已验证示例

提示词库

真实提示词与对应生成图片。浏览灵感、查看效果,并复制完整提示词作为创作起点。

图文组合
5,541
模型
6
来源
10
5,541 组已验证提示词与图片 · 第 61 / 116 页
Example generated from “Realistic Japanese Idol Close-Up”
GPT Image 2en
Realistic Japanese Idol Close-Up

A hyper-realistic close-up portrait of a young Japanese female idol, shot vertically from the shoulders up, with her head slightly tilted and looking directly at the camera. She has {argument name="hair color" default="jet black"} Her medium-length straight hair, with thick, straight bangs partially covering her forehead and gently framing her face, is striking. Her skin is incredibly smooth and fair, appearing evenly lit. Her large, deep brown eyes hold a soft glow, her nose is small and natural, and her slightly parted pink lips reveal slightly uneven front teeth, one of which is noticeably irregular and overlapping, creating a captivating yet imperfect smile. Her expression is calm, gentle, and innocent. She wears an exquisite light blue ruffled lace top with sheer pleats at the shoulders. The background is a soft, blurred palette of cool tones—lavender, blue, and pink—suggesting an indoor stage or event setting, with a bokeh effect reminiscent of light spots. The design emphasizes realistic skin texture, accurate asymmetrical teeth, an idol photography style, shallow depth of field, and a clean, refined Japanese portrait aesthetic.

Example generated from “Realistic Japanese Newspaper Front Page”
GPT Image 2en
Realistic Japanese Newspaper Front Page

{ "type": "Photos of print pages of Japanese newspapers", "style": "The paper has a realistic newspaper texture, a slightly tilted overhead angle, natural light, and slight wrinkles.", "header": { "date": "{argument name=\"date\" default=\"Thursday, April 23, 2026 (Reiwa 8)\"}", "newspaper_name": "{argument name=\"newspaper name\" default=\"Asahi Shimbun\"}", "publisher_info": "Issuing Office: Asahi Shimbun Tokyo Headquarters", "top_right_ad": "Water and Life SUNTORY" }, "main_headlines": [ "{argument name=\"main headline 1\" default=\"OpenAI releases new model "Spud"\"}", "{argument name=\"main headline 2\" default=\"The "Codex SuperApp" will be released simultaneously.\"}" ], "sub_headline_bar": "It comprehensively surpasses existing benchmarks in terms of network attack and coding performance.", "layout": { "sections": [ { "position": "center left", "type": "Article text bar", "content": "Japanese news text" }, { "position": "central", "type": "Embedded Images", "description": "Dark blue graphics showcasing three simulated mobile UIs.", "overlay_text": "{argument name=\"center image text\" default=\"Codex SuperApp\"}", "caption": "Codex SuperApp concept image (provided by OpenAI)" }, { "position": "center right", "type": "Vertical text headings", "text": "The competition to develop \"the cornerstone of the AGI era\" is intensifying." } ] }, "lower_section": { "headline": "Experts gave it high praise, calling it \"overwhelming progress\".", "columns": 3, "left_column": { "title": "■Main Features", "bullet_points_count": 3 }, "middle_column": "Japanese article text", "right_column": "Japanese article text" }, "footer_teasers": { "count": 3, "labels": [ "Tariff negotiations: Japan and the US to renegotiate", "Heavy rains hit Kyushu region, injuring two people.", "Opinion" ] } }

Example generated from “Realistic Korean School Classroom Portrait”
GPT Image 2en
Realistic Korean School Classroom Portrait

Ultra-realistic portrait of a beautiful Korean woman standing inside an empty Korean classroom during late afternoon golden hour, captured in a cinematic film photography style. She faces slightly to the side with her gaze turning toward the camera, expression soft and introspective — lips gently parted, eyes calm and luminous. Both hands loosely holding the strap of a small canvas tote bag hanging off one shoulder, effortlessly cool posture. Wearing an oversized vintage cream knit sweater slightly off one shoulder, paired with high-waisted dark navy pleated mini skirt and white low-top sneakers. Soft layered gold necklaces, minimal jewelry, understated Korean street fashion aesthetic. Long dark brown hair with subtle highlights, half-tucked behind one ear, loose natural waves. Dewy glass skin, flushed cheeks, barely-there tinted lip, soft eye makeup — natural Korean beauty. Warm amber golden light flooding in through tall classroom windows at a low angle, casting long dramatic shadows across wooden desks and the floor. Dust particles softly visible in the light beams. Authentic Korean classroom setting — rows of wooden desks, vintage chalk-dusted green board with Korean writing, South Korean flag, old wall clock, stacked textbooks, worn polished flooring. 35mm analog film grain texture, Kodak Portra 400 color palette — warm honey tones, muted greens, soft creams. Shallow depth of field, background bokeh on classroom furniture. Nostalgic coming-of-age Korean drama visual tone, quiet emotional storytelling atmosphere, editorial lifestyle photography, cinematic still frame, hyper-realistic skin and fabric texture, golden hour haze, peaceful solitude mood.

Example generated from “Realistic League of Legends Cosplay”
GPT Image 2en
Realistic League of Legends Cosplay

League of Legends Cosplayer, this character is{argument name="character" default="Miss Fortune the Bounty Hunter"} .{argument name="facial features" default="French face"} .Proportion{argument name="aspect ratio" default="9:16"}

Example generated from “Realistic Male Character Sheet”
GPT Image 2en
Realistic Male Character Sheet

{"type":"character reference sheet","style":"clean fashion lookbook mixed with a realistic model sheet","subject":{"name":"{argument name=\"character name\" default=\"KEIGO\"}","name_kana":"けいご","gender_presentation":"young adult Japanese man","appearance":{"hair_color":"dark brown to black","hair_style":"medium-length wavy tousled hair with soft volume and a slightly messy natural texture","build":"slim, average height","face":"realistic Japanese facial structure, soft youthful features"},"outfit":{"count":4,"featured_variant":"light spring casual outfit that looks gentle and neat","pieces":["beige lightweight jacket","light blue button-up shirt","off-white or cream straight pants","white low-top sneakers"]}},"layout":{"background":"very light gray studio background inside a rounded white presentation card","overall_format":"vertical character sheet with large rounded corners and minimalist editorial spacing","sections":[{"title":"head portrait","position":"top left","count":1,"labels":["front close-up"]},{"title":"full body","position":"right large panel","count":1,"labels":["standing pose with one hand in pocket"]},{"title":"side views","position":"middle left","count":2,"labels":["left profile","right profile"]},{"title":"lower portrait","position":"bottom left","count":1,"labels":["three-quarter close-up"]}],"total_images":5},"composition":{"top_center_text":"{argument name=\"headline text\" default=\"KEIGO\"}","subtitle_text":"けいご","pose_details":"full-body figure standing casually, slight lean, relaxed shoulders, one hand in pocket, clean catalog-style posture","camera_views":"front close-up, full body front, left profile, right profile, lower three-quarter portrait"},"rendering":{"photorealistic":true,"lighting":"soft even studio lighting with almost no harsh shadows","color_palette":"beige, pale blue, cream, white, soft gray","mood":"minimal, polished, springlike, calm, commercial fashion sheet"},"quality_notes":"high facial consistency across all 5 views, realistic clothing folds, clean magazine-like layout, no extra props, no dramatic styling, no heavy text beyond the name"}

Example generated from “Realistic Manga Receipt Generator”
GPT Image 2en
Realistic Manga Receipt Generator

Generate a purchase{argument name="item" default="comics"} Image of the receipt.{argument name="title" default="The title of a comic can be anything."}

Example generated from “Realistic McDonald's Selfie Portrait”
GPT Image 2en
Realistic McDonald's Selfie Portrait

A highly realistic smartphone selfie taken inside an Asian McDonald's restaurant features a half-body portrait of a young man seated at a table. The camera is held at arm's length, slightly above the table, creating a natural frontal selfie perspective, with one forearm extending towards the lens. The subject wears a black T-shirt with a small white Tesla logo on the left chest, has short, straight black hair, and thin-rimmed black glasses are faintly visible. One hand lightly supports his chin, his posture relaxed. His face is completely obscured by a large, solid rectangular blur, preserving his privacy while maintaining the visibility of his hair, ears, jawline, and the temples of his glasses. A white McDonald's paper beverage cup with a lid and a white straw is placed in the lower left foreground, cropped at the edge of the image. The background showcases the restaurant counter area with warm ambient lighting; to the left is a large red wall panel featuring the yellow McDonald's arches logo and the white slogan "I'm lovin' it" with Chinese text below; above hang four illuminated digital menu items displaying burgers, fries, combo meals, prices, and Chinese text; near the counter are two people, one with their back to the camera carrying a backpack, and the other partially visible behind the cashier. Use natural color balance, subtle interior grain, realistic reflections, shallow to medium depth of field, and the casual feel of an unstaged social media snapshot.

