GPT Image 2 prompts

Create an image: A page of a scientific and medical dossier in a {argument name="style" default="biopunk horror manga style"}, with a detailed and complex anime-style technical interface design. In the center, the attached model is dressed as a {argument name="role" default="scientist"}. Strict trichromatic color palette: only white, black, and bright blood red. Inked manga art style, shaded with dot patterns (screentones), with detailed hand-drawn line drawing textures. The background is a worn technical document with a grunge aesthetic, filled with English warning texts and Japanese kanji in blocks, barcodes, biological data boxes, plant diagrams, and a red seal. Old paper texture, photocopy grain, high contrast, dark and suspenseful atmosphere—a masterpiece of conceptual graphic art.

Use the uploaded product image as the primary reference. Preserve the original product exactly as shown, including its packaging design, logo placement, typography, branding elements, colors, proportions, materials, textures, reflections, shadows, product shape, food appearance, beverage appearance, ingredients, label details, and overall brand identity. DO NOT redesign, simplify, reinterpret, replace, crop, distort, recolor, or modify any packaging, logos, branding, structural details, or label elements. The product must remain 100% photorealistic and completely unchanged. Transform the product into the hero of a premium commercial advertising campaign while keeping the product itself entirely realistic. Before generating the scene, intelligently analyze the product's category, purpose, flavor, ingredients, packaging style, brand personality, color palette, and emotional appeal. Based on that analysis, automatically develop a fun, playful, visually engaging concept that feels naturally connected to the product. Choose an activity that authentically reinforces the product's identity, usage occasion, lifestyle, season, flavor, or personality—never something random or disconnected. Automatically generate matching doodle props that support the chosen activity and strengthen the visual storytelling. Select a clean {argument name="background" default="pastel studio background"} inspired by the dominant colors of the product packaging and branding, ensuring strong visual harmony while maintaining enough contrast for the product to remain the clear focal point. Position the product in a dynamic tilted pose that conveys movement, energy, and personality. Surround the product with playful hand-drawn doodle illustrations created using {argument name="doodle style" default="loose black marker lines"}. Illustrate ONLY the following as doodles: Arms, Hands, Legs, Shoes, Facial expression (only if appropriate), Activity props, Motion lines, Speed lines, Swirls, Impact effects, Decorative doodles. Keep the product itself completely photorealistic. All doodles must share the same hand-drawn black marker style with consistent line weight, slightly imperfect strokes, expressive energy, and a fun mascot-like personality. Generate additional decorative doodles that naturally complement the chosen concept, such as leaves, stars, bubbles, sparkles, confetti, splashes, crumbs, petals, clouds, geometric shapes, or other relevant elements. Add realistic, soft floating shadows beneath the product to create convincing depth. Use premium commercial product photography lighting with soft diffused studio illumination, realistic reflections, subtle shadows, shallow depth of field, premium retouching, and luxury advertising quality. Leave generous negative space suitable for marketing headlines, branding, or promotional copy.

A technical drawing of a {argument name="subject" default="futuristic sports car"} in blueprint style. Include line drawings of the sports car from the front, side, and rear views, exploded parts sketches, parts assembly diagrams, and structural diagrams of disassembled components. Use abundant lines and measurement values to indicate the dimensions of each part, with grayscale tones expressing the overall sketch relationship. In addition to the main design, also show scattered thumbnails from different angles.

Restore this {argument name="item" default="photo"}. Reduce and correct the overexposed areas, recover highlight and shadow details, sharpen the {argument name="subject" default="faces"} naturally, balance the colors and skin tones, enhance overall clarity, and restore fine details without making the image look artificial or overprocessed.

