GPT Image 2 prompts

Create a realistic studio portrait of {argument name="subject" default="a bald middle-aged man"} from the chest up, centered against a flat {argument name="background color" default="muted mustard yellow"} backdrop. The subject faces forward with a slight three-quarter turn, ears and the top of a shaved head visible, wearing {argument name="outerwear" default="a beige canvas work jacket"} over {argument name="layers" default="a black crewneck shirt and a burnt orange ribbed turtleneck"}. Cover the entire face with one large perfectly square opaque censor block centered over the head, extending from just below the scalp to below the chin; the square has a smooth skin-tone gradient from warm tan on the left to darker brown on the right, with no facial features visible. Use soft even studio lighting, natural fabric texture, shallow portrait framing, high realism, clean minimal composition, no text, no logo, no watermark.

Create a cinematic 3D miniature architectural diorama map of {argument name="city" default="London"} at night, presented as a physical museum archive board on a dark tabletop. The board is a wide landscape rectangle viewed from an elevated oblique angle, with beveled edges, deep navy paper, thin bronze-gold cartographic linework, a coordinate grid, and warm amber city lights. Render a dense low-relief city street map with the river cutting through the middle, raised landmark buildings and bridges, realistic miniature textures, and a refined historical-atlas aesthetic. Use exactly 6 tall callout pins along the top edge labeled: “BUCKINGHAM PALACE”, “BIG BEN / PALACE OF WESTMINSTER”, “ST. PAUL’S CATHEDRAL”, “THE SHARD”, “THE GHERKIN”, and “CANARY WHARF”. Include additional small map labels across the city such as Regent’s Park, Paddington, Marylebone, Oxford Street, Hyde Park, Green Park, St. James’s Park, Buckingham Palace, Victoria, Westminster, Westminster Abbey, City of London, Southwark, The Shard, The Gherkin, London Bridge, Tower Bridge, The Tower of London, Bermondsey, Greenwich, Kensington, and “RIVER THAMES” following the water. Emphasize the raised landmarks: Buckingham Palace lit at lower left, the Palace of Westminster and Big Ben beside the river, St. Paul’s dome near the center, the Shard as a glowing glass pyramid tower, the Gherkin as a dark rounded tower, Canary Wharf as a cluster of golden high-rises, and Tower Bridge spanning the river. At the bottom of the board, create an information panel with exactly 5 main compartments: a legend box with small icons for major landmark, historic building, park/green space, water body, bridge, and railway station; a central title block reading “{argument name="map title" default="LONDON — NOCTURNAL URBAN ARCHIVE"}”; a scale bar marked in miles and kilometers; a compass rose with N; and a small index-grid inset plus coordinate-system notes. Add the subtitle “LANDSCAPE ARCHIVE MODEL / 地景档案模型” and archive code “{argument name="archive code" default="LDN-NAM-2024-001"}”. Lighting should be moody and nocturnal, with gold edge highlights, tiny illuminated windows, glowing roads, dark blue water, soft shadows, shallow depth of field, and highly detailed miniature craftsmanship. Keep all typography crisp, elegant, and atlas-like; avoid people, vehicles larger than tiny boats, modern UI elements, logos, or watermarks.

Goal: Create a cinematic 3D miniature architectural map of {argument name="city name" default="New York City"}, presented as a nocturnal urban archive board with raised buildings, glowing streets, waterways, landmark callouts, and technical cartographic annotations.

Canvas: Landscape 16:9 image, viewed from an elevated oblique angle as if looking down at a physical diorama map on a dark tabletop. Use a deep navy, charcoal, warm brass, and amber-light palette. The board has a thick dark frame, thin gold border lines, grid coordinates along the top/bottom and left/right edges, and a premium museum-archive feel.

Main subject: A highly detailed raised-relief 3D map of Manhattan and surrounding boroughs at night. Manhattan is centered and built as a dense miniature city with hundreds of extruded buildings, warm window lights, streetlight networks, bridges, docks, and dark reflective water. The tallest illuminated focal tower is the {argument name="central landmark" default="Empire State Building"} near the center, with a bright warm glow around Midtown. The southern tip includes a tall One World Trade Center tower. A long dark rectangle labeled Central Park sits in upper Manhattan. Surrounding flat map areas show New Jersey to the left, Queens to the right, and Brooklyn at lower right, with street grids drawn as thin golden lines.

Top title block: In the upper-left corner, add a boxed archive label reading exactly “NEW YORK CITY” with the subtitle “NOCTURNAL URBAN ARCHIVE” and a small archive code line “NYC-NAM-2024-001.” Add tiny header text along the top edge such as “LANDSCAPE ARCHIVE MODEL / 地景档案模型,” column letters across the top, and a small study label at the upper-right reading “EAST RIVER CORRIDOR STUDY” and “EDITION 01.”

Callouts and labels: Include exactly 11 named map labels/callout boxes: 1) “NEW JERSEY / HUDSON COUNTY,” 2) “HUDSON RIVER,” 3) “ONE WORLD TRADE CENTER / 1,776 ft / 541 m,” 4) “TIMES SQUARE / THEATER DISTRICT,” 5) “EMPIRE STATE BUILDING / 1,454 ft / 443 m,” 6) “CENTRAL PARK,” 7) “CHRYSLER BUILDING / 1,046 ft / 319 m,” 8) “EAST RIVER,” 9) “BROOKLYN BRIDGE / 1,595 ft / 486 m,” 10) “QUEENS / LONG ISLAND CITY,” 11) “BROOKLYN / DOWNTOWN BROOKLYN.” Use thin leader lines connecting callout boxes to landmarks.

Bottom information panels: Add exactly 4 technical panels along the bottom edge: 1) a legend panel with 6 symbol entries labeled “PRIMARY ROAD,” “SECONDARY ROAD,” “RAIL LINE,” “WATER BODY,” “PARK / GREEN SPACE,” and “MAJOR LANDMARK / BUILDING FOOTPRINT”; 2) a north arrow panel showing true north and grid north; 3) a scale panel labeled “SCALE 1:25,000” with imperial and metric scale bars; 4) an archive notes panel with small text including “ARCHIVE NOTES,” “This model represents a nocturnal capture of New York City urban form, infrastructure, and landmark illumination,” plus boxes for “ARCHIVE CODE NYC-NAM-2024-001,” “MODEL NO. 001,” and “SHEET NO. 01 / 01.”

Visual style: Ultra-detailed photorealistic miniature model mixed with technical blueprint/map design, crisp micro typography, precise gold linework, physical paper texture, slight bevels and shadows, moody night lighting, realistic reflections on rivers, and warm amber building illumination. The city should feel like a carefully directed miniature urban world, not a flat map.

Constraints: Keep all text in English except the tiny bilingual archive header already specified. Do not add people, vehicles as focal subjects, logos, or watermarks. Maintain consistent architectural-diorama language across the entire image. Use {argument name="lighting mood" default="dark nocturnal amber city lights"} and an archival cartography aesthetic with {argument name="map base color" default="deep navy blue"} background and {argument name="linework color" default="aged brass gold"} annotations.

Goal: Create a premium 3D architectural diorama map of {argument name="city name" default="Paris"} at night, presented as a physical archival board titled {argument name="map title" default="PARIS — NOCTURNAL URBAN ARCHIVE"}.