Example generated from “Realistic Person and Doodle Shadow Portrait”
GPT Image 2en
Realistic Person and Doodle Shadow Portrait

Use the {argument name="character reference" default="uploaded character reference image"} as the strict identity and outfit reference. Preserve the reference character’s: face identity, facial proportions, eye shape, nose, lips, skin tone, hairstyle, hair color, visible hair accessories, overall recognizable vibe, outfit and styling shown in the reference image. Do not hardcode any specific character traits that are not present in the uploaded reference. All identity, hairstyle, accessories, and clothing details must be inferred directly from the uploaded reference image. Create a high-quality mixed-style vertical portrait showing: 1. a realistic full-body version of the uploaded character/person 2. a black hand-drawn doodle-shadow version of the same character/person on the wall beside them. Core concept: The real person and their doodle-shadow are doing a {argument name="pose" default="playful mischievous pose"} together. The real person performs a cute realistic version of the pose, while the doodle-shadow performs a much more exaggerated, chaotic, cartoonish version of the same pose idea. The mood should feel cute, playful, mischievous, stylish, funny, and social-media-friendly. Real person: Must remain realistic and photogenic, wear the same outfit style and key visible details from the uploaded reference image. Expression should be {argument name="expression" default="cute, slightly confused, mildly embarrassed, playful"}, as if thinking: “Why am I doing this with my shadow?” Doodle-shadow: Must be a black hand-drawn sketch version of the same person, drawn directly on the wall. Black sketch line-art only. Should resemble the person through hairstyle silhouette, accessories, outfit silhouette, and pose structure. Add manga-like motion lines, hearts, stars, sparkles, and comic marks around the doodle if helpful. Create a different mischievous pose each time. Possible pose directions: playful idol pose, silly dance pose, leaning sideways with one arm curved overhead.

Example generated from “Realistic Pirate Cosplay Portrait”
GPT Image 2en
Realistic Pirate Cosplay Portrait

A highly realistic, professionally retouched photograph featuring a person dressed in {argument name="character" default="Miss Fortune, the Bounty Hunter in League of Legends"} A female cosplayer. She possesses {argument name="facial features" default="French face"} She had exquisite makeup, green eyes, and a head of fluffy hair.{argument name="hair color" default="red"} She had long, wavy hair. She was dressed in exquisite pirate attire, including an aged black tricorn hat with gold filigree embellishments, a black leather corset with a deep neckline and gold accents, white ruffled off-the-shoulder sleeves, and a wide brown leather belt. In her right hand she held a magnificent sword with brass details.{argument name="prop" default="flintlock rifle"} The background features a wooden mast and rigging of a sailboat against a bright blue sky. Cinematic natural lighting accentuates the rich textures of leather, metal, and skin, creating a photorealistic, edited portrait effect.

Example generated from “Realistic Sam's Club Shopping Snapshot”
GPT Image 2en
Realistic Sam's Club Shopping Snapshot

A highly realistic snapshot taken with a mobile phone inside a bustling Sam's Club warehouse store in China. The composition centers on a young East Asian man, half-length, standing upright facing the camera, his face obscured by a rectangular blurred block for privacy. He wears a plain black short-sleeved T-shirt with a small white Tesla logo on the left chest, and carries a large bag of red apples in his left hand, tied with a blue cable tie. A white tag on the bag reads "envy," with smaller Chinese characters below. Behind him is the brightly lit retail interior, with high industrial ceilings, warehouse shelves, shoppers, shopping carts, and stacks of produce. The blue Sam's Club wall sign is prominently displayed in the upper left corner, reading "Sam's CLUB" and "Sam's Club Membership Store"; behind him is a large blue banner that reads "A Better Life at Sam's" and "Life is Better in the Club." A tall white vertical banner hangs in the upper right corner, proclaiming "Great Products, Great Prices" and adorned with blue and green geometric decorations. The background includes numerous slightly out-of-focus customers to create a realistically crowded shopping atmosphere. Lighting should be bright and even from retail overhead lighting; colors should be natural; the perspective should be eye-level; the aesthetic should have a documentary snapshot feel; the foreground subject should be sharp, while the background should be slightly softened; and the composition should be portrait-like.

Example generated from “Realistic Sam's Club Shopping Snapshot”
GPT Image 2en
Realistic Sam's Club Shopping Snapshot

A candid, highly realistic smartphone photo inside a busy Sam’s Club warehouse store in China. Center the composition on a young East Asian man from mid-torso up, standing upright and facing the camera, with his face obscured by a rectangular blur block for privacy. He wears a plain black short-sleeve T-shirt with a small white Tesla logo on the left chest and holds a large clear plastic bag of red apples in his left hand, tied at the top with a blue band. The apple bag has a white label reading “envy” with small Chinese text beneath it. Behind him is a bright retail interior with tall industrial ceilings, warehouse shelving, shoppers, shopping carts, and stacked produce displays. Prominently show the blue diamond Sam’s Club wall sign at upper left with the text “Sam’s CLUB” and “山姆会员商店,” and a large blue wall banner behind him reading “更好的生活尽在山姆” and “Life is Better in the Club.” On the upper right, hang a tall white vertical banner that reads “优质优价” and “Great Products Great Prices,” with blue and green geometric accents. Include many background customers, slightly out of focus, to create an authentic crowded shopping atmosphere. Lighting should be bright, even retail overhead lighting; colors natural; perspective eye level; documentary snapshot aesthetic; sharp foreground subject with mildly softened background; portrait orientation.

Example generated from “Realistic Social Media Food Photography”
GPT Image 2en
Realistic Social Media Food Photography

Based on the bottom left corner{argument name="subject" default="Hand-Pulled Noodle"} The image generates a picture that looks like it is for {argument name="purpose" default="Instagram during lunchtime"} The location where the photos were taken was {argument name="location" default="Ramen shop"} n=4

Example generated from “Realistic Subway Candid Portrait”
GPT Image 2en
Realistic Subway Candid Portrait

Inside the subway car, a{argument name="person" default="young women"} She was sitting in the seat near the car door, focused intently on her phone, not looking at the camera, her expression relaxed and natural. She had delicate features, smooth skin, and a refreshing air. She was wearing {argument name="clothing" default="Simple white top, black skirt and white sneakers"} Long hair falls naturally. The subject is positioned in the right third of the frame, with a slightly blurred foreground obscuring the left side, creating a candid perspective. The vertical lines formed by the subway doors and handrails guide the eye. The overall lighting is cool-toned subway light, with soft overhead lighting and natural, not overly harsh shadows. There's a shallow depth of field, the subject is sharp, and the background is slightly blurred. The image has realistic camera grain and minor compositional flaws, resembling street photography—not posed, highly realistic, and with high detail. Additional quality keywords: high definition, hyper-realistic, realistic skin texture, natural skin tone, realistic lighting, 35mm lens, portrait photography, cinematic quality.

Example generated from “Realistic Subway Snapshot”
GPT Image 2en
Realistic Subway Snapshot

{argument name="subject" default="A beautiful woman looking down at her phone on the subway"}; {argument name="style" default="Candid photos"}。

Example generated from “Realistic Textbook Desk Scene”
GPT Image 2en
Realistic Textbook Desk Scene

There are two open books on the table, one of which is {argument name="first textbook" default="Primary school Chinese textbook"} The other one is{argument name="second textbook" default="Elementary school math textbook"} The contents of the book are clearly visible.