以参考人物为唯一人像身份，保持五官、脸型、眼神和长黑发高度一致。生成一组 9:16 竖版写实人像写真，场景为安静的复古室内空间，深色木地板、灰米色墙面、藤编木椅、米色针织毯和一本旧书。

Create a high-end studio portrait of a smiling {argument name="subject description" default="young adult man with short dark brown hair, blue eyes, light skin, and neatly trimmed stubble"} facing straight toward the camera, centered and cropped from the upper chest upward. He wears a fitted {argument name="shirt color and style" default="premium crimson red crew-neck t-shirt"}. Split the entire image into exactly 2 vertical halves with a razor-sharp thin white divider line running perfectly from the top edge to the bottom edge through the center of his face and torso: the left half has a solid {argument name="left background color" default="deep crimson red"} studio backdrop, and the right half has a solid {argument name="right background color" default="clean white"} studio backdrop. Keep the subject symmetrical and aligned with the divider, with one half of his face on each side, a natural friendly smile, crisp facial detail, realistic skin texture, polished commercial lighting, soft shadows, sharp focus, and premium editorial retouching. Use a vertical 3:4 composition, minimal background, no text, no logos, no props, no border other than the central divider.

Goal: Create a square side-by-side comparison image showing two dark fantasy generations of the same concept: a white-clad gothic nun holding a black bass guitar inside a ruined cathedral, with a black title bar across the top.

Canvas: Square 1:1 image, split into exactly 2 equal vertical panels separated by a thin white divider line. Add a black header band across the full width at the top.

Top text: Place exactly 2 white sans-serif labels in the header band: left label “{argument name="left model label" default="GPT Image 2"}” centered over the left panel, right label “{argument name="right model label" default="Seedream 5.0 pro"}” centered over the right panel. Use large clean white text, no other typography.

Left panel: A clean cinematic hero shot. Center exactly 1 young pale gothic nun with long platinum-blonde hair under a white hooded veil, stern expression, wearing layered white robes with a dark belt and subtle armor-like shoulder pieces. She stands full body, facing forward, holding exactly 1 glossy black electric bass guitar diagonally across her waist. Behind her is exactly 1 enormous dark stone cross suspended or mounted high in the ruined cathedral. The cathedral is narrow, gray, decayed, filled with vines, broken arches, dust, mist, and cold shafts of light from tall windows. Around her are exactly 6 visible crouching alien/demon silhouettes: 2 near the left foreground/midground, 2 near the right midground, and 2 partially visible in the lower foreground edges. Mood is restrained, flat, scanned-film, monochrome gray, clean composition, strong centered symmetry.

Right panel: A more dramatic action-horror version of the same scene. Center exactly 1 young pale gothic nun with long platinum-blonde hair under a white hooded veil, wearing white robes and a dark belt, holding exactly 1 black electric bass guitar across her waist. Behind her is exactly 1 enormous dark stone cross, backlit by intense god rays pouring through broken cathedral windows. The ruined cathedral is wider, deeper, more chaotic, with twisted organic vines, arched walls, rubble, smoky haze, and high-contrast teal-gray lighting. Include exactly 14 visible demon/alien figures: 5 on the left side, 5 on the right side, 2 in the center background, and 2 partial foreground creatures at the bottom corners. They should appear as a menacing swarm approaching the nun. Mood is dread-filled, dynamic, cinematic, with stronger rim light and volumetric beams.

Visual style: Dark fantasy gothic realism, ultra-detailed but not glossy, cold desaturated palette of black, charcoal, bone white, silver gray, and faint blue-green. The nun’s white outfit must remain the brightest subject in both panels. Use dramatic cathedral scale, wet stone floor texture, dust particles, and horror atmosphere.

Constraints: Keep the image as a comparison diptych, not two separate files. Include exactly 2 panels, exactly 2 top labels, exactly 2 nuns total, exactly 2 bass guitars total, and exactly 2 large crosses total. No watermark, no extra captions, no logos, no additional text.

Using the provided reference image, transform the person on the motorcycle into a Grand Theft Auto VI / Vice City inspired illustrated poster. Preserve the rider’s recognizable identity, pose, sunglasses, white jacket with the large “P” patch, jeans, shoes, and the motorcycle as the main subject, but stylize everything as polished game key art with bold outlines, glossy highlights, saturated neon colors, and dramatic cel-shaded rendering.

Change the setting from the forest roadside to a tropical coastal city at sunset: add a wide boulevard, palm trees, pastel art-deco buildings, ocean water, a distant skyline, and a warm pink-orange-to-blue gradient sky with soft clouds. Make the motorcycle look slightly more heroic and pristine with reflective chrome and teal/pink neon rim lighting. Use a low-angle poster composition with the rider and bike large in the foreground, facing slightly right.