Canvas: Landscape 3:2 view, photographed from an oblique overhead angle on a dark tabletop. The map is a thick rectangular black board with beveled edges, thin gold border lines, coordinate grid labels along the top and sides, and a refined museum-archive aesthetic.

Visual style: Dark navy and charcoal physical map texture, raised miniature city blocks, engraved streets, warm golden illumination tracing major boulevards and intersections, glossy deep-blue water, subtle cinematic shadows, high-detail miniature model craftsmanship, luxury cartography, brass-and-ink color palette.

Main map content: Show a nighttime miniature urban plan with illuminated road networks and raised landmark models. Include exactly 4 prominent 3D landmarks: 1) Eiffel Tower glowing gold in the lower-left/central-left area near the river, 2) Arc de Triomphe in the upper-left area with radiating avenues, 3) Notre-Dame-like cathedral on an island or riverbank near center-right, 4) Sacré-Cœur-like basilica on a hill in the upper-right area. Include one winding river crossing the map diagonally from left to lower-right, with multiple bridges, dark reflective water, and golden riverside lights.

Map labels: Use elegant small-cap serif labels in gold ink. Include exactly 20 arrondissement-style district labels: I Louvre, II Bourse, III Temple, IV Hôtel de Ville, V Panthéon, VI Luxembourg, VII Palais Bourbon, VIII Élysée, IX Opéra, X Entrepôt, XI Bastille, XII Reuilly, XIII Gobelins, XIV Montparnasse, XV Vaugirard, XVI Passy, XVII Batignolles-Monceau, XVIII Butte-Montmartre, XIX La Villette, XX Ménilmontant. Also label major routes and areas including Champs-Élysées, Boulevard Haussmann, Boulevard Saint-Germain, Boulevard Voltaire, Bois de Boulogne, and the River Seine.

Bottom information band: Add a wide cartographic footer divided into exactly 5 panels: 1) left legend box, 2) compass rose box, 3) large centered title block, 4) scale bar box, 5) right inset-map box. The legend box must list exactly 5 items: Monument, Major Axis, Boulevard, Bridge, Park / Garden, Water. The centered title block contains the map title, a subtitle reading "CITY OF LIGHTS · HISTORIC CORE · HAUSSMANN NETWORK", and an archive code reading "ARCHIVE CODE: PAR-NAM-2024-001". The scale panel shows "SCALE 1:10,000" and a 0 to 1000 m scale bar. The right inset panel shows a small grid map and a tiny outline map labeled "CENTRAL PARIS FRANCE".

Constraints: Keep all typography crisp and plausible, no people, no vehicles, no modern UI, no photoreal skyline beyond the raised map surface. The entire image should feel like directing a miniature nocturnal urban world: layered landmarks, physical map materials, coherent architectural-diorama language, and warm night lighting.

Goal: Create a highly detailed 3D miniature diorama map titled {argument name="map title" default="TOKYO — NOCTURNAL URBAN ARCHIVE"}, showing Tokyo at night as a physical architectural landscape archive model with glowing city lights, raised buildings, roads, rail lines, waterways, and landmark labels.

Canvas: Landscape 3:2 composition, viewed from an elevated oblique angle like a museum display table. The map sits on a dark tabletop as a thick rectangular board with beveled edges, a dark navy-black base, thin cream border lines, and a printed coordinate grid.

Main scene: A dense nocturnal cityscape of {argument name="city" default="Tokyo"} rendered as a miniature urban model. Use thousands of tiny buildings and warm window lights, fine street networks, rail corridors, expressways, and dark blue water surfaces. The overall feeling should be a premium physical map texture, not a flat digital map: raised terrain-like urban fabric, tiny architectural blocks, engraved labels, and realistic depth.

Landmarks and labels: Include exactly 10 black label plaques with bilingual Japanese/English styling, placed over the city: 1) Shinjuku, 2) Ueno, 3) Asakusa, 4) Tokyo Skytree, 5) Imperial Palace, 6) Shibuya, 7) Tokyo Tower, 8) Ginza, 9) Odaiba, 10) Tokyo Bay. Make Tokyo Tower glow orange near the center-left, Tokyo Skytree glow blue-violet on the right, the Imperial Palace appear as a large dark green park enclosure near the center, and Rainbow Bridge span the bay in the lower right.

Grid and map markings: Add a printed grid around the map with numbered columns 1 through 10 along the top and lettered rows A through E down the left side. Add subtle latitude and longitude markings along the edges. Use thin cream cartographic lines and small technical annotations.

Bottom information band: Along the bottom edge, create a wide printed legend and archive panel divided into sections. Include exactly 6 legend line items with colored line samples: 1) Yamanote Line, 2) Chuo Line, 3) Ginza Line, 4) Asakusa Line, 5) Yurikamome Line, 6) Water. Include a center title block reading {argument name="map title" default="TOKYO — NOCTURNAL URBAN ARCHIVE"}, with subtitle text “Landscape Archive Model” and archive code {argument name="archive code" default="TKY-NAM-2024-001"}. Add a scale bar labeled 1:25,000 and a north arrow in the lower right. Add a small index grid with cells labeled A1 through E5.

Visual style: Cinematic macro photography of a handcrafted architectural model, ultra-detailed, sharp foreground, slight atmospheric haze, realistic miniature lighting, warm gold and cool blue contrast, premium museum exhibit aesthetics, dark navy and charcoal color palette with cream typography. Emphasize layered landmarks, physical map textures, and consistent architectural-diorama language.

Constraints: Keep the image free of people and vehicles as dominant subjects. Do not add extra landmark labels beyond the 10 listed. Avoid cartoon styling, flat infographic aesthetics, or modern app UI elements. Make all text appear printed or as small physical plaques on the map board.

Using REFERENCE_0 as the base, transform the real convenience-store night photo into a Korean lifestyle magazine illustration with a watercolor-and-ink sketch aesthetic. Preserve the same subject, pose, outfit, convenience-store setting, rainy night window view, ramen cup, chopsticks, triangular kimbap, and drink case, but make the scene wider and more editorial, with cinematic Seoul night lighting, hand-drawn texture, soft paper grain, and scrapbook collage annotations.

Goal: Create a cozy Korean convenience-store diary spread titled {argument name="main title" default="서울의 밤"}, like a magazine page about late-night convenience-store food.

Style: Watercolor wash + fine ink outlines, warm indoor convenience-store light contrasted with wet neon city lights outside. Keep the face softly anonymized/blurred as in the reference. Add torn-paper edges, masking-tape style labels, doodled hearts, moons, stars, arrows, and handwritten Korean notes.

Composition additions: Expand the canvas into a square illustrated poster. Show a more detailed Seoul street outside the window with glowing signs including “SEOUL” and “GS25,” wet pavement reflections, and a fuller convenience-store refrigerator wall on the right. Add a small wooden table surface in front of the subject.

Add exactly 5 illustrated food/lifestyle feature items around the scene: 1 large Shin Ramyun cup close-up, 1 triangular kimbap/onigiri close-up, 1 strawberry milk carton, 1 small compact camera doodle, and 1 small taped Polaroid photo labeled “Seoul Night.”