Example generated from “Realistic Textbook Desk Scene”
GPT Image 2zh
Realistic Textbook Desk Scene

书桌上摆着两本摊开的书,一本书是{argument name="第一本书" default="小学语文课本"},一本是{argument name="第二本书" default="小学数学课本"},课本内容清晰。

Example generated from “Realistic VTuber Merchandise Mockup”
GPT Image 2en
Realistic VTuber Merchandise Mockup

For standard{argument name="character type" default="Vtuber"} create{argument name="items" default="Doujin merchandise"} The style is{argument name="style" default="Real shot"}

Example generated from “Realistic White Dress Mirror Selfie”
GPT Image 2en
Realistic White Dress Mirror Selfie

A realistic indoor mirror selfie shows a slender young woman standing in the warm light of a modern apartment living room. She wears a fitted ivory one-shoulder midi dress with pleats at the torso and hips. She has shoulder-length, dark brown, center-parted hair and wears minimalist jewelry, including two rings, a silver bracelet, and delicate earrings. She holds her phone in one hand, the black case adorned with red cloud patterns obscuring her face. Her body is slightly turned to the side, her posture relaxed, one arm hanging naturally, showcasing her natural feminine silhouette. The image is vertically composed, presenting the perspective of a casual social media selfie. Behind her is a beige fabric sofa with a gray geometric-patterned cushion and a taupe leather handbag. The background includes a textured bronze-brown accent wall, glossy cream-colored tiled floors, a dimly lit hallway to the left (with three framed paintings), and warm recessed ceiling lights. Soft evening ambient light, natural skin tones, realistic fabric tension and folds, delicate reflections in the mirror, clean and upscale apartment aesthetics, high realism, photorealistic skin texture, smartphone shooting perspective.

Example generated from “Realistic White Paper Craft Style Conversion”
GPT Image 2ja
Realistic White Paper Craft Style Conversion

{argument name="対象" default="このイラスト"}を非常に精密でリアルな{argument name="色" default="白い"}紙の{argument name="スタイル" default="ペーパークラフト"}風に変換してください。厚紙をレーザーカットしたようなシャープで正確な切り口、細かい折り目、複数のレイヤーによる豊かな立体感、紙の繊維感と自然な影を高ディテールで表現してください。光の当たり方による紙の白さと微妙なグラデーションを活かし、本物のような立体的なペーパーアート作品のようなリアリティを出して。全体を洗練されていて芸術的な雰囲気に。元のイラストの構図・キャラクターを忠実に維持。

Example generated from “Realistic X Profile Page Mockup”
GPT Image 2en
Realistic X Profile Page Mockup

Create a highly realistic dark mode social media interface mockup, closely replicating the French layout of the desktop X/Twitter, presented as a full browser screenshot with a black background and subtle dark gray dividing lines. The left sidebar contains 11 vertically arranged navigation items with white icons and labels: "Accueil" (Home), "Explorer," "Notifications" (with a small blue badge displaying the number 3), "Messages," "Grok," "Signets," "Communities," "Premium," "Organisations certifiées," "Profil," and "Plus." Place a white X logo at the top of this sidebar, add a large circular white button labeled "Poster" slightly below, and display Martin Dubois's compact login account bar at the bottom, including his username @martindubois and a small circular avatar. The middle bar is {argument name="profile name" default="Martin Dubois"} The main profile page features a back arrow, the name "Martin Dubois," and a blue verification badge at the top, with "1,248 posts" displayed below. Below this is a wide panoramic cover photo showcasing the breathtaking scenery of the misty Alps, forests, and lakes. Overlaid on the lower left of the banner is a circular portrait of a young man in dark clothing, his face slightly blurred or obscured against a light background with a thin border. To the right of the profile header are three small circular action buttons and a prominent light-colored, rounded button labeled "Éditer le profil" (Edit profile). Below the portrait are the display name Martin Dubois, the blue verification badge, the account @martindubois , and a French bio: "Créateur de contenu • Passionné de tech, de design et de productivité." (Content creator • Passionate about technology, design, and productivity.) Next, add a row containing metadata for four items: location "Lyon, France", website "martindubois.com", and "A rejoint X en janvier 2021". This is followed by a row of followers showing "320 abonnements" and "12,4k abonnés". Add a horizontal tab bar with five labels: "Posts", "Replies", "Featured Tweets", "Media", and "Likes", with "Posts" active and underlined in blue. Below the tabs, display a feed containing three visible posts by Martin Dubois. The first post is pinned, marked "Épinglé" (pinned), dated "12 mai" (May 12th), and contains the French text "10 leçons que j'ai apprises en construisant mon projet pendant 2 ans" (10 lessons I learned in 2 years of building this project) with an emoji. The interaction count includes five figures: 24 replies, 142 retweets, 1,000 likes, 85,000 views, and a bookmark/share area. The second post is dated "2h" (2 hours ago), and contains the French text "Le meilleur moment pour commencer était hier. Le deuxième meilleur moment, c'est maintenant." (The best time to start was yesterday. The second best time is now.) It has 12 replies, 37 retweets, 256 likes, and 8,000 views. The third post, dated "1j" (1 day ago), reads in French, "Un week-end à Chamonix, toujours ressourçant" (A weekend in Chamonix is ​​always revitalizing) and includes a mountain emoji along with a large embedded image of a snow-capped mountain peak bathed in soft sunlight. Each post uses a circular avatar on the left and is separated by thin lines. A rounded search box with the placeholder "Chercher" (search) is at the top of the right sidebar. Below this is a card titled "Vous pourriez aimer" (You might be interested in), listing three suggested profiles to follow, each with an avatar, display name, account, blue verified badge, and a white rounded "Suivre" (follow) button: "Théo Catarino" @theocatarino , "Aurélie Martinez" @aurel_martinez , and "Ulrich Rozier" @ulrichrozier . A "Voir plus" (see more) link is added at the bottom of the card. Below, create a trending card titled "Tendances: France," listing four trends by ranking, including category and number of posts: 1. "#RolandGarros" with "32,100 posts," 2. "#FêteDesMères" with "18,400 posts," 3. "Mbappé" with "45,200 posts," and 4. "#LundiMotivation" with "12,300 posts." Add a "Voir plus" link below the trend list. At the very bottom, place a compact footer containing several small gray legal links and "© 2024 X Corp." The overall image should look almost identical to a real desktop screenshot: clean typography, authentic spacing, realistic UI proportions, delicately rounded card corners, black and white text on a dark background, blue accent color, refined anti-aliased icons, and a clean, modern product design aesthetic.

Example generated from “Realistic X-Style French Profile Screen”
GPT Image 2en
Realistic X-Style French Profile Screen

A highly realistic screenshot mockup of a smartphone app showcases a social media interface almost identical to X/Twitter, displayed on a full-screen iPhone-style display with rounded corners and a dark blue-black background. The top status bar displays the time 9:41 on the left and cellular, Wi-Fi, and battery icons on the right. (Screen display){argument name="profile name" default="Marie Dubois"} The French profile page features a large cover image at the top: the Eiffel Tower and the Seine River against a dramatic orange-blue sky at sunset. A circular portrait partially obscures the face, positioned in the lower left corner of the header. A dark, semi-transparent circular back arrow button is located in the upper left corner, and a matching circular three-dot menu button is in the upper right corner. Below the header is the profile name, displayed with a blue verification badge.{argument name="profile name" default="Marie Dubois"} ,account{argument name="handle" default="@MarieDubois"} The feed includes a French description: the first line displays "Journaliste | Paris," the second line displays "✨ Curieuse & voyageuse," and below it displays "Rejoint en mai 2012." Follower statistics are displayed in French: "1,245 Abonnements" and "32,8k Abonnés." A tab bar with four labels—"Publications," "Reponses," "Médias," and "J'aime"—is added, with "Publications" selected and underlined in blue. The feed contains three visible posts. The first post is by {argument name="profile name" default="Marie Dubois"} Released with a blue certification badge and a French title.{argument name="post caption" default="Balade matinale le long des quais de la Seine ☀️ #Paris #Promenade"} The first post is followed by a collage of four images: 1) The Eiffel Tower at Golden Hour along the river, 2) A riverside promenade in Paris with bridges and autumn leaves, 3) A cozy terrace café overlooking the Eiffel Tower, and 4) A cup of coffee next to an open laptop by the water. Below are interaction icons and counts: 258 replies, 642 retweets, and 3,1k likes. The second post comes from the verified news account "Le Monde" ( @LeMonde ), with a profile picture of a black circle with white letters. The French text reads: "Dernières infos : Le nouveau gouvernement français a été annoncé 📰 Tous les details ici ↓", followed by a blue link "lemonde.fr/article123"; below are counts of 187, 512, and 2,7k. The third post, from "Thomas Lefevre" ( @Thomas_Lf ), reads "Juste incroyable! 😍 Le Mont-Saint-Michel vu du ciel 📷," followed by a wide-angle aerial photo of Mont Saint-Michel. The bottom features an iOS-style navigation bar with five elements: a home icon, a search icon, a notification bell marked with a red number 4, a message envelope marked with a red number 4, and a large blue floating post button with a white quill icon. A slim, black iPhone home screen indicator is added at the very bottom center. The design utilizes clean, modern UI typography, subtle gray dividing lines, realistic spacing, authentic X-style icons, beautiful mobile app rendering, and near-photographic screenshot realism.