Add exactly 2 visible text/logo elements: 1) large block title text in the bottom-left reading {argument name="location title" default="LEONIDA STATE"}; 2) a small semi-transparent badge/logo in the top-right reading {argument name="studio badge text" default="FICTIONAL LIFESTYLE STUDIO"}. Keep the overall mood cinematic, tropical, stylish, and like official open-world crime game promotional artwork. No photorealism, no extra people, no clutter.

Create an ultra-realistic cinematic editorial photograph titled “Deep Work in Nature” featuring {argument name="subject description" default="a focused Black male professional in his early 30s"} working alone in the middle of a vast lush green meadow. The subject sits on one simple black folding camping chair, positioned in the lower center of the frame, wearing {argument name="outfit" default="a tailored navy blue business suit, white dress shirt, dark tie, black shoes, and eyeglasses"}. He has over-ear {argument name="headphones color" default="silver and white"} headphones on and is quietly writing in one open notebook or planner held across his lap with one pen, looking downward in deep concentration. Surround him with tall wet grass in the foreground, rolling meadow hills in the midground, and a soft blurred forested valley in the background, with one pale winding dirt path disappearing into the distance. Include one large leafy tree canopy entering from the upper right, its branches framing the scene. Use moody natural overcast light, rich greens, shallow depth of field, cinematic compression from a telephoto lens, realistic texture, subtle bokeh, soft background blur, and a calm contemplative editorial mood. Vertical 4:5 composition, eye-level to slightly elevated perspective, subject small against the expansive landscape, no text, no logos, no extra people, no buildings, no devices other than the notebook and headphones.

Create a square, close-up editorial portrait of {argument name="subject" default="a young East Asian man"} in a dreamy high-fashion style, with his head tilted and resting on a bent arm near the bottom of the frame. Use extreme soft focus, long-exposure motion blur, and double/triple exposure ghosting so the face appears repeated and smeared diagonally across the image, with one clearer eye near the center-left and another translucent face shifted toward the upper right. The expression is calm, distant, and introspective, lips relaxed, gaze slightly off-camera. Show dark tousled hair blending into the blur, pale skin, a white shirt at the lower right, and a small silver chain bracelet on the wrist near the lower center. Lighting should be ethereal and overexposed, with a cool misty background in pale blue and white. Add translucent color washes of {argument name="accent colors" default="cyan, teal, deep blue, and magenta-pink"} across the hair and face, like prism flare or light leaks. Make the image feel analog, romantic, hazy, and painterly, with no sharp edges except faint facial features; use shallow depth of field, airy glow, and smooth pastel gradients. No text, no logos, no hard outlines, no realistic studio crispness.

Goal: Create a clean futuristic Chinese tech product matrix slide for {argument name="brand name" default="Z-Seven"}, showing the team’s core AI-native product lineup.

Canvas: 16:9 horizontal presentation slide, white to very pale blue background, subtle circuit-board traces, tiny dotted tech patterns, thin connector lines, soft shadows, glossy SaaS/AI startup aesthetic.

Header: Large centered headline in dark navy and bright blue: {argument name="headline text" default="Z-Seven 团队核心产品矩阵"}. Under it, a smaller gray-blue subtitle: {argument name="subtitle text" default="围绕 AI 原生开发体验，构建设计、记忆、框架与终端工具链"}.

Layout: Arrange exactly 4 rounded rectangular product cards in a 2×2 grid around a central circular hub. The cards are large white panels with subtle shadows, soft borders, and colored accent strokes on the left edge or corner. Connect all four cards to the center with faint blue connector lines and small node dots. In the center, place a circular badge containing a stylized blue Z logo and the text “ZSEVEN”.