Add exactly 12 scrapbook annotation blocks, arranged around the subject without covering the main face or hands: 1 top-left title block with {argument name="main title" default="서울의 밤"}; 1 pink label reading {argument name="mood label" default="편의점 감성 ♡"}; 1 checklist block titled “오늘의 메뉴” listing three checked items, “신라면 작은컵,” “참치마요 삼각김밥,” and “딸기우유”; 1 small sign callout “GS25”; 1 speech bubble near the window reading “역시 갓진라면! 국물까지 완벽해..♡”; 1 top-right pink tape quote reading “비 오는 밤, 편의점은 나만의 작은 아지트..♡”; 1 white ramen note saying the warm ramen tastes like a small comfort after a tiring day; 1 note pointing to the kimbap saying tuna mayo inside convenience-store samgak kimbap is the best combination; 1 note pointing to strawberry milk saying sweet strawberry milk completes the meal; 1 pink playlist memo titled {argument name="playlist title" default="오늘의 플레이리스트"} with four short bullet songs; 1 bottom diary caption reading “서울의 밤은 언제나 특별해요. 혼자여도 괜찮아, 이런 시간이 소중하니까. 내일도 힘내보자! 화이팅 ♡”; and 1 bottom-right note beside the Polaroid reading “오늘의 한줄 작은 행복들이 모여서 내 하루를 빛나게 해준다.”

Text treatment: Use casual handwritten Korean for the notes, with readable but organic lettering. Keep annotations decorative and magazine-like, not like a clean infographic.

Constraints: Do not change the subject’s clothing or pose significantly. Do not replace the convenience-store meal. Do not make it photorealistic; the result should clearly look like an illustrated Korean lifestyle magazine spread.

Goal: Create a dark, vertical anime collage illustrating {argument name="mood" default="my current incompetent state"}: an overwhelmed student/programmer trapped inside a gloomy machinery-and-productivity nightmare.

Canvas: Tall vertical poster, about 9:16 aspect ratio, black background with low-contrast charcoal gray details, muted sepia highlights, desaturated colors, soft film grain, vignette, smoky atmosphere, dense but readable collage composition.

Central subject: A tired, emotionless anime girl with {argument name="character appearance" default="short black hair, pale skin, heavy-lidded eyes, oversized dark hoodie"}. Show her multiple times: 1 small bust at the upper center tangled in cables, 1 chibi/sleepy version near the lower middle reaching forward from inside gears, and 1 full-body version at the bottom lying exhausted on a messy bed or blanket surrounded by papers and trash. Her expression should stay flat, drained, and sleep-deprived.

Main background and machinery: Fill the upper half with server racks, tangled black cables, smoke vents, and steampunk gears. Include exactly 4 prominent antique clock faces: 1 large clock at upper left, 1 medium pocket-watch clock near center-left, 1 large hanging clock in the vertical center, and 1 smaller clock integrated among gears. Add chains, cogwheels, paper airplanes, scattered notes, and faint sparks.

Productivity and school-task UI collage: Place exactly 8 floating interface/task panels across the middle area, all dim and slightly skewed like screenshots pasted into the scene: 1 OneNote window on the left with Chinese class notes and bullet points, 1 Microsoft To Do panel in the upper middle with several overdue tasks, 1 torn sticky-note cluster on the upper right with daily routines, 1 small chat window labeled like friends on the left, 1 class-group chat window below it, 1 small blue homework/deadline card on the right, 1 cloud-classroom/deadline card below it, and 1 red warning or reminder message card below those. The visible text may be mostly Chinese, but keep it tiny, imperfect, and secondary to the mood.

Media, fandom, and distraction strip: In the lower-middle area, include exactly 5 distraction panels/items in a row: 1 small anime sleep panel with three chibi characters, 1 dark computer/software panel, 1 music-production app icon resembling FL Studio, 1 grayscale band/anime screenshot panel with Chinese title text, and 1 grayscale panel labeled with an excited phrase like “MyGO!!!!!”.

Lower activity scenes and clutter: Add exactly 6 bottom-life vignettes/items around the exhausted character: 1 baseball game scene at lower left, 1 crumpled thought scribble nearby, 1 coding laptop at lower right showing terminal output, 1 electronics/breadboard or soldering setup with smoke and wires near center-right, 1 plate-and-bowl meal scene at far right, and 1 snack pile at lower left with potato chips and melon bread packaging. Scatter additional notebooks, phones, cans, tissues, receipts, and small papers around the final lying figure.

Visual style: Melancholic anime illustration, gothic productivity hell, detailed collage, dark academia meets cyberpunk servers and steampunk clocks. Use mostly black, charcoal, dusty brown, faded gray, muted blue, and occasional tiny warm lights. Keep everything intentionally dim, cluttered, and emotionally heavy, with the central character as the only recurring human focal point.

Text constraints: Use a few legible UI labels such as {argument name="task app label" default="Microsoft To Do"}, {argument name="note app label" default="OneNote"}, and {argument name="band panel label" default="MyGO!!!!!"}. Do not add a clean infographic layout, bright colors, watermark, or photorealistic style.

Create a dark, cinematic isometric anime illustration of {argument name="character name" default="a lonely exhausted young man"} in a tiny futuristic capsule apartment floating in a black void. The room is shown as a cutaway 3/4 overhead diorama, with thick rounded sci-fi walls, a raised loft bed along the back wall covered in rumpled dark bedding, a cramped desk under it, an open laptop showing lines of code, small glowing monitors, books, cables, notes, empty cups, snacks, takeout containers, laundry, trash, and clutter covering the floor. The character lies face-down on the messy floor in the center-left, barefoot, wearing a dark T-shirt and shorts, one arm stretched toward the desk, looking drained and incompetent. Use dim warm desk-lamp lighting as the main light source, with heavy shadows, muted browns and charcoal blacks, soft painterly anime detail, melancholy late-night atmosphere, and a sense of burnout and isolation. Include visible room details: a wardrobe with hanging shirts on the left, a small shelf with three cute figurines or plush mascots, a ladder leading to the loft bed, a narrow kitchenette or utility corner on the right with sink, bowls, bottles, and a wall calendar, plus scattered papers and electronics everywhere. Around the room in the surrounding black void, add exactly seven faint miniature memory/island vignettes: 1) a lit sports stadium at upper left, 2) a handheld game console at left, 3) a small group of anime friends or idols at lower left, 4) a telescope on grass at bottom left, 5) a ferris wheel amusement park at upper right, 6) a small forest cottage scene at far upper right, and 7) a small plate of food floating at mid-right. Make the vignettes ghostly and dim so the apartment remains the focus. No captions, no readable text except indistinct code-like marks on the laptop. Aspect ratio 3:2, highly detailed, moody, atmospheric, no watermark.

Create a polished anime-style BL/romance illustration set in a cozy modern apartment living room at twilight. Two young adult men are the only characters, drawn in detailed semi-realistic Japanese anime style with clean linework, soft cel shading, warm indoor lighting, and a gentle intimate mood. The seated man sits cross-legged on a beige shag rug in the foreground, wearing dark navy satin pajamas with light piping and buttons, holding an open book on his lap with one hand resting on the pages and the other hand raised slightly to the right in a delicate gesture. He has {argument name="front character hair color" default="light ash blond"} tousled layered hair and a slim build. Behind him, the second man kneels close and wraps both arms around him in an affectionate back hug, wearing a loose light gray sweatshirt and dark pants; he has {argument name="rear character hair color" default="black"} messy tousled hair and a slightly broader build. Their faces are intentionally obscured by two simple vertical rectangular blur blocks in skin-toned gradient colors, one centered over each face, preserving the rest of the illustration clearly. The embrace should be close and protective, with four visible embracing hands/arms: the rear man’s left arm across the front man’s chest, rear man’s right arm around the waist, the front man’s left hand on the open book, and the front man’s right hand raised near shoulder height. The room contains exactly five notable background furnishing groups: a low wooden coffee table on the left with a white saucer and glass cup of amber tea, a small stack of three books on the table, a dining table with two chairs near the back window, a tall open shelving unit on the right with boxes and a small potted plant, and large floor-to-ceiling windows showing a blue-purple evening sky, balcony railing, and distant greenery. Use warm beige walls, wooden flooring, a soft rug, realistic perspective, cinematic composition, and no visible text except unreadable book page texture. Avoid extra people, avoid explicit nudity, avoid logos or watermarks.

Goal: Create a dense Japanese tech infographic about {argument name="main topic" default="CLAUDE OPUS 4.8"} and market reaction, styled as a GPT-image-2 generated editorial slide.

Canvas: Portrait 2:3 infographic, off-white graph-paper background with faint grid lines, orange and black headline branding, rounded rectangular cards, thin gray dividers, orange section headers, small flat icons, and compact Japanese typography. Use a clean business-tech layout with a slightly hand-assembled newsletter feel.

Top header: Large black title “CLAUDE OPUS 4.8” with an orange radial starburst logo at far left. To the right, an orange Japanese headline: {argument name="headline text" default="最新情報・市場反応マップ"}. Under it, a smaller Japanese subtitle: “公式発表＋Xサーチ統合：派手なAGI宣言ではなく、長時間エージェント実務の信頼性アップに市場が反応”.

Overall layout: Use exactly 7 numbered main sections, each with an orange circular number badge and orange title bar. Arrange sections 1–3 across the upper half, sections 4–6 in the middle, and section 7 plus a workflow strip across the bottom. Keep all content packed but readable.

Section 1, left column, “発表の中身 — Opus 4.7からの実務寄りアップグレード”: Show exactly 6 stacked spec rows with square black line icons on the left and labels/values on the right: 1) API ID, value “claude-opus-4-8”; 2) Bedrock ID, value “anthropic.claude-opus-4-8”; 3) 通常価格, orange price pills “$5 / $25 per MTok” and note “Opus 4.7から据え置き”; 4) Fast mode, orange “2.5× speed” and price pills “$10 / $50 per MTok”, note “従来Fast比で3分の1”; 5) Context window, orange pill “1M tokens”; 6) Max output, orange pill “128k tokens”; add a bottom small row “Reliable knowledge cutoff / training data cutoff” with orange date {argument name="training cutoff date" default="Jan 2026"}.

Section 2, top center, “公式ベンチマーク比較”: Create a comparison table with exactly 4 model columns labeled “Opus 4.8”, “Opus 4.7”, “GPT-5.5”, and “Gemini 3.1 Pro”. Use orange for Opus 4.8, gray for Opus 4.7, blue for GPT-5.5, green for Gemini. Include exactly 7 benchmark rows with small numbered circles, metric names, percentages/scores, horizontal mini bars, and small crown icons where Opus 4.8 leads: 1) Agentic coding / SWE-Bench Pro: 69.2%, 64.3%, 58.6%, 54.2%; 2) Agentic terminal coding / Terminal-Bench 0.1: 74.6%, 66.1%, 78.2%, 70.3%; 3) Multidisciplinary reasoning / Humanity’s Last Exam no tools: 49.8%, 46.9%, 41.4%, 44.4%; 4) Multidisciplinary reasoning / Humanity’s Last Exam with tools: 57.9%, 54.7%, 52.2%, 51.4%; 5) Agentic computer use / OSWorld-Verified: 83.4%, 82.8%, 78.7%, 76.2%; 6) Knowledge work / GDPval-AA: 1890, 1753, 1769, 1314; 7) Agentic financial analysis / Finance Agent v2: 53.9%, 51.5%, 51.8%, 43.0%. Add a small Japanese footnote beneath the table.

Section 3, top right, “Xで刺さった価値”: Show exactly 4 stacked insight cards with large icons and quotation marks: 1) orange target icon, title “判断の鋭さ”, note about catching dangerous blind spots and noticing your own mistakes; 2) green clock/network icon, title “長時間エージェント”, note “Claude Code dynamic workflows / サブエージェント / 大規模移行”; 3) green enterprise-building icon, title “企業利用”, note listing “API / Bedrock / Vertex / Foundry / 日本リージョン文脈”; 4) orange price-tag icon, title “価格据え置き”, note “通常価格据え置き + Fast改善”.

Section 4, middle left, “市場反応 — 3つの温度帯”: Create exactly 3 sentiment bands plus one summary band: 1) green “ポジティブ（グリーン）” with smiling face icon and three bullets about winning back trust in Anthropic, strong real benchmarks, and self-correction/reliability; 2) yellow “プラクティカル（イエロー）” with neutral face icon and three bullets about Terminal being strong for GPT-5.5, usage model selection, and Bedrock availability; 3) red “スケプティカル（レッド）” with sad face icon and three bullets about quota burn, heavy token burn, and suspicion of 4.7/4.6 comparison; 4) white summary row with tiny bar-chart icon: “全体はポジティブ寄りだが、初動不満も明確。”

Section 5, middle center, “競争地図”: Draw a quadrant map with vertical axis “高性能・高信頼性” at top to “低コスト・高速” at bottom, horizontal axis “汎用／会話寄り” left to “実務・長時間タスク寄り” right. Include exactly 3 model cards: a blue “GPT-5.5” card on the upper-left/center labeled “terminal codingで強い”; an orange “Claude Opus 4.8” card in the upper-right with a brain icon and text “長期エージェント / SWE / OS操作 / 金融文書”; a green “Gemini 3.1 Pro” card in the lower-left with Google-style G icon and text “Google統合・低コスト / 速度軸で別評価”. Add a dashed circular guide and a bottom orange trophy ribbon reading “最強1択ではなく用途別使い分け”.

Section 6, middle right, “Anthropic 企業価値ストーリー”: Split into exactly 2 cards. Top pink card titled “市場会話（X上の拡散）” with people icon, growth arrow, and three bold funding lines: “$65B 調達”, “$965B post-money”, “$47B revenue run-rate”; add small caveat text that X treats these as market conversation and they are not official announcement figures. Bottom green card titled “公式一次情報（Anthropic / 公式発表）” with exactly 2 rows: thumbs-up icon “受け止め 企業向けAI基盤の覇権候補”; question icon “懐疑 高すぎるvaluationを正当化するリリースでは?”.

Section 7, bottom left, “よっしー向け実務示唆”: Make exactly 4 rows with colored icons: 1) green check “使うなら” followed by tasks such as long-duration, many-step, self-verification, Claude Code, migration, and refactoring; 2) yellow warning “避けるなら” followed by low-latency single-shot, simple searches, and high-cost large-volume processing; 3) red exclamation “やらないリスク” followed by missing early signals in engineering and product domains; 4) blue clipboard “導入方針” followed by comparing shadow mode with Opus 4.7/4.8 and logging failures, correction count, token burn, and completion rate.

Bottom workflow strip, to the right of section 7: Title it “推奨ワークフロー（Shadow比較で学習・最適化）”. Show exactly 5 process steps connected by arrows: 1) task-definition clipboard icon “タスク定義”; 2) comparison block with two pills “Opus 4.7” and orange “Opus 4.8”, label “実行（並走）”; 3) magnifier/chart icon “評価・計測”; 4) scales icon “比較・分析”; 5) target icon “改善・定着”. Beneath, add a metrics ledger row with exactly 5 chips/icons: “失敗率”, “修正回数”, “token burn（コスト）”, “完了率”, “所要時間”.

Footer: Add tiny source and confidence text along the bottom: “Source: Anthropic announcement/docs + X Search, observed 2026-05-29 JST.” and “Confidence: official specs high / X sentiment medium / funding numbers medium.” Include a small shield icon at bottom right.

Visual constraints: Use mostly orange, black, gray, green, blue, yellow, and red accents; include many small icons but no photorealism; keep the page crisp, flat, legible, and information-dense; no watermark; no extra sections beyond the 7 numbered sections and the workflow strip.

Create a surreal cinematic landscape showing an enormous explosive fountain of glossy liquid paint or molten cream erupting from the center of a barren desert plain. The eruption is made of thick white fluid streaked and marbled with vivid orange, forming branching splash tendrils that shoot upward and outward like a frozen tree of liquid. Surround it with dozens of floating glossy spheres in different sizes, all suspended in midair; the spheres are white ceramic or pearl-like with orange marbling, plus some smaller solid orange droplets, connected here and there by thin strands and splashes of liquid. Include exactly three dominant foreground spheres: one large marbled white-and-orange sphere at the lower left edge, one large marbled white-and-orange sphere hovering at the upper right, and one large orange-and-white sphere resting near the lower right edge. The background is a vast rocky desert with distant low mountains, scattered stones, wet reflective streaks on the ground, and a dramatic cloudy sky lit by warm golden sunlight. Use a wide-angle perspective, ultra-detailed photorealistic 3D rendering, high dynamic range lighting, sharp focus, turbulent motion frozen in time, glossy reflections, volumetric atmosphere, epic scale, no people, no text, no watermark.

Goal: Create a vintage hand-drawn Japanese infographic poster titled {argument name="headline text" default="AIイラスト用手描き風／水彩・スケッチ系画風Tier表３"}, presenting a tier/list of AI illustration prompt styles for hand-drawn, watercolor, and sketch aesthetics.

Canvas: Vertical poster, 4:5 aspect ratio, aged cream parchment paper background with stained edges, subtle grid texture, washi tape scraps, ink dots, small stars, botanical doodles, and scrapbook-like decorative framing. Use navy blue handwritten Japanese title typography with watercolor accents.

Top area: Large title across the upper left. Under it, add a smaller Japanese subtitle: 「フリーハンド・水彩スプラッシュ・柔らかい線画表現に使える画風プロンプト一覧」. Place a decorative memo card on the left with four checkbox bullets about soft lines, blurring, mixed media, and warmth. Across the center top, show exactly 4 taped sample swatches: 1 graphite scribble sketch sheet, 1 blue watercolor blot sheet, 1 pink-purple watercolor blot sheet, and 1 black urban line sketch sheet. On the upper right, show a seated young woman artist with long dark hair wearing a patterned shawl, surrounded by brushes, ink bottles, a watercolor palette, and a sketchbook; include a small speech bubble saying 「画材感で使い分けてね。」 and a small note reading “Create Inspire Explore”.

Main layout: A clean 3-column by 4-row grid of exactly 12 rounded rectangular information cards. Each card has a colored circular number badge from 1 to 12 at the top left, a tiny category line reading “C. 手描き風 / 水彩 / スケッチ (Hand-drawn / Watercolor / Sketch)”, an English style name, Japanese style name, short Japanese explanation, “Sample Prompt:” in English, and a small hand-drawn sample illustration. Use dark navy text, warm beige card backgrounds, thin blue-gray borders, and delicate watercolor shadows.

The 12 cards must be:
1. “English Style: Freehand Line Art” / 「日本語名：フリーハンドラインアート」 with a loose ink drawing of a brainstorming session around a table.
2. “English Style: Watercolor Splash” / 「日本語名：水彩スプラッシュ」 with a bright watercolor splash surrounding a glowing lightbulb icon.
3. “English Style: Soft Pencil Sketch” / 「日本語名：ソフト鉛筆スケッチ」 with a soft pencil desk scene with plants near a window.
4. “English Style: Mixed Media Sketch” / 「日本語名：混合技法スケッチ」 with a creative designer’s desk combining layered textures, paper, paint, and pen.
5. “English Style: Handmade Sketch Illustration” / 「日本語名：手作りスケッチイラスト」 with four people shaking hands in a warm handmade business illustration.
6. “English Style: Gentle Watercolor” / 「日本語名：優しい水彩」 with a teacher helping students at a classroom table in gentle colors.
7. “English Style: Fine Line Drawing” / 「日本語名：ファインラインドローイング」 with a luxury wristwatch drawn in intricate fine lines.
8. “English Style: Whimsical Sketch” / 「日本語名：ファンシースケッチ」 with a whimsical flying office desk carrying papers and coffee.
9. “English Style: Expressive Brushstroke” / 「日本語名：表現的な筆使い」 with a runner breaking through a finish line in dynamic ink and color strokes.
10. “English Style: Casual Doodle Sketch” / 「日本語名：カジュアルドゥードルスケッチ」 with casual doodle icons for brainstorming: an IDEA lightbulb, gears, sticky notes, pie chart, and bar chart.
11. “English Style: Loose Colored Pencil Sketch” / 「日本語名：ラフな色鉛筆スケッチ」 with a happy team in a casual meeting, loose colored pencil texture.
12. “English Style: Soft Sketch Watercolor Combo” / 「日本語名：ソフトスケッチ水彩コンボ」 with a cozy home office setup, laptop, shelves, plants, and sketch tools.

Bottom section: Add a horizontal explanation panel. On the left, a dark blue vertical label block says 「解説」 in large white Japanese characters with floral ornamentation. To the right, include five concise Japanese bullet points explaining that the chart helps choose hand-drawn, watercolor, and sketch styles for AI illustration prompts, that English style names can be inserted into prompts, and that brush strength ranges from doodle and colored pencil to expressive brushstrokes and fine line art. On the bottom right, place a still-life of art supplies: stacked books labeled “DESIGN”, “PROMPT”, and “AI NOTE”, a watercolor palette, brushes in a cup, ink bottles labeled “INK”, pencils, pens, and a small note saying 「目的に合わせて選んでね。」.

Visual style: Handwritten infographic, watercolor sketch, analog stationery aesthetic, soft ink outlines, lightly desaturated beige and navy palette with pastel watercolor accents, cozy art-journal mood, Japanese magazine layout, high detail but readable.

Constraints: Use exactly 12 numbered style cards, exactly 4 top sample swatches, and the bottom explanation area. Keep all text legible and arranged like a polished educational poster. Do not add logos, watermarks, QR codes, or extra cards.

Goal: Create a vintage travel-poster infographic for {argument name="destination" default="Umeda, Osaka, Japan"}, combining a simple skyline illustration, large decorative typography, an isometric landmark map, and travel-ad copy.

Canvas: Vertical poster, about 2:3 aspect ratio, warm ivory paper background with a thin double-line ornamental gold border and clipped art-deco corner details. Use clean vector line art, muted pastel colors, subtle grain, beige shadows, and a retro Japanese tourism poster feel.

Layout: Top section is a wide rectangular sunset skyline panel with a turquoise-to-pink-to-orange gradient sky, flat beige city silhouette, the Umeda Sky Building centered, and a two-car train on tracks at the lower right. Middle section has one huge word, {argument name="large title word" default="UMEDA"}, in bold outlined block letters filled with miniature city scenes; add a tan offset shadow and small handwritten Japanese note arrows near the right side. Main section is an isometric illustrated area map of Umeda with labeled landmarks and connecting leader lines. Bottom section contains location/date text, a large Japanese slogan, a bold English slogan, access information, railway line legend, must-see list, URL/hashtags, coordinates, and a small circular map inset.

Top skyline details: Include exactly 3 main skyline elements: 1 tall Umeda Sky Building with two linked towers and a bridge-like top, 1 simplified beige urban skyline stretching across the horizon, and 1 modern train made of exactly 2 connected cars.

Large typography details: The word UMEDA must contain exactly 5 giant letters, each filled differently: U with the Umeda Sky Building, M with modern buildings and greenery, E with a red ferris wheel, D with tall office towers, and A with urban street/building fragments. Use thick dark brown outlines, warm beige inner outlines, and colorful fills.

Isometric map landmark count: Show and label exactly 10 landmarks around the central map: 1 Umeda Sky Building, 2 Grand Green Osaka, 3 Grand Green Osaka (Umekita Park), 4 HEP FIVE Ferris Wheel, 5 LUCUA, 6 Grand Front Osaka, 7 Yodobashi Umeda, 8 Hankyu Department Store, 9 Hanshin Department Store, 10 Osaka Station. Also include exactly 1 additional label, Whity Umeda, near the lower left walkway area. The map should include towers, shopping complexes, station buildings, plazas, a green park with trees, curved pedestrian paths, and a red ferris wheel as the most saturated object.

Bottom text content: Use the heading “OSAKA, JAPAN | UMEDA 2026”. Below it, place the Japanese slogan {argument name="Japanese slogan" default="そうだ、梅田行こう。"} in large bold black type. Under that, write “UMEDA — LET’S GO!” in very large bold uppercase letters. Add the subtitle “VIBRANT STREETS. INCREDIBLE FOOD. ENDLESS MODERN ENERGY.” Add access text: “Access: 関西国際空港 → JRはるか（最速約45分）”. Include a rail legend with exactly 5 colored line items: JR Line, Hankyu, Hanshin, T, and Y. Add the must-see line listing: “Must-See: Umeda Sky Building · HEP FIVE Ferris Wheel · Grand Green Osaka · Yodobashi Umeda · Grand Front Osaka · Osaka Station · Hankyu Department Store · Hanshin Department Store · LUCUA · Whity Umeda”. Add small footer text “osaka-info.jp    #UmedaOsaka    #VisitOsaka” and coordinates “34.701°N 135.496°E”.

Map inset: Bottom right contains exactly 1 circular street-map inset with pale roads, light green blocks, blue waterways, a red location pin, and a tiny compass marker.

Style constraints: Keep the illustration intentionally simple and clean, with thin black-brown outlines, flat fills, minimal shading, and a hand-drawn travel guide aesthetic. Avoid photorealism, 3D rendering, excessive detail, extra labels, logos, or watermarks. Preserve the visible Japanese text exactly where specified while keeping all other design instructions in English.

Goal: Create a square Japanese social-media comic infographic about trying GPT Image2 for making a campaign poster, with a cheerful office-work atmosphere and cute anime styling.

Canvas: 1:1 square image, bright pastel palette, soft daylight office background, white cloud-shaped panels outlined in blue, pink, purple, and green, with sparkles, arrows, doodles, stickers, and expressive reaction icons. The overall design should feel like a polished explanatory X/Twitter post image.

Main subject: A cute chibi/anime-style office woman with short wavy brown hair, white blouse, small gold accessories, and expressive poses, appearing repeatedly across the design. Keep faces anonymized/blank with a soft rectangular blur over each face. The same character appears exactly 7 times: 1) small at top left using a laptop, 2) upper center admiring a generated poster, 3) upper right giving a thumbs-up, 4) mid-left looking uncertain beside prompt instructions, 5) large right-side rear/side view seated at a desk looking at a monitor, 6) bottom left distressed with hands on head, 7) bottom right tucked in bed saying good night.

Layout: Place six numbered story steps in rounded cloud panels around a central desktop monitor. Use large circular step numbers with white numerals: step 1 in blue at top left, step 2 in blue at top center, step 3 in pink at top right, step 4 in purple at left middle, step 5 in purple at bottom left, and step 6 in green at bottom right. Connect the steps with thick colorful arrows, including a blue arrow from step 1 to step 2, a pink arrow to step 3, a purple downward arrow to step 5, and a dotted green path toward step 6.

Central object: A large computer monitor on a desk showing a design app interface labeled “GPT Image2” at the top. On the canvas inside the monitor is a skincare campaign poster for {argument name="brand name" default="Lumière"}, with Japanese headline text, product bottles, white flowers, green leaves, and gold circular benefit badges. Include a keyboard, mouse, coffee mug, notebook checklist, pen, small potted plant, and a tidy office desk. In the background, show a softly blurred modern office window with greenery.

Text content: Use Japanese text in the artwork. Step 1 headline: “AIで画像生成を試してみよう！” with a question mark and lightbulb doodle. Step 2 headline: “GPT Image2すごい” followed by a surprised emoji, and show a small generated campaign poster preview. Step 3 headline: “感動しました！” with a heart; include a checklist with exactly 4 green check items: “テーマ◎”, “雰囲気◎”, “訴求ポイント◎”, “ロゴも完璧!”. Step 4 headline: “他のAIは少し難しい” with a scribble icon; include a mini card titled “他社AIツール” with exactly 3 stacked instruction boxes: “プロンプトを詳細に入力”, “構成を細かく指定してください”, “ネガティブプロンプトも設定してください…”. Step 5 headline: “ポイントがどんどん減っていく…” with a crying/sweating emoji; include a vertical point meter titled “保有ポイント” listing exactly 5 values: “1000 pt”, “600 pt”, “230 pt”, a downward arrow, and “50 pt” in red. Step 6 headline: “明日またトライします” with a sparkle. Bottom-right bedtime bubble: “おやすみなさい また明日！” with a heart.

Poster details: The skincare ad should read {argument name="campaign headline" default="うるおい続く 私らしい毎日へ。"} and smaller copy “肌も心も、やさしく満たす。” Include exactly 3 gold benefit circles: “高保湿”, “敏感肌処方”, “うるおいキープ”. Show a clean logo mark of leaves above “Lumière / ルミエール”.

Desk details: The open notebook at the bottom center should be labeled “やることリスト” and include exactly 5 checkbox items, with several checked in red; the visible items should include “企画”, “ビジュアル”, “キャッチコピー”, and “予算確認”. Add small decorative point coins flying away near the distressed character.

Style constraints: Cute Japanese manga infographic, clean commercial design, rounded typography, thick white outlines around character cutouts, soft cel shading, high detail but readable at social-post size. Preserve the exact six-step count, the seven character appearances, and the visible Japanese labels. No watermark, no photorealism, no extra panels.

Create a realistic vertical street-fashion photograph of {argument name="subject" default="a young Japanese woman"} standing centered in a narrow urban backstreet in Japan during soft morning daylight. She is shown from about mid-thigh upward, facing the camera, with both arms raised and hands resting on top of her head, giving a playful little-devil smile implied by the pose. Her face is intentionally obscured by a centered soft-edged rectangular blur/mosaic block, while the rest of the image remains sharp and natural. She has long slightly messy {argument name="hair color" default="black hair"} falling over her shoulders and wears a {argument name="outerwear" default="light beige cardigan"} over a fitted {argument name="top" default="white cropped tank top"}, plus high-waisted blue denim jeans with the waist and navel visible. The background is a realistic Japanese alley with low buildings on both sides, a metal shutter storefront and bicycle on the left, stacked crates near the curb, potted plants and a vending machine on the right, overhead utility wires, and shallow depth of field. Use a candid smartphone/editorial photo look, natural skin tones, cool daylight, slightly wide-angle perspective, subtle film grain, no text, no watermark, no extra people.

Goal: Create a realistic 2x2 instructional sports photo collage showing a female golfer completing a dynamic golf swing sequence, photographed from an extreme low ground-level angle on a sunny golf course.

Canvas: Vertical 4:5 image, split into exactly 4 equal panels with thin seams, two columns and two rows. Each panel should look like a high-speed action photograph taken outdoors under a vivid blue sky with scattered white clouds, green fairway grass, and trees near the horizon.

Subject details: The same athletic young woman appears in all 4 panels, wearing a sleeveless {argument name="outfit color" default="white"} golf dress with a pleated skirt, white golf gloves, and white athletic shoes. She has long {argument name="hair color" default="brown"} hair in a high ponytail, blowing dramatically with the motion. Her face should be obscured by a soft square blur for privacy. Use a driver golf club with a dark club head.

Layout and exact panel count: Include exactly 4 numbered action panels with small white Chinese labels in the upper-left corner of each panel: 1. 「站位准备」 showing the golfer at address position, feet apart, club head near the ball on the grass; 2. 「上杆顶点」 showing the top of the backswing, arms raised overhead, body arched, skirt and hair moving; 3. 「击球瞬间」 showing the impact moment, club sweeping through the ball, dirt and grass exploding toward the camera, strong motion blur on the club head; 4. 「收杆定格」 showing the follow-through finish, club swung around behind her, hair flying, grass and soil particles still suspended in the air.

Visual style: Ultra-realistic sports photography, cinematic high-speed shutter mixed with subtle motion blur, dramatic low-angle perspective that makes the golfer appear tall and powerful, bright midday sunlight, crisp shadows, saturated natural colors, energetic dynamic capture, shallow depth of field near the grass.

Constraints: Keep exactly 4 panels and exactly the 4 listed labels, no extra captions, no logos, no watermark, no product-ad layout, no shopping-poster styling.

Goal: Create a clean anime concept art reference sheet for a {argument name="character name" default="Samurai Mecha Girl"}, combining a school-uniform samurai heroine with exposed light-mecha prosthetic limbs, presented as high-resolution monochrome production-design line art with small technical notes.

Canvas: Vertical portrait character sheet, off-white paper background, thin black outer border, 3:4 aspect ratio. Use crisp black and gray pencil/ink linework, minimal shading, no full color except a small color guide. The overall look is a Japanese setting-materials sheet, precise and highly detailed.

Layout: Top-left large English title “Samurai Mecha Girl” with smaller Japanese subtitle “侍機巧娘” below it. The left and center two-thirds contain the main full-body character illustration. The right third is a boxed head-turnaround panel. The bottom area contains weapon and mechanical-part diagrams. Keep the page airy, technical, and neatly ruled with thin divider lines.

Main subject: A 17-year-old anime girl with {argument name="hair style" default="short straight bob hair with blunt bangs"}, serious eyes, and circular mechanical ear/head sensor modules. She wears a sailor-style school uniform with scarf, short pleated skirt, thigh-high stockings, and mechanical armor integrated into one arm and one leg. Pose her kneeling dynamically: one knee down, one cybernetic leg raised, torso angled forward, one mechanical hand braced near the ground, and the other hand holding a katana horizontally above her head. The katana has a wrapped handle and a long slightly curved blade. Her mecha limbs are detailed with pistons, circular joints, armor plates, cables, segmented fingers, and heavy boot components.

Visible text and labeled sections: Include exactly 7 major text/detail blocks: 1) top-left title block with “Samurai Mecha Girl” and “侍機巧娘”; 2) a stats box listing “TYPE : ASSAULT”, “CLASS : LIGHT MECHA”, “HEIGHT : 158cm”, “WEIGHT : 49kg”, “AGE : 17”, “AFFILIATION : UNKNOWN”; 3) a “CONCEPT” block with short Japanese paragraph lines; 4) a “SPEC” block with Japanese bullet points; 5) a bottom weapon section labeled “武装” and “ENERGY KATANA / エネルギー式刀剣”; 6) a bottom mechanical parts section labeled “ARM UNIT / 腕部ユニット” and “LEG UNIT / 脚部ユニット”; 7) a right-side lower notes area with “COLOR GUIDE（参考）” and “DETAIL”.

Right-side head turnaround panel: Add one bordered panel titled “頭部ラフスケッチ三面図”. Inside it, show exactly 3 head drawings stacked vertically and labeled “FRONT”, “SIDE”, and “BACK”: front view with face and headband, side profile showing the circular ear mechanism, and back view showing the bob haircut and rear of the mechanical module.

Bottom diagrams: In the weapon box, show exactly 1 horizontal energy katana diagram with a wrapped grip. In the mechanical parts box, show exactly 5 separate component drawings: two arm-unit views on the left and three leg-unit views on the right, each rendered as detailed mechanical line-art studies.

Color guide: Include exactly 6 small square swatches with labels: main armor dark iron, sub armor white gray, accent red, mechanical joints gold, cloth dark blue, hair brown, plus a small orange eye/accent swatch if needed; keep these as the only colored elements on the page.

Style constraints: Use refined monochrome anime drafting, thin clean outlines, technical callout-sheet composition, high-detail mechanical design, subtle paper texture, sharp readable typography, no background scenery, no dramatic lighting, no painterly rendering, no extra characters.

Create a cinematic anime-style night street portrait of a young adult male urban character in a neon-lit Japanese city district. The subject is centered slightly right in a medium shot from waist to head, leaning toward the camera with a confident streetwear attitude, one hand in his jacket pocket and the other extended toward the viewer in a finger-gun gesture. Give him messy layered {argument name="hair color" default="dark brown"} hair with detailed strands, visible ear piercings, and a deliberately obscured face covered by a plain square blur/censor block in a soft skin-tone gradient. Dress him in an all-black layered outfit: an oversized high-collar technical jacket with zippers, straps, snaps, sleeve patches, and glossy folds; a black hoodie underneath; a graphic dark T-shirt; and multiple silver chain necklaces. The background is a rainy, bustling Tokyo-like nightlife street with wet reflective pavement, metal guardrails along the sidewalk, blurred pedestrians, storefronts, streetlights, and many vertical glowing signs with indistinct Japanese-style lettering in cyan, pink, orange, yellow, and white. Use shallow depth of field with the background softly blurred, strong neon rim lighting on the character, high-contrast cel-shaded anime rendering, crisp line art, rich highlights, and a moody cyberpunk color palette. Horizontal 16:9 composition, no readable brand names, no watermark, keep the face censor block perfectly square and centered over the facial features.

Create a cinematic anime close-up of {argument name="character name" default="a smug young man"} leaning toward the viewer in a dark industrial setting. He has messy, tousled {argument name="hair color" default="black hair"} falling over his forehead and partially covering one eye, sharp narrow gray eyes, tan skin, strong angular facial features, and a confident teasing smirk. His right hand is thrust forward toward the camera in extreme foreground, palm angled upward with fingers curled in a beckoning gesture, heavily foreshortened and slightly out of focus to emphasize depth. Frame the shot from below at a slight upward angle, with the face occupying the left-center and the hand dominating the right foreground. Use dramatic hard lighting with deep shadows across half of his face, a bright highlight on the cheek and nose, and muted cool tones. The background should be blurred gray metal or concrete panels with a gritty, worn texture, suggesting an alley, warehouse, or underground facility. Style: high-quality modern anime key visual, crisp linework, expressive eyes, cinematic shading, shallow depth of field, widescreen 16:9 composition, intense moody atmosphere. No text, no watermark, no extra characters.

Create a hyper-realistic cinematic DSLR street-fashion portrait of a {argument name="character description" default="stylish Pakistani girl"} standing full-body against a tall rough concrete urban wall covered with black graffiti splatter and abstract bird silhouettes, in a rainy city alley with wet reflective pavement and a softly blurred modern street background on the left. She faces the camera in a relaxed fashion pose, one leg crossed in front of the other, long straight brown hair over her shoulders, wearing a luxury western-fusion outfit without dupatta: a knee-length grey and black printed kurti with black embroidered floral patterns, black neckline and sleeve panels, patterned border hem, slim white trousers, chunky white sneakers with black accents, and a silver wristwatch. Add exactly 5 birds around her: 2 large blue-and-yellow macaws/parrots with wings fully spread, one flying in from the left and one from the right; 2 white gull-like birds near her lower left and lower right; and 1 tiny colorful hummingbird above her head. Make the parrots appear dynamic and sharp, with dramatic shadows cast onto the concrete wall. Use moody overcast lighting, high detail fabric texture, realistic skin and hair, shallow depth of field, cinematic contrast, 8K editorial photography quality, vertical 9:16 composition. Add a small elegant handwritten signature watermark at bottom right reading {argument name="watermark text" default="Created by Mr. Tariq"}.

Create a clean printable sheet of nine identical ID photo portraits arranged in a 3 by 3 grid on a white background. Each portrait shows the same adult woman from the chest up, centered and facing forward, with long light brown/blonde hair worn down, wearing a dark navy business blazer over a white V-neck blouse. Use neutral passport-photo lighting, a plain white studio backdrop, symmetrical composition, realistic photographic style, and consistent scale across all copies. The face must be anonymized/covered in every copy by a smooth vertical rectangular blur block in a natural skin-tone color, placed over the full face area from forehead to chin while leaving hair, neck, ears, blazer, and blouse visible. The sheet contains exactly 9 portraits: top-left, top-center, top-right, middle-left, middle-center, middle-right, bottom-left, bottom-center, and bottom-right. Keep generous white gutters between portraits, crop each photo just below the shoulders, no text, no borders, no watermark, no decorative elements. Use {argument name="subject gender" default="adult woman"}, {argument name="hair color" default="light brown blonde"}, {argument name="outfit" default="dark navy blazer and white V-neck blouse"}, {argument name="grid count" default="3 by 3, nine portraits"}, and {argument name="face anonymization" default="smooth skin-tone rectangular blur over the face"}.

Create a cinematic luxury perfume advertisement still in a warm, dark, candlelit interior. Show a glamorous {argument name="subject" default="woman"} in close-up from the chest up, positioned slightly left of center, with her face intentionally obscured by a large solid medium-brown rectangular block covering the central upper face area. She has dark wavy hair pinned loosely with strands falling around her neck, wears long sparkling gold chandelier earrings, and a shimmering metallic gold evening dress with pleated sequins and a low neckline. Her right hand is raised near the right side of the frame holding an elegant faceted rectangular glass perfume bottle filled with amber liquid, with a geometric gold cap catching the light. Use {argument name="lighting style" default="golden candlelight"}, shallow depth of field, soft bokeh flames in the background, dark reflective walls, and a sensual high-fashion editorial mood. Add a small white “Pollo.ai” watermark in the top-right corner. Horizontal 16:9 composition, photorealistic, cinematic color grading, rich amber and bronze tones, no extra text.

Create a realistic vertical smartphone selfie photo of a stylish young adult woman seated on a white tufted leather sofa in an upscale café or lounge. She has long voluminous wavy {argument name="hair color" default="blonde"} hair swept dramatically to one side, with loose strands and salon-like highlights. Her face is intentionally obscured by a centered matte square censor block in a warm tan skin-tone color, covering the entire facial area from forehead to chin. She wears a fitted black ribbed short-sleeve crop top with a bold white ornamental heart graphic on the chest and the visible text {argument name="shirt text" default="THE TOTAL"} arched above it, plus light blue high-waisted denim jeans. Add exactly 3 visible accessories: a black lace choker, a thin silver necklace with a small rectangular pendant, and small heart-shaped stud earrings. Pose her at arm’s length as if taking a selfie, with one arm extended toward the camera, torso angled slightly, confident casual posture. The background should be softly blurred with warm bokeh chandeliers, dark cabinetry, framed wall décor, mirrors or glossy black trim, a bar/counter area, and creamy white seating, giving a refined boutique café atmosphere. Use natural warm indoor lighting, shallow depth of field, high-detail photorealism, crisp fabric texture, glossy hair detail, and a modern social-media portrait aesthetic. No extra people, no watermark, no additional text besides the shirt graphic.

Create a realistic vertical smartphone selfie photo of a young adult woman with {argument name="hair color" default="long tousled blonde hair"} sitting on a white couch in an upscale modern interior. Her arm extends toward the camera as if holding the phone, creating a close, slightly wide-angle selfie perspective from chest to head. A large opaque square censor block in a warm tan skin-tone color completely covers the central face area from forehead to chin, leaving the hair, ear, neck, shoulders, and outfit visible. She wears a black ribbed cropped tank top with an ornate symmetrical white floral-vine graphic across the chest, a black lace choker, a delicate silver chain necklace with a small spiral pendant, and a small heart-shaped earring. The background shows glossy black cabinetry and mirrors, a warm crystal chandelier, a vase of white flowers, and a black-and-white striped cushion on the couch. Use warm indoor lighting, realistic skin texture, casual glamorous lifestyle photography, natural messy hair strands, shallow depth of field, and an unpolished social-media selfie composition. No text, no watermark.