Example generated from “Reasoning Context and Prompt Cache Slide”
GPT Image 2en
Reasoning Context and Prompt Cache Slide

Goal: Create a clean 16:9 Japanese business/technical infographic slide about how continuous reasoning context and explicit prompt caching improve long-running task efficiency. Use a polished PowerPoint-native infographic look with dense but readable explanatory text. Canvas: 16:9 landscape slide, warm off-white background, dark navy text, teal and olive green accent colors, thin rounded borders, subtle shadows, generous margins. Leave no photo-realistic elements; use flat vector icons and crisp typography. Top text area: Place a large bold Japanese headline at the top left: {argument name="headline text" default="継続推論と明示的キャッシュが長期タスクを効率化"}. Under it, add a smaller semibold lead sentence: {argument name="lead text" default="過去ターンの推論と再利用可能なプロンプトを選択的に保持することで一貫性と効率を高められるが、前提変更や利用回数を踏まえて保存範囲を設計する必要がある。"} Main layout: Two large rounded rectangular panels sit side by side beneath the lead sentence. The left panel is teal and titled “reasoning.context”; the right panel is olive green and titled “明示的プロンプトキャッシュ”. Connect the panels with a thin line and a small circular chain-link icon centered between them near the bottom, suggesting coordination between reasoning context and cache. A small upward green arrow near the right panel points back toward it. Left panel details: Include exactly 3 vertically stacked turn markers on the far left, connected by a vertical line with a downward arrow. The markers are labeled “1ターン”, “2ターン”, and “3ターン”. Next to each marker, show a small person icon and speech bubble. To the right, include exactly 3 horizontal rounded cards with teal icon blocks and Japanese explanatory text: 1) gear icon, “auto|モデルの既定動作を利用”; 2) checklist icon, “all_turns|安定した目標、前提、優先順位を複数ターンで再利用”; 3) clock icon, “current_turn|以前の推論が不要になった場面で現在ターンに限定”. Right panel details: Include exactly 3 horizontal rounded cards with olive icon blocks and Japanese explanatory text: 1) document icon, “再利用する共通プレフィックスを開発者が指定”; 2) database cylinder icon, “キャッシュ書き込みは通常入力単価の1.25倍”; 3) percent tag icon, “読み出しは割引されるため、反復利用の回数が効果を左右する”. Bottom conclusion band: Across the bottom, create one wide rounded rectangle callout with a thin yellow border. On the left of the band place a gold target icon with an arrow. To the right, place a bold Japanese conclusion: {argument name="conclusion text" default="長期タスクの一貫性と効率を高める一方、前提変更時には過去の推論や不要なキャッシュを切り離す設計が必要"}. Visual style: Japanese corporate slide design, crisp vector line icons, rounded rectangles, light cream background, teal left section, olive right section, yellow conclusion accent, strong hierarchy, highly legible Japanese sans-serif typography, no decorative clutter. Constraints: Preserve the exact count of elements: 2 main panels, 3 left turn markers, 3 left option cards, 3 right cache cards, 1 central chain-link connector, 1 bottom conclusion callout, and 1 target icon. Do not add extra panels, extra cards, watermarks, logos, or decorative characters. Keep all slide text inside the infographic sharp and readable.

Example generated from “Recursive King of Spades Card”
GPT Image 2en
Recursive King of Spades Card

Use my uploaded face image as the primary identity reference. Preserve my exact facial identity with extremely high fidelity: identical facial structure, jawline, cheekbones, eye shape, eyebrows, nose, lips, beard pattern, hairstyle, hair texture, skin tone, skin texture, and overall recognizable appearance. Do not beautify, alter, or reinterpret my face. Maintain realistic anatomy and authentic likeness. Create an ultra-detailed, cinematic, surreal luxury playing-card artwork. The main subject is me as the King of Spades, occupying the full design of a magnificent, royal Ace-quality playing card. I wear elaborate black-and-silver spade-themed regalia, a crown forged from obsidian and polished steel, intricate embroidered armor, and flowing royal garments decorated with subtle spade symbols. My expression is calm, intelligent, and powerful. In my right hand, I am holding a pristine Ace of Spades card. The creative twist: the Ace of Spades I am holding is not a normal card. It contains another complete playing-card illustration. Inside that card, I again appear as the King of Spades, and that entire card is being elegantly held between the fingers of a majestic Queen of Hearts. The Queen is graceful, regal, and visually striking, dressed in rich crimson and gold royal attire adorned with heart motifs. The illusion continues with subtle recursive storytelling: the Queen of Hearts appears to be examining the card with fascination, creating a "card within a card" visual paradox. The composition should feel like a legendary tale of power, strategy, love, and destiny intertwined. Add impossible Escher-inspired visual design elements: •Infinite recursion effect •Card-world folding into itself •Floating spade and heart symbols transforming into ravens and rose petals •Ornate golden borders extending beyond physical card edges •Royal chess pieces suspended in midair •Fractal patterns hidden within the card engravings •Elegant smoke forming suit symbols •Dimensional portals emerging from card corners Style: ultra-realistic fantasy realism mixed with luxury casino art, Renaissance royal portraiture, and modern cinematic concept art. Lighting: dramatic chiaroscuro, volumetric god rays, rich contrast, glowing metallic highlights, subtle magical energy around the Ace of Spades. Color palette: •Deep blacks •Silver chrome •Ivory white •Crimson red •Antique gold accents Composition: •Vertical masterpiece •Museum-quality detail •Hyper-realistic textures •Intricate card engravings •Perfectly symmetrical playing-card aesthetics blended with cinematic depth •Sharp focus on my face •Extremely high identity fidelity •Epic storytelling through visual symbolism The final image should feel like the cover of a legendary fantasy card game where the King of Spades has become self-aware, existing across multiple layers of reality while being held in the hands of fate itself, represented by the Queen of Hearts.

Example generated from “Red Alert Style Pixel Sprite Sheet”
GPT Image 2en
Red Alert Style Pixel Sprite Sheet

Please try to analyze all the scenes in Red Alert, and then generate all the resources used in a large image of a pixelated sprite sheet, making sure it does not contain any irrelevant elements so that I can crop it.

Example generated from “Red Idol Concert Close-Up”
GPT Image 2en
Red Idol Concert Close-Up

A dynamic close-up photo of a concert, with the main subject being the performers on stage.{argument name="idol performer" default="Young J-pop female idol"} Shot from the audience's perspective, the scene features a shallow depth of field. She leans towards the camera, one hand reaching towards the foreground, creating a strong perspective compression effect; the other hand holds a black handheld microphone with a red decorative ring close to her mouth. She wears a richly detailed red and white ruffled idol costume: dark brown twin ponytails with bangs, two fluffy red pom-pom hair accessories tied with ribbon bows, red pendant earrings, a red choker with lace and heart-shaped gemstones, a lace-trimmed crop top with a large satin bow at the chest, off-the-shoulder ruffled sleeves, a cinched waist with red ribbon ties and small metal embellishments, a layered satin skirt decorated with sparkling beaded chains, and a heart-shaped gemstone ornament at the waist. Her nails are glossy red with cute white decorative nail art. The lighting is vibrant pink and deep red stage lighting, creating a glamorous live idol atmosphere. The background is filled with soft blurring effects, LED screens, stage setups, and countless glowing and floating heart-shaped confetti, creating a dreamy and romantic concert scene. Employing a photorealistic style, it showcases clear fabric textures, shimmering satin highlights, lace details, concert atmosphere, subtle dynamic energy, and the overall {argument name="color theme" default="Red and pink"} Color scheme. It blends a subtle sense of excitement with a fan-oriented stage presence, like {argument name="theme" default="Cute and romantic idol concert"} Photos of the promotional performance.