Product cards: Include exactly 4 cards, each with one logo, one product name, one two-line description, and exactly 3 pill-shaped tags.
1. Top-left card: Product name “OpenPencil” in bright cyan-blue. Large angular blue pencil/Z style logo on the left. Description: “AI 原生矢量设计工具，\n让 Prompt 直接连接实时画布。” Tags, exactly 3: “Design-as-Code”, “开发 Agent Teams”, “实时画布”.
2. Top-right card: Product name “Noema” in vivid blue. Minimal blue gradient N-like logo with three short horizontal motion lines. Description: “面向 Coding Agent 的\n本地优先记忆系统。” Tags, exactly 3: “Local-first”, “词法召回”, “审核队列”.
3. Bottom-left card: Product name “Jian” in green. Logo resembles the Chinese character “简” inside a thin green circular outline with a small bamboo-like vertical mark. Description: “Rust 原生跨平台 UI 框架，\n.op 文件即应用。” Tags, exactly 3: “Rust-native”, “Cross-platform”, “.op App”.
4. Bottom-right card: Product name “Zode” in purple-blue gradient. Logo is a bold futuristic Z/terminal symbol in purple and navy. Description: “AI 原生终端编码 CLI，\n面向高效开发工作流。” Tags, exactly 3: “CLI”, “插件架构”, “全屏 TUI”.

Footer: Add a rounded horizontal bottom bar with a thin blue line and dot accents. In the center, show exactly 4 workflow steps separated by arrows: “Design → Memory → Framework → CLI”. Beneath it, add the small Chinese caption: “从创作到编码，形成完整的 AI 工具协作闭环”.

Visual style: Modern enterprise AI product slide, crisp vector-like typography, blue/cyan/purple/green accents, soft gradients, high readability, no photo texture, no watermark. Use balanced margins and make all text sharp and legible.

Goal: Create a richly illustrated vintage storybook poster cover for a fantasy children's tale titled {argument name="title text" default="The Name Dragon"}, with a warm, magical library scene and ornate hand-lettered typography.

Canvas: Vertical portrait poster, 3:4 aspect ratio, like an aged illustrated book cover. Use a dark teal-black background with distressed parchment texture, copper-orange decorative border, corner flourishes, stars, moons, and tiny magical sparkles.

Top typography: The upper half is dominated by large ornate serif lettering reading {argument name="title text" default="The Name Dragon"}. The letters are warm antique gold with engraved line texture, subtle shadowing, curled swashes, and a hand-printed vintage look. Above and around the title are small stars, a crescent moon, a small flame icon at the top center, and symmetrical vine-like flourishes. Beneath the title is a small shield crest with a flame emblem, then the subtitle {argument name="subtitle text" default="Every roar remembers home"} in cream-gold serif type.

Main scene: The lower half shows a cozy magical library inside an arched alcove glowing with sunset light. Bookshelves, lanterns, scrolls, framed art, plants, and warm amber light fill the background. The floor is covered with scattered parchment name cards and golden sparkles.

Characters: Include exactly 4 main characters. 1) A small orange baby dragon seated at center, cute and expressive, with round green eyes, little gray horns, red-orange scales, tiny wings, a curled tail, and clawed feet; it holds a parchment reading {argument name="dragon parchment text" default="Who am I?"}. 2) A young girl sitting on the left, brown hair tied in a messy bun with a red ribbon, wearing a teal embroidered cloak and red scarf, smiling gently with her chin in her hand. 3) A wise owl on the right wearing round spectacles and a dark robe, reading a brown book labeled {argument name="book title" default="The Book of Names"}. 4) A small orange fox at the lower right wearing a green scarf, looking toward the dragon with curiosity.

Visible text and countable items: Include exactly 6 loose name cards on the floor labeled: “Scorch”, “Spark”, “Ember”, “Puff”, “Home”, and “Blazy”. Add one hanging red banner on the left bookshelf with the text {argument name="banner text" default="NAMES LIVE IN MEMORIES AND HEARTS"}. Keep the title, subtitle, dragon parchment, owl book label, banner, and six name cards readable.

Visual style: Whimsical hand-painted children's book illustration, vintage fantasy cover, warm orange and gold lighting, deep teal shadows, textured paper grain, slightly imperfect ink lines, ornate decorative framing, cozy emotional atmosphere. Avoid photorealism, modern UI elements, or extra characters.

Create a vertical 9:16 ultra-realistic lifestyle travel selfie shot on an {argument name="phone model" default="iPhone 17 Pro"}, with a candid wide-angle front-camera feel and high dynamic range. In the foreground, show exactly 2 young East Asian women dressed in elegant fantasy hanfu/xianxia-inspired outfits, crouching together on stone steps and happily looking down at a large unfolded tourist map spread across a rough stone block. The woman on the left wears a pale ice-blue robe with deep blue trim, silver embroidered patterns, metallic silver-blue wrist bracers, dangling earrings, delicate forehead ornament, and long black hair in a half-up style with ornate silver hairpieces; she smiles softly while pointing at the map with one finger. The woman on the right wears a darker navy and charcoal robe with blue inner layers, silver embroidery, silver wrist bracers, dangling earrings, a white forehead marking, and black hair pinned up with a hair stick; she laughs brightly toward the camera side while pointing at the map. Behind them, place the Ruins of Saint Paul's in Macau centered and towering in the background, with its historic stone facade, arched openings, statues, columns, cross at the top, and sunlit plaza steps leading up to it. Add light tourist activity in the distance, potted plants and European-style buildings along both sides, clear blue sky, warm midday sunlight, natural shadows, realistic skin texture, detailed fabric texture, slight smartphone lens distortion, shallow but readable depth of field, and a spontaneous social-media travel-photo mood. Include a small subtle watermark-like Chinese text near the bottom center reading {argument name="watermark text" default="蓝心｜蓝心 AI"}. No extra people in the foreground, no modern casual clothing on the two women, no cartoon style, no blur on the faces, no deformed hands, and keep the map visibly paper-like with printed roads and place markings.

Goal: Create a widescreen philosophical typography poster in a bold Chinese big-character-poster style, where the typography itself forms an ascending staircase-like composition and integrates a human silhouette.

Canvas: 16:9 horizontal poster, dark navy-black textured paper background with subtle scratches, grain, faded fold lines, and vintage print wear. Use an off-white ink color for all main type, slightly distressed and roughened like screen-printed paper.

Main typography: Fill most of the canvas with exactly 5 enormous blocky Chinese characters reading {argument name="main phrase" default="我思故我在"}. The characters should be custom geometric slab forms, extremely heavy, angular, architectural, and arranged from left to right with a gradual rising step rhythm: the first character lower and wide, the second taller, the middle characters higher, and the final character towering at the far right. Make the characters look like massive concrete structures or stairs, with rectilinear cutouts and negative spaces. The layout should feel intelligent and intentional, not standard font rendering.

Figure integration: Place exactly 1 dark silhouette of a seated thinker inside the negative space between the third and fourth large characters, roughly lower center-right. The person sits hunched forward with one elbow on a knee and hand near the chin, evoking Rodin’s thinker pose. The silhouette should be the same dark color as the background, appearing carved out of the white typography.

Small text elements: Add exactly 9 small typographic text groups around the poster: 1 small Chinese quote in the upper left reading {argument name="quote text" default="思想不是被肯定明的方式，而是使我们得以存在的理由。——笛卡尔"}; 1 bottom-left English title reading “COGITO ERGO SUM”; 1 small bottom Chinese translation group reading “我 思 故 我 在” and “RENE DESCARTES”; 1 bottom label reading “PHILOSOPHY EXHIBITION”; 1 bottom label reading “EXISTENCE IS NOT GIVEN. IT IS REALIZED.”; 1 bottom label reading “THINK. QUESTION. THEREFORE. BE.”; 1 bottom-right roman code reading {argument name="roman code" default="MMXXIV"}; 1 bottom-right copyright line reading “© ALL RIGHTS RESERVED”; and 1 minimalist phi-like circular symbol at the far bottom right. Keep all small text in condensed uppercase sans serif with wide tracking, thin strokes, and careful baseline alignment.

Visual style: Minimalist brutalist graphic design, East Asian poster aesthetics, Bauhaus-inspired grid discipline, high contrast, cinematic negative space, off-white ink on deep blue-black, subtle paper texture, aged edges, no gradients except natural texture. The poster should feel like an art exhibition announcement about René Descartes and the phrase “I think, therefore I am.”

Constraints: Use exactly the 5 large Chinese characters and exactly 1 seated silhouette. Do not add extra figures, icons, illustrations, or decorative objects beyond the 9 small text groups. Preserve the stark poster composition, monumental typography, and ascending stair-step structure.

Goal: Create a clean McKinsey-style strategy cover slide about {argument name="topic" default="compound-interest growth methodology"}, combining Chinese headline typography with an engineered growth flywheel diagram embedded inside the Chinese word “增长”.

Canvas: Wide horizontal presentation cover, about 16:5 aspect ratio, off-white paper background with subtle grain, thin black top rule and faint gray construction grid lines.

Top header: At top left, write “复利时代的增长方法论 | METHODOLOGY FOR GROWTH IN THE AGE OF COMPOUND INTEREST”. At top right, write “STRATEGY COVER | GROWTH SYSTEM | EXECUTIVE INSIGHT”. Use small uppercase condensed sans-serif text, black.

Central layout: Place the huge dark navy Chinese characters “增长” across the center, occupying most of the slide width. Make the characters feel like architectural, mechanical letterforms: thick navy strokes with black outlines, subtle bevels, and embedded rectangular modules. Integrate a circular flywheel mechanism directly into and behind the characters, centered between the two characters.

Discrete central elements: Include exactly 1 large word made of 2 Chinese characters: “增” and “长”. Include exactly 1 main circular flywheel with 3 concentric rings, a light gray hub, blue curved arrows, and many small gear-like white rectangular blocks around the outer arc. Include exactly 3 labeled points above the flywheel connected by a thin blue arc: “INPUT”, “OUTPUT”, and “REINVESTMENT”. Include exactly 1 label below the flywheel: “GROWTH ACCELERATION”. Include exactly 2 orange accent nodes on the flywheel and several small blue nodes on the guide lines. Include exactly 2 diagonal segmented ramps attached to the right character, each made of small white rectangular steps. Include exactly 1 horizontal navy arrow extending to the right from the center-right of the composition. Include small blueprint coordinate markings in pale gray near the grid, but keep them secondary and mostly decorative.

Bottom title: Centered beneath the graphic, write the bold Chinese title “构建可持续的螺旋增长飞轮”, and below it in smaller uppercase English write “BUILDING A SUSTAINABLE SPIRAL GROWTH FLYWHEEL”.

Visual style: Corporate consulting deck cover, precise vector illustration, blueprint engineering aesthetic, dark navy, black, off-white, pale gray, and small orange highlights. Use crisp linework, restrained colors, lots of white space, and balanced symmetry. The diagram should look disciplined, technical, and premium rather than playful.

Constraints: Preserve the exact visible Chinese and English text. Do not add people, photos, logos, watermark, shadows, or extra sections. Keep the flywheel physically embedded into the word “增长”, not floating separately.

Create a highly realistic vertical smartphone photo of {argument name="character description" default="a young East Asian woman factory technician with a short dark-brown bob haircut and wispy bangs"} seated at an industrial assembly workstation, viewed from a slightly high wide-angle perspective as she turns her head up to look directly into the camera with large expressive eyes. She wears exactly 3 visible protective/work items: a navy-blue zip-front work uniform jacket, a light-blue disposable surgical face mask, and bright blue nitrile gloves on both hands. Her hands are positioned at the right side of the frame, handling small shiny cylindrical metal components on a stainless-steel roller conveyor or inspection rail. The setting is a clean modern factory line with metal frames, rails, machinery, bolts, rollers, green flooring, fluorescent overhead lights, and cool gray industrial surfaces; keep the background detailed but slightly softened by depth of field. Use soft bright overhead lighting, realistic skin and fabric texture, mild lens distortion from a close phone camera, cool teal-blue color grading, and a candid documentary feel. Include exactly 2 small visible watermark-style text marks: a rounded translucent label in the top-left reading {argument name="top left label" default="AI生成"}, and a white logo/text mark in the bottom-right reading {argument name="bottom right logo text" default="即梦AI"}. No extra people, no extra text, no cartoon styling, no glamour studio lighting.

Create a cinematic photorealistic wide shot in a dusty desert mountain wasteland, blending ancient Chinese wuxia fantasy with a modern punk survivor character. Show exactly 4 young women standing in a dramatic lineup, all facing the camera with serious, defiant expressions: 1) left woman in a pale mint-green flowing hanfu robe with long black hair half tied in a top bun, holding a sheathed ornate sword diagonally at her waist; 2) center-front woman in a black layered martial-arts robe with messy high bun hair, gripping a short curved dagger forward in a threatening pose; 3) center-rear woman in a faded beige-green robe with long dark hair, aiming a tall wooden bow directly toward the viewer with an arrow drawn; 4) right woman with short {argument name="hair color" default="pink"} hair wearing a black tank top, ripped dark denim shorts, black headphones around her neck, one hand on her hip and the other holding a cigarette near her mouth. Set them in a sandy, windblown canyon with hazy mountains in the background, warm beige sunlight, airborne dust, faint smoke from the cigarette, and a partially visible old car wreck behind the modern woman on the right. Use a gritty movie still look, realistic skin and fabric texture, shallow depth of field, muted earth tones, high detail, natural lens perspective, 16:9 aspect ratio. The overall mood should feel like {argument name="mood" default="post-apocalyptic wuxia standoff"}, with no visible text, no watermark, and no extra people.

Create a vertical cinematic fantasy portrait of {argument name="character name" default="a beautiful ancient Chinese forest goddess"} standing among twisted dark tree branches in a sun-dappled woodland. She is a young ethereal woman with porcelain skin, soft red lips, delicate makeup, a small forehead ornament, and a calm slightly wistful gaze looking toward the camera with her head gently tilted. Her {argument name="hair color" default="dark brown-black"} hair is styled in an elaborate high Tang/Song-inspired updo with ornate gold filigree hairpins, red beads, floral ornaments, dangling chains, and long jeweled earrings. Dress her in an opulent translucent hanfu in {argument name="robe colors" default="pale pink, misty blue, lavender, and warm cream"}, with layered sheer gauze sleeves, embroidered floral borders, gold thread patterns, beaded tassels, and a flowing sash; the fabric should glow where backlit and ripple softly as if caught by a light breeze. Place her three-quarter length in the center, body angled slightly left, one hand relaxed near her waist and the other lowered along the robe. The forest should feel enchanted and ancient: gnarled black-brown branches framing her like an arch, small leaves and blossoms, deep green shadows, bright white sunlight streaming from behind her head and shoulders, sparkling bokeh, volumetric haze, rim light around her hair and sleeves, and a dreamy film-photography texture. Use shallow depth of field, realistic skin detail, high-fashion editorial composition, soft bloom, subtle grain, natural lens flare, delicate pastel highlights, and rich shadow contrast. No text, no watermark, no modern objects, no extra people.

Goal: Create a dark neon Chinese timeline infographic poster showing the evolution history of an AI programming partner stack, with four horizontal stage cards and paired software tools.

Canvas: Square 1:1 image, about 768×768 px, black-to-deep-navy background with subtle glow, thin neon strokes, high-contrast modern tech aesthetic.

Header: Large bold Chinese headline at the top: {argument name="headline text" default="我的 AI 编程搭档进化史 🚀"}. Use a white-to-cyan-to-purple gradient emphasis across the AI/programming words. Under it, add a smaller gray subtitle: {argument name="subtitle text" default="不同阶段，不同选择，适合自己的才是最好的。"} with thin horizontal divider lines on both sides.

Layout: Four stacked rounded rectangular timeline cards, evenly spaced vertically. A thin vertical timeline line runs down the far left with exactly 4 glowing circular nodes aligned to the 4 rows: blue, green, yellow, purple. Each row has a left date/stage block and a larger right tools block. The right block shows exactly 2 app tiles separated by a large plus sign.

Rows: Use exactly 4 rows. Row 1 is blue and labeled {argument name="row one text" default="1-2 月份｜探索尝试阶段"}; tools are Cursor + Claude. Row 2 is green and labeled {argument name="row two text" default="3-4 月份｜效率提升阶段"}; tools are CodeBuddy + Claude. Row 3 is yellow and labeled {argument name="row three text" default="5-6 月份｜深度协作阶段"}; tools are Pi + Codex. Row 4 is purple and labeled {argument name="row four text" default="7 月份｜专注交付阶段｜Now!"}; tools are Orca + Codex.

App tiles and icons: Create exactly 8 app icon/name groups total: 1) Cursor, black rounded-square icon with a white faceted diamond/cursor mark, sublabel “AI IDE”; 2) Claude, orange rounded-square icon with a white radial starburst, sublabel “AI Assistant”; 3) CodeBuddy, bright green rounded-square icon with a cute black robot face and white eyes, sublabel “AI Coding Assistant”; 4) Claude, same as above; 5) Pi, cream rounded-square icon with large yellow “Pi”, sublabel “AI Developer”; 6) Codex, teal rounded-square OpenAI knot style icon, sublabel “by OpenAI”; 7) Orca, purple rounded-square icon with a black-and-white orca whale, sublabel “AI IDE”; 8) Codex, same as above. Tool names are large bold white text; sublabels are smaller light gray text. Plus signs are large, centered, and light gray.

Footer: At the bottom center, add a small colorful handwritten-style line: “💗 工具在变，热爱不变，持续进化 🚀”.

Visual style: Cyber UI dashboard look, rounded cards with translucent dark fill, colored neon borders matching each row, soft outer glows, crisp typography, clean alignment, no clutter, no watermark.

Create a photorealistic vertical smartphone mirror selfie that looks like a natural post on X: an adult {argument name="subject age and ethnicity" default="Japanese woman in her 20s"} casually taking a photo while choosing clothes in a closet. She is close to the mirror, framed from head to upper thighs, sitting or crouching slightly with a relaxed posture, holding a silver triple-camera smartphone in her right hand so the phone partly covers the right side of the image. She has {argument name="hair style and color" default="soft light-brown shoulder-length bob with airy bangs"}, large natural eyes, subtle everyday makeup, smooth skin, and a calm, slightly shy neutral expression while looking at the phone screen/mirror. Outfit: modest casual loungewear, exactly 2 visible clothing pieces: a fluffy dark charcoal-gray cardigan and a white ribbed T-shirt underneath; no glamorous styling, no revealing pose, no gravure look. Background: a real closet area with exactly 6 visible hanging garments in soft neutral tones: cream, blue-gray, white, beige, black, and pale pink; sheer white curtain on the left with gentle morning daylight. Composition should feel unposed and domestic, with soft natural lighting, shallow depth of field, realistic mirror edge visible on the right, slight smartphone-camera softness, authentic skin texture, and no added text, watermark, studio lighting, or overly polished fashion-shoot atmosphere. Use {argument name="mood" default="quiet morning, natural, approachable everyday feeling"}.

Goal: Create a side-by-side comparison image showing an AI-generated fashion portrait before and after Photoshop-style color correction, featuring the same young woman in pastel pink sweet lolita/gyaru street fashion on a European boutique shopping street.

Canvas: Landscape 4:3 image, split exactly into two equal vertical panels with a hard center divide. Photorealistic fashion editorial style, full-body subject in both panels, street-level perspective, shallow depth of field, soft daylight.

Left panel: Show the uncorrected version labeled in bold black Japanese text near the bottom: 「ノーマル生成」. The model stands in front of a bright cream-white boutique facade with large display windows and a pale sign. Overall color is slightly flat, pale, low-contrast, airy, and softly washed out. The model faces forward with a gentle smile, one hand near her hair, wearing a pastel pink corset top with puff sleeves, pink short bottoms, decorative waist chains, patterned pastel tights, and tall pink lace-up platform boots.

Right panel: Show the corrected version labeled in bold black mixed text near the bottom: 「Photoshop補正互換で生成」. The same model stands on a similar elegant shopping street in front of a boutique with a visible hanging sign reading {argument name="shop sign text" default="LIZ LISA"}. The lighting and color correction should look more polished: warmer highlights, richer contrast, clearer skin tones, slightly deeper shadows, brighter pinks, crisp details, and a more professional retouched fashion-photo finish. The model poses with one hand in her hair and one leg slightly bent, smiling at the camera.

Subject details: A cute young Japanese fashion model with fair skin, long wavy twin-tail hair in {argument name="hair color" default="pink blonde"}, soft bangs, ribbons and pearl hair accessories, glossy idol-like makeup, