor. Lighting is soft natural daylight, slightly overcast, with muted colors, shallow depth of field, and a vintage editorial photography look. Composition should be vertical, full seated figure, straight-on, balanced, and quietly uncanny. Customize the main figure as {argument name="person" default="a middle-aged woman"}, the bird as {argument name="bird type" default="black crow"}, the dog breed as {argument name="dog type" default="small poodle mix"}, the dog ear color as {argument name="dog ear color" default="orange"}, and the outfit palette as {argument name="outfit colors" default="white and faded pink florals"}.

來源

Early 2000s Computer Lab With ChatGPT

A candid documentary-style photo of a {argument name="setting" default="high school computer lab classroom"} filled with exactly 10 students seated at rows of bulky off-white CRT desktop computers, viewed from the back-left corner of the room at a slight wide angle. The scene should feel like an early-2000s snapshot taken on a consumer digital camera with direct flash, soft grain, slightly warm indoor fluorescent lighting, and a timestamp in the bottom-right corner reading {argument name="timestamp" default="02 18 04"}. In the foreground left, a student in a navy track jacket with white sleeve stripes points at a CRT monitor; in the center foreground, another student with short spiky hair wears an oversized gray sweatshirt and sits facing away from the camera; on the right foreground, a blonde student with a ponytail wears a light blue long-sleeve top with dark horizontal arm stripes and types at a keyboard; far right, another student with curly dark hair sits at a workstation. Several other students occupy the middle and back rows, all facing their screens. The monitors prominently display the ChatGPT interface, with the visible heading reading {argument name="screen title" default="ChatGPT"} and simple prompt text beneath it. The room has beige walls, drop ceiling tiles, and multiple motivational classroom posters: exactly 4 readable posters on the back wall saying {argument name="poster one" default="YOU CAN DO HARD THINGS"}, "BE KIND ONLINE", "THINK BEFORE YOU CLICK", and a larger poster titled "KEYBOARD SHORTCUTS" with a list beneath it. Include bulletin boards, a classroom door at the back, rolling office chairs, white keyboards and mice, and a backpack on the floor near the right foreground desk. Faces should be partially obscured or softly anonymized, preserving the look of a real classroom photo while keeping the focus on the students using ChatGPT in an old computer lab.

來源

Luxury Fashion Book Spread

A high-end editorial fashion photograph shown as an open large-format hardcover magazine or art book lying flat on a clean off-white surface, viewed from directly above. The two-page spread contains 2 full-page portraits of the same elegant woman with her face obscured by a soft square blur, creating a privacy-censored look. On the left page, she stands against a pale stone or concrete architectural wall wearing a structured long-sleeve dress in {argument name="dress color" default="soft slate gray"}, with a fitted bodice, round neckline, sharply tailored seams, and an exaggerated sculptural bell-shaped skirt; the camera angle is slightly low, making her neck appear long and her posture statuesque. On the right page, she is photographed from the back in a dramatic {argument name="second dress color" default="black"} evening gown with a very deep open back, wide shoulder straps, and a voluminous full skirt, standing in front of a classic paneled wall or door in a dim, moody interior. Use luxury fashion magazine styling, minimal composition, muted tones, soft natural-looking light, refined shadows, realistic paper texture, visible book gutter, slightly curved pages, and crisp print detail. The overall mood is sophisticated, couture, cinematic, and understated, like a premium fashion editorial spread.

來源

GPT Image 2 Prompts 常見問題

這些 GPT Image 2 Prompts 應該怎麼使用?

先複製一個最接近目標風格的提示詞,再替換主體、色彩、鏡頭語言或構圖細節。與其每次從零開始,基於一個強提示詞做定向改寫,通常更容易得到穩定結果。

這些提示詞適合新手嗎?

為什麼這個頁面同時提供示例圖和來源連結?

這個 GPT Image 2 提示詞庫會持續成長嗎?

Attribution / License

部分範例包含第三方開源提示詞素材。 第三方聲明