Example generated from “Red Triptych Build the Future Poster”
GPT Image 2en
Red Triptych Build the Future Poster

{ "type": "Triptych advertising poster", "style": "A bold and modern toy brand event style, combining high-contrast studio photography with graphic design.", "format": "Three vertical panels of equal width side by side", "color_palette": { "dominant_background": "Vibrant and saturated red", "accent": "Bright yellow", "secondary": "Orange-red, black, white, gray" }, "subject": { "theme": "Creative construction and imagination", "brand_style": "Building block toy aesthetics", "headline": "{argument name=\"headline text\" default=\"BUILD THE FUTURE\"}", "tagline": "{argument name=\"tagline\" default=\"Play. Build. Repeat.\"}" }, "layout": { "sections": [ { "title": "left portrait panel", "position": "Left third", "count": 1, "labels": [ "Children's portraits" ] }, { "title": "Central Build Panel", "position": "The middle third", "count": 1, "labels": [ "Place blocks onto the toy building with your hands" ] }, { "title": "Right-side layout panel", "position": "right third", "count": 1, "labels": [ "Title and Brand Graphics" ] } ], "panel_dividers": "There are thin, dark vertical dividing lines between the three panels." }, "left_panel": { "subject": "Close-up portrait of a child from the chest up", "counted_elements": { "people": 1, "garments": 2, "visible_facial_features": 2 }, "appearance": { "age": "young children", "hair": "Short dark brown hair", "clothing": [ "Red crew neck shirt", "Red zipper jacket with silver zipper teeth" ], "visible_features": [ "Left ear visible", "The neck and lower half of the face are visible." ] }, "framing": "The top is slightly trimmed and located in the center of the panel.", "background": "Solid color, vibrant red, seamless photography studio background" }, "center_panel": { "subject": "Toy building scene being assembled", "counted_elements": { "hands": 1, "main_building": 1, "mini_figures": 6, "trees": 1, "vehicles": 1, "loose_bricks_being_placed": 1 }, "action": "A hand reaches in from the upper left and places a red block on the roof.", "building": { "style": "Modern multi-story residential or municipal buildings constructed from building blocks", "colors": [ "red", "yellow", "grey", "Transparent dark window panel" ], "features": [ "left arch structure", "Single-story roof", "Multiple balconies or terraces", "large windows", "gray baseboard", "Front steps" ] }, "surroundings": { "mini_figures": "Six small toy figures shaped like construction workers are arranged around the structure, dressed in yellow and red clothes.", "trees": "A small green coniferous tree near the left front.", "vehicle": "A small black and red wheeled vehicle located on the lower left edge." }, "background": "Solid color, vibrant red, seamless photography studio background", "surface": "Red desktop that blends into the background" }, "right_panel": { "subject": "Graphic poster layout", "counted_elements": { "headline_words": 3, "small_repeated_word_instances": 9, "decorative_blocks_top_right": 3, "small_horizontal_bar_groups_top": 2, "tagline_lines": 3, "logo_boxes": 1, "decorative_hatched_boxes": 1, "decorative_dot_grids": 1 }, "typography": { "main_text": "{argument name=\"headline text\" default=\"BUILD THE FUTURE\"}", "arrangement": "Three lines stacked, one word per line, in very large, compressed, bold, sans-serif yellow font, occupying most of the panel's height.", "tagline": "{argument name=\"tagline\" default=\"Play. Build. Repeat.\"}", "tagline_position": "The lower left corner of the right panel displays three lines of small yellow text.", "repeated_side_text": "The word BUILD is repeated vertically nine times in the middle right corner, displayed in small yellow text." }, "graphics": { "top_details": [ "A short yellow horizontal line in the upper left corner", "One long, thin yellow striped bar at the top center" ], "top_right": [ "A combination of 3 yellow and orange squares" ], "bottom_details": [ "A yellow shaded rectangle in the bottom left center", "One small yellow dot at the bottom center" ], "logo": "{argument name=\"brand name\" default=\"LEGO\"} The logo is located in the bottom right corner within a thin black outline box." }, "background": "solid bright red" }, "lighting": "The clean, bright, and saturated lighting in the commercial studio, with minimal shadows, creates a sophisticated and polished advertising campaign.", "composition": "The striking triptych composition features a photographic portrait on the left, a product-building action in the middle, and an oversized typographical slogan panel on the right." }

Example generated from “Red Vinyl Soundtrack Ad”
GPT Image 2en
Red Vinyl Soundtrack Ad

A luxury product advertisement photograph with a cinematic quality features a vintage wooden record player playing a translucent, bright red vinyl record in a warm, dimly lit living room. Behind the record player, a vertical album cover dominates the image against a striking distressed red background, with large, cream-white, compressed text printed in the upper left corner.{argument name="album title" default="BELLA CIAO"} Below is smaller black text.{argument name="subtitle text" default="SOUNDTRACK OF THE RESISTANCE"} The cover art features a silhouette of a figure in a black hoodie against a red background, with a row of smaller silhouettes of figures moving along the bottom edge. Subtle architectural lines in the lower left corner create an atmosphere of political resistance and a dramatic soundtrack. The record player is presented in a three-quarter view on a wooden table, featuring a brushed metal platter edge. A black tonearm rests to the right of the record, and a small metal control knob is visible in the front left. The design employs shallow depth of field, high-end photographic lighting, and rich amber highlights. Soft candlelight illuminates the right side of the background, while bookshelves and furniture details are blurred, creating an aesthetic that blends high-end fashion advertising with a music product. At the bottom of the image, a striking promotional banner overlay is added, featuring large, compressed white all-caps text.{argument name="bottom headline" default="SOUNDTRACK OF THE RESISTANCE"} A red rectangular button is placed in the center below, containing white all-caps text.{argument name="button text" default="SHOP NOW"} It contains a small white [item] in the upper right corner.{argument name="brand watermark" default="Pollo.ai"} Watermark. Hyperrealistic photography, with the vinyl record and cover clearly in focus, creating a somber and luxurious atmosphere, suitable for social media advertising, in a vertical 2:3 aspect ratio.

Example generated from “Redacted Rooftop Portrait”
GPT Image 2en
Redacted Rooftop Portrait

A cinematic, realistic rooftop portrait, taken at the golden hour. An older man faces the camera, his face completely obscured by a large, vertical rectangular brown occluder. Visible features are clearly visible: thick, tightly curled grey hair with silver highlights; clearly defined ears; a long, slender neck showing the marks of time; a charcoal grey suit jacket over a tieless, off-white collared shirt. The composition employs a vertical cropping technique starting from the mid-chest. The scene is set on a real New York rooftop, with details including: a weathered HVAC unit in the foreground and left, a cylindrical rooftop water tower in the middle ground, two striking Art Deco skyscrapers in the background (the left resembling the Empire State Building, the right resembling the Chrysler Building), low brick buildings, rooftop pipes, vents, railings, and a soft, atmospheric haze. Using warm sunset lighting, the sky takes on an orange-pink hue, with subtle contour lighting around the ears and hair. The horizon has a shallow depth of field, and the textures of the suit fabric, rooftop concrete, metal machinery, and curly hair are richly detailed. The overall atmosphere is mysterious, editorial, and slightly surreal, like a magazine cover photo where the identity is obscured.

Example generated from “Reference-Based Character Portrait Prompt”
GPT Image 2en
Reference-Based Character Portrait Prompt

exist{argument name="location" default="Taiwan Metro carriages"} Inside, a young woman sits in a seat near the door. Her appearance and clothing must reference the woman in the uploaded photo (including facial features, demeanor, hairstyle, and clothing style; the clothing must be exactly the same as in the photo). She is looking down at her phone, her expression natural, not at the camera. The woman is positioned in the right third of the frame, with a blurred foreground obscuring her left side.

Example generated from “Reference-Based Cinematic World Expansion”
GPT Image 2en
Reference-Based Cinematic World Expansion

Using the provided reference images as the seed of visual civilization, this concept is expanded into a completely new cinematic world. While maintaining the grand East Asian fantasy design language, monumental scale, complex world-building centered on architecture, and ultra-detailed realism, the scene is transformed from a bright, cloud-top palace into ancient mechanical ruins on the ground. The gilded fairyland temple environment is replaced with a massive circular structure, embedded giant gears, heavy industrial-style stone and metal buildings, broken pipes, and ruins overgrown with pale-colored flowering trees. The atmosphere is adjusted to a cold, hazy, low-saturation blue-gray tone, complemented by a soft atmospheric fog, while retaining the epic and breathtaking world-building quality of the reference images. A solitary wandering swordsman, with his back to the camera, wearing a flowing white robe and a scabbard, is placed directly below the largest circular structure in the foreground to emphasize the scale. This gives the scene the quality of a high-budget fantasy film, with extremely rich environmental details and showcasing a coherent visual language derived from the reference images.

Example generated from “Reference-Based Cinematic World Expansion”
GPT Image 2en
Reference-Based Cinematic World Expansion

Using the provided reference image as the visual civilization seed, expand it into a new cinematic world scene that keeps the same grand East Asian fantasy design language, monumental scale, intricate architecture-first worldbuilding, and ultra-detailed realism, but shift the setting from a luminous cloud palace into an ancient mechanical ruin on the ground. Replace the gilded celestial temple environment with colossal circular ring structures, giant embedded gears, heavy industrial stone-and-metal architecture, broken conduits, and overgrown ruins with pale flowering trees. Change the mood to cold, misty, desaturated blue-gray lighting with soft atmospheric fog, while preserving the epic, awe-inspiring worldbuilding feel from the reference. Add 1 lone wandering swordsman seen from behind in the foreground, wearing flowing white robes and carrying a sheathed sword, centered beneath the largest ring to emphasize scale. Make it feel like a frame from a high-budget fantasy film, with extremely rich environmental detail and a cohesive visual language derived from the reference image.

Example generated from “Refreshing Beverage Commercial Portrait”
GPT Image 2en
Refreshing Beverage Commercial Portrait

A bright and refreshing commercial photograph, featuring a {argument name="subject description" default="Handsome young Asian man with short black hair"} Wearing a{argument name="clothing" default="Light blue knitted sweater"} He wore a gentle smile, looking directly at the camera, holding a clear plastic water bottle with water droplets clinging to it. The bottle had a bright blue label with white lettering.{argument name="product label" default="POCARI SWEAT"} The background is divided into two parts: on the left, a clear light blue sky can be seen through a window; on the right, a soft-focus, cozy interior is depicted, complete with a picture frame and a warm-toned table lamp. In the lower left corner, there is clear white overlay text that reads, "..."{argument name="overlay text line 1" default="포카리 스웨트"} Below that is a line of larger text.{argument name="overlay text line 2" default="이온을 채워, 순간을 살아"} The design features a white, curved, dynamic graphic at the bottom. The overall lighting is soft, natural, and translucent, creating a pure and hydrating atmosphere.

Example generated from “Refreshing Summer Udon Ad”
GPT Image 2zh
Refreshing Summer Udon Ad

少し暑くなってきた今の時期に、さわやかにさっぱりしたい、みずみずしさ、みたいなところをもっと強く感じたい。冷たいうどんやナス、つゆを口に含んだ時の爽快感、みたいなものをもっと感じるように

Example generated from “Research-Driven Ad Layout Designer”
GPT Image 2en
Research-Driven Ad Layout Designer

The advertising creative was generated in a 2x2 grid format, producing four different types.{argument name="campaign name" default="Golden Week travel plans"} Advertising Plans. Based on online research of high-conversion advertising campaigns in Japanese domestic tourism over the past 10 years, four distinct and highly appealing advertising designs were designed and created. Aspect Ratio ={argument name="aspect ratio" default="1:1"} Used for A/B testing.

Example generated from “Research-Driven Ad Layout Designer”
GPT Image 2ja
Research-Driven Ad Layout Designer

広告クリエイティブ制作、{argument name="キャンペーン名" default="ゴールデンウィーク用の旅行プラン"}の広告を2x2グリッドで4パターン。日本国内旅行で過去10年間で高い実績を持つプランをwebでリサーチした上で、極めて高い訴求力のある広告デザインを完全に異なるパターンで4種類設計・作成。A/Bテスト前提でアスペクト比={argument name="アスペクト比" default="1:1"}

Example generated from “Restaurant Logo Guidelines Board”
GPT Image 2en
Restaurant Logo Guidelines Board

Using the provided reference images as brand inspiration, transform fried chicken burgers and waffle fries into a beautifully designed one-page logo guide project for a brand called [Project Name].{argument name="brand name" default="Can Dende"} The restaurant brand. Based on the food in the reference image, create a cheerful, modern fast-casual brand image: transform the crispy fried chicken toppings into a playful mascot/icon above the text label, and build a striking custom logo inspired by the warm tones in the photo. Design the output as a clean, editorial-style brand page with an off-white background, navy blue and red font, yellow fan-shaped borders at the top and bottom, and a few touches of glitter doodles. Please include the following sections and quantities: 1 main heading with "LOGO GUIDELINES" and the brand name below; 1 "PRIMARY LOGO" section containing the main logo and a "ABOUT THE LOGO" text block; 1 "LOGO VARIANTS" section containing 3 variations labeled "FULL COLOR," "SINGLE COLOR," and "REVERSED"; 1 "COLOR VARIATIONS" section containing 4 color blocks labeled "YELLOW BACKGROUND," "NAVY BACKGROUND," "PINK BACKGROUND," and "NAVY ON CREAM"; 1 "MINIMUM SIZE" section showcasing the logo with width guide lines and the text "1.25 in / 120 px MINIMUM WIDTH"; 1 "CLEAR SPACE" section containing a spacing diagram based on the height of the letter C; 1 "INCORRECT USAGE" section containing 4... There are 10 error examples, labeled "Don't change the colors," "Don't stretch or distort," "Don't rotate the logo," and "Don't add effects or shadows"; 1 "USAGE NOTES" section containing 4 key points; and a handwritten closing statement in the bottom right corner.{argument name="tagline text" default="It's a vibe. It's a smile. It's yours."} Maintain a clear, printable, highly legible, and presentation-quality layout, just like a real restaurant brand's deliverables.

Example generated from “Restaurant Logo Guidelines Board”
GPT Image 2en
Restaurant Logo Guidelines Board

Using the provided reference image as the brand inspiration, transform the fried chicken burger and waffle fries meal into a polished one-page logo guidelines board for a restaurant brand named {argument name="brand name" default="Can Dende"}. Create a cheerful, modern fast-casual identity derived from the food in the reference: turn the crispy fried topping into a playful mascot/icon above the wordmark, and build a bold custom logo in a warm palette inspired by the photo. Design the output as a clean editorial brand sheet on an off-white background with navy and red typography, yellow scalloped borders at the top and bottom, and small sparkle doodles. Include these exact sections and counts: 1 main header reading "LOGO GUIDELINES" with the brand name beneath it; 1 "PRIMARY LOGO" section with the main logo and an "ABOUT THE LOGO" text block; 1 "LOGO VARIANTS" section with 3 variants labeled "FULL COLOR", "SINGLE COLOR", and "REVERSED"; 1 "COLOR VARIATIONS" section with 4 tiles labeled "YELLOW BACKGROUND", "NAVY BACKGROUND", "PINK BACKGROUND", and "NAVY ON CREAM"; 1 "MINIMUM SIZE" section showing the logo with a width guide and the text "1.25 in / 120 px MINIMUM WIDTH"; 1 "CLEAR SPACE" section with a spacing diagram based on the height of the letter C; 1 "INCORRECT USAGE" section with 4 wrong examples labeled "Don't change the colors", "Don't stretch or distort", "Don't rotate the logo", and "Don't add effects or shadows"; 1 "USAGE NOTES" section with 4 bullet points; and a handwritten-style closing tagline at bottom right that says {argument name="tagline text" default="It's a vibe. It's a smile. It's yours."}. Keep the composition crisp, print-ready, highly legible, and presentation-quality, like a real branding deliverable for a restaurant.

Example generated from “Restaurant POV Change Comparison”
GPT Image 2en
Restaurant POV Change Comparison

A side-by-side comparison image of social media posts against a dark background showcases a test of camera angle changes inside a busy restaurant. At the top is large white sans-serif text that reads:{argument name="headline text" default="Show me the view of this crowded restaurant from behind the bar. Don't change anything in the scene except the perspective."} Two rounded rectangular photos are stacked vertically in the center of the page. The upper image has the large white text "Source" on its left; the lower image has the large white text "Output" on its left. The upper image shows the original restaurant scene from the customer side, looking towards the bar at the back: a warm, upscale, and crowded restaurant at night, with a huge illuminated wall of bottles in the background, wooden ceiling beams, and hanging spherical chandeliers. Bartenders and servers are at work, and the foreground and midground are filled with customers. The lower image shows the same restaurant, maintaining the same lighting, decor, crowd density, bottle wall, chandeliers, and overall atmosphere, but with a changed perspective: the camera is now positioned behind the bar, looking out over the crowded restaurant. The foreground of the lower image includes the bar countertop, glassware, bottles, metal bar tools, and the illuminated POS touchscreen in the left front, reinforcing the "behind the bar" perspective. Both photos were designed to maintain photorealistic quality, warm amber lighting, and a vibrant nighttime atmosphere, ensuring the lower image appears to be the same scene as the upper one, only with a different camera angle. It features a clean social media posting style with ample margins and a neat layout, making it suitable for use as a tweet screenshot or AI image comparison sticker.

Example generated from “Restaurant POV Change Comparison”
GPT Image 2en
Restaurant POV Change Comparison

A side-by-side comparison image against a black background showcases the changing camera angles within the same restaurant scene. At the top, bold white sans-serif text reads: "Show me the perspective of someone standing behind the bar overlooking this crowded restaurant. Don't change anything in the scene except the perspective." Below, vertically centered, are two stacked rectangular photos: the top image is labeled "Source" in large white letters on the left, and the bottom image is labeled "Output" in large white letters on the left. The top photo shows a warm, upscale, and crowded restaurant interior, viewed from the dining area towards the bar. A full wall behind the bar is filled with glowing bottles, and bartenders and customers are at the bar. The ambiance features amber lighting, spherical chandeliers, a wooden ceiling, beige pillars, and diners in the foreground. The bottom photo shows the exact same restaurant, the same customer density, the same warm lighting, the same decor, the same bar shelves, the same spherical pendant lights, and the same overall compositional elements, but the perspective has changed to looking out from behind the bar at the crowded restaurant; the foreground includes the bar with glassware, metal bar tools, bottles, and a point-of-sale (POS) screen visible in the lower left corner, the central area is packed with customers and staff, and the dining area extends into the background. Please maintain the visual impression that only the camera position has changed between the two images, with no other alterations to the scene.

Example generated from “Restore and Colorize Damaged Vintage Photo”
GPT Image 2en
Restore and Colorize Damaged Vintage Photo

Using REFERENCE_0, restore the old damaged family photo into a clean, high-resolution modern scan. Remove the heavy scratches, stains, fading, tears, and surface damage while preserving the original composition, pose, framing, and background curtain. Reconstruct realistic natural skin tones, sharpen details, and colorize the image with believable warm indoor tones. Keep the woman holding the baby in the same position, restore the black sleeveless dress, the baby's light outfit, the reddish-brown hair, and the subtle necklace. Maintain the same portrait crop and photographic realism, as if the original print was professionally repaired and remastered in 4K.

Example generated from “Retro 3D Claymation Desk Setup”
GPT Image 2en
Retro 3D Claymation Desk Setup

A vibrant and charming 3D claymation-style illustration depicts a retro yellow desktop computer setup. The central yellow CRT monitor screen is cyan, displaying…{argument name="OS logo text" default="Microsoft Windows 98"} The monitor displays a logo and seven desktop icons labeled "My Computer," "Online Services," "My Documents," "Outlook Express," "Internet Explorer," "Connect to the Internet," and "Recycle Bin." The bottom taskbar includes a "Start" button and a clock displaying "5:49 PM." A yellow sticky note on the monitor reads, "..."{argument name="monitor sticky note" default="Be Happy"} A small yellow bird perches on top. The computer stand is labeled "CREATIVE PC". In front is a yellow mechanical keyboard labeled "KBD-98", with orange and cream keycaps, and a yellow mouse labeled "MS". To the left is a green desk lamp, a smiling cactus, and a sign that reads "..."{argument name="floppy disk label" default="FUN STUFF"} A blue floppy disk with the word "", and a container with the word "{argument name="mug text" default="COFFEE TIME!"} A blue mug with a smiley face. To the right are two stacked books (green "WINDOWS 98 MANUAL," blue "INTERNET EXPLORER") and two smiling balloons (orange and yellow). The foreground includes a green cutting mat, a yellow pencil, a green tape dispenser labeled "TAPE IT!", two crayons, a pink sticky note that says "Take a Break!", and a white notebook with five "TO DO LIST" items: "1. Be Awesome," "2. Play More," "3. Call Mom," "4. Eat Snacks," and "5. Smile :)." A green and cream clock hangs on the background wall, along with a sign that reads "..."{argument name="poster text" default="HAVE A NICE DAY!"} The white poster featured a smiling sun and the words "Do Your Best!", along with three additional sticky notes: a yellow one that read "Dream Big Work Hard!", a pink one that read "You've Got This!", and a green one that read "Meeting 10:00 AM!"

Example generated from “Retro Anime Gender Swap Visual”
GPT Image 2ja
Retro Anime Gender Swap Visual

提供された画像をオリジナルキャラクターの基準リファレンスとして使用すること。オリジナルキャラクターと、その性別反転版キャラクターが同時に登場する、高品質なアニメキービジュアルを制作すること。オリジナルキャラクターは画面左側に配置すること。性別反転版キャラクターは画面右側に配置すること。両者は一目で同一人物だと認識できる必要がある。顔立ち、目の形、顔の比率、髪型のデザイン言語、髪色、シルエット、表情の癖、性格表現、視覚モチーフ、全体的なキャラクターアイデンティティを高い精度で維持すること。性別反転版は単なるコスプレや衣装交換ではなく、「もし同じ人物が最初から逆の性別として生まれていたら」という自然な解釈でデザインすること。単純に異性用の服を着せるだけにしないこと。衣装は性別に合わせて自然に再設計しながらも、元デザインのファッション性、色彩設計、アクセサリー、象徴性、職業性、社会的立場、世界観、デザイン哲学を維持すること。観客が即座に「同じ人物の男女版だ」と理解できること。両デザインには強い視覚的共通性を持たせること。【全体の雰囲気】1990年代後半〜2000年代前半の日本アニメ雑誌やアニメ販促ポスターのような空気感。存在しない名作性別反転ラブコメアニメの公式キービジュアル。楽しい混乱。軽快な騒動。ラブコメ的な距離感。コミカルなテンポ感。深刻なSF作品ではない。深刻な並行世界作品ではない。世界の危機ではない。何か予想外のトラブルが発生してしまったような楽しい空気感を優先すること。視聴者が「絶対面白いやつだ」と思えるようなアニメ作品の宣伝ビジュアルを目指すこと。【キャラクター演技】二人のキャラクターは、同じ人物の男女版が偶然出会ってしまった状況として描写すること。互いを認識していること。互いにリアクションしていること。感情表現は自由に選択してよい。例:・驚き・混乱・困惑・苛立ち・好奇心・照れ・対抗心・自己嫌悪・自画自賛・呆れ・挑発・不信感・警戒・変な親近感など。複数の感情が混ざっていてもよい。重要なのは、「自分自身の性別反転版を見ている」という面白さが伝わることである。【構図】構図は固定しないこと。毎回異なる演出を選択すること。例:・向かい合う・背中合わせ・肩を組む・言い争う・睨み合う・指差し合う・鏡を見るように見つめ合う・同じポーズを取る・全く違うポーズを取る・お互いを観察する・追いかけ回す・逃げる/追う・偶然ぶつかる・入れ替わり直後の混乱状態・同じ物を取り合う・同時にリアクションする・互いの服装を確認する・距離を取って警戒する・近付きすぎて気まずくなるなど自由に変化させること。左右対称構図に依存しないこと。毎回異なるアニメ作品の宣伝ビジュアルになるように構図を変化させること。【背景】背景はキャラクターの世界観、職業、時代設定、ジャンル、文化圏、デザイン言語、に自然に適合すること。背景はキャラクターを引き立てるための高品質なアニメ背景として設計すること。必要に応じて、・学園・都市風景・商店街・ファンタジー世界・和風空間・魔法世界・サイバーパンク都市・レトロポップデザイン・遊園地・文化祭・ステージ・抽象グラフィックなどを使用してよい。背景は作品ジャンルを自然に伝えるものであること。【画風】1990年代後半〜2000年代前半のアニメ雑誌イラスト。レトロポップアニメ美学。セル画風アニメ塗り。クリーンな線画。大胆なシルエット設計。鮮やかで楽しい色彩設計。商業アニメの公式販促イラスト品質。アニメオープニング映像のワンシーンのような躍動感。雑誌表紙のような強いアイキャッチ。【禁止事項】ロゴ禁止。タイトル文字禁止。キャプション禁止。ウォーターマーク禁止。比較表レイアウト禁止。設定資料集レイアウト禁止。ターンアラウンドシート禁止。キャラクター比較画像にしないこと。シリアスなマルチバース戦争にしないこと。ダークファンタジー終末決戦にしないこと。性別反転そのものよりも戦闘や世界観説明を主役にしないこと。

Example generated from “Retro Anime Pastel Sleep Scene”
GPT Image 2en
Retro Anime Pastel Sleep Scene

A sheet using{argument name="art style" default="90s Retro Anime Aesthetics"} The illustration depicts a cute girl sleeping peacefully in bed. She has fluffy hair.{argument name="hair color" default="Soft rainbow colors"} Her hair was disheveled, and a pair of large, retro sunglasses perched on her forehead, the frames yellow with purple gradient lenses. She wore brightly colored printed pajamas, heart-shaped earrings, and clutched a pink blanket adorned with soft rainbow, heart, and star patterns. The bedroom was bathed in {argument name="color palette" default="Vibrant neon soft colors"} The lighting is neutral, with a soft glow and shimmering effect. In the upper right corner is a glowing cloud-shaped sign that reads "..."{argument name="neon sign text" default="Sweet Dreams"} The window to the left revealed a deep blue night sky, with a crescent moon and stars visible above the city silhouette. Star decorations hung from the ceiling. Beside her on the bed lay a small yellow plush toy with a pink bow. On the bedside table in the lower right corner were a star-shaped lamp, a heart-shaped mug, and a book with the title "..."{argument name="book title" default="Good night"} ".

Example generated from “Retro Audio Concept Poster Board”
GPT Image 2en
Retro Audio Concept Poster Board

A minimalist retro-futuristic showcase project, set against a soft periwinkle blue backdrop, employs a minimalist editorial layout, horizontally arranging four analog-style audio product concept visuals. On the far left is a partially cropped reference image showcasing a retro open-reel tape recorder and a silver stereo receiver on a wooden stand against a warm off-white wall; this is complemented by two large white scrolls with dark green centers, brushed metal knobs, a blue backlit display, and hardware in light mint green and silver. To the right are three equally spaced, rounded, tall poster cards, all using the same cream-colored paper tone, deep forest green typography, mint green product shapes, and soft studio shadows. The first poster: a spherical retro speaker/radio sits on a base, with a light mint green body, a round front, a small display, and two side knobs, with a bold, capitalized title displayed at the top.{argument name="headline text 1" default="SOUND, REIMAGINED."} The second poster features a mint green retro oscilloscope-style device with a CRT screen displaying waveforms, silver knobs, switches, and a metal tilting stand, with a bold, capitalized title displayed centered above.{argument name="headline text 2" default="WAVEFORM"} The third poster: A compact, mint green portable cassette recorder/speaker with a handle, front cassette window, buttons, speaker grille, and silver knobs, with a bold, capitalized title displayed at the top center.{argument name="headline text 3" default="ANALOG STILL HITS."} Each poster features micro-typography, a faux brand logo, and a footer icon, presenting a sophisticated Swiss modern advertising style. A restrained color palette is used, employing shades of gray-blue, cream, sage mint green, forest green, silver, black, and warm wood tones. Thin horizontal guide lines are added to the background, and small footer text is added at the bottom, with [text missing] on the left.{argument name="left footer text" default="MJ REFERENCE"} The right side is{argument name="right footer text" default="KREA // GPT 2"} Overall Appearance: A high-concept design project that translates high-fidelity inspiration into a consistent aesthetic presentation across three exquisite product posters, featuring soft lighting, sophisticated graphic design, minimalist layout, clear typography, and a subtle touch of nostalgic realism.

Example generated from “Retro Audio Concept Poster Board”
GPT Image 2en
Retro Audio Concept Poster Board

A clean retro-futurist presentation board on a muted periwinkle-blue background with a minimalist editorial layout, showing 4 analog-inspired audio product concept visuals arranged horizontally. On the far left, a partially cropped reference image of a vintage reel-to-reel tape deck and silver stereo receiver on a wooden shelf against a warm off-white wall; two large white reels with dark green centers, brushed metal knobs, a blue-lit display, and pale mint and silver hardware. To the right, 3 tall rounded-corner poster cards evenly spaced, all using the same cream paper tone, dark forest-green typography, mint-green product styling, and soft studio shadows. First poster: a spherical retro speaker/radio on a pedestal base, pale mint body with a circular front face, small display, and two side knobs, centered under a large bold uppercase headline reading {argument name="headline text 1" default="SOUND, REIMAGINED."}. Second poster: a mint-green vintage oscilloscope-like device with a CRT screen displaying a waveform, silver knobs and switches, and a metal tilt stand, centered under a bold uppercase headline reading {argument name="headline text 2" default="WAVEFORM"}. Third poster: a compact mint-green portable cassette recorder/boombox with handle, front cassette window, buttons, speaker grille, and silver knobs, centered under a bold uppercase headline reading {argument name="headline text 3" default="ANALOG STILL HITS."}. Each poster includes tiny microtype, fake brand marks, and footer icons in a refined Swiss-modern advertising style. Use a restrained palette of dusty blue, cream, sage mint, forest green, silver, black, and warm wood. Add thin horizontal guide lines across the background and small footer text at the bottom reading {argument name="left footer text" default="MJ REFERENCE"} on the left and {argument name="right footer text" default="KREA // GPT 2"} on the right. Overall look: premium concept design board, consistent aesthetic translation from analog hi-fi inspiration into three polished product posters, soft lighting, high-end graphic design, minimal clutter, crisp typography, and subtle nostalgic realism.

Example generated from “Retro ChatGPT Space Poster”
GPT Image 2en
Retro ChatGPT Space Poster

A vintage medieval space race poster, affixed to a rough white plaster wall, is photographed directly in natural light. Slightly wrinkled, creased, and weathered, with worn edges and small tears, the poster presents a realistic, aged paper texture. The design employs a limited vintage color palette of deep navy, faded cream, soft reddish-orange, and grayish-blue. A prominent headline in the upper left corner dominates the design:{argument name="headline text" default="ChatGPT Images 2.0"} The text features "ChatGPT" in an oversized cream-colored sans-serif font, with "Images 2.0" below in bold reddish-orange. On the right, a towering vintage rocket stands vertically, launching upwards through cream-colored smoke, accompanied by vertical lettering on its fuselage.{argument name="rocket text" default="OpenAI"} The color is reddish-orange. A thin orbital trajectory line runs from left to right across the upper half of the poster, encircling the rocket. Slightly to the left of center, a small, detachable spacecraft floats along the trajectory path. The lower left shows the Earth's curved horizon, presented in a retro illustration style, with cyan oceans and cream-colored land. At the bottom is a long, solid reddish-orange bar containing evenly spaced small cream-colored text.{argument name="tagline text" default="EXPLORING TOMORROW"} The design should include four visible stars or celestial objects in a dark sky: two tiny starbursts, one small cream dot, and one larger cream dot. The overall composition should evoke the feel of a 1950s-1960s science fiction travel poster or Soviet/American space age advertising print, featuring silkscreen textures, subtle creases, nostalgic typography, and a cinematic retro-futuristic atmosphere.

Example generated from “Retro Chinese Catering Badge Logo”
GPT Image 2zh
Retro Chinese Catering Badge Logo

请根据用户输入的【{argument name="品牌名" default="品牌名 / 项目名"}】【{argument name="产品名" default="副标题 / 产品名"}】【类型 / 行业】【地域风味】【品牌定位】【核心关键词】【核心图形】【核心食物 / 品类元素】【情绪气质】【主色调】【辅助色】【画幅比例】,设计一个高完成度的「国潮复古餐饮徽章 Logo / Chinese Retro

Example generated from “Retro Constructivist Fight Poster”
GPT Image 2en
Retro Constructivist Fight Poster

A rough, retro-style poster illustration, using a limited palette of black, deep red, and aged beige paper textures, presents a bold, vintage screen-print action poster feel. It features two stylized women in a dramatic diagonal composition, frozen in a moment of combat. On the left is a woman in a short white uniform and a sharply angled nurse's cap, knee raised, arm bent, lunging forward, her body exhibiting a strong perspective shortening effect. On the right is a woman in a fitted black dress and a wide-brimmed black hat, leaning back in a defensive stance, one arm raised above her head, the other close to her torso, long black hair flowing behind her. Both figures are rendered as hard-edged geometric shapes with exaggerated limbs, sharp shadows, and faces in a high-contrast film noir style. A radical red diagonal beam sweeps across the background, framing the scene with black architectural lines and abstract constructivist shapes. Aged paper textures, ink wear, and slight misregistration effects are added to create a vintage print feel. A blurred monkey head badge is placed in the upper right corner. Add three large, blocky poster text lines in the bottom left corner: "SCOTTISH RECIPE:", "MINCE AND", and "TATTIES". The first two lines are beige, and the last line is a striking red, using a heavy, square sans-serif font. Add a graphic element resembling a red cross in the bottom right corner. The overall atmosphere is violent, satirical, cinematic, and visually impactful, like an advertisement.{argument name="dish name" default="Scottish Recipe: Mince and Tatties"} A poster for a Soviet Constructivist vulgar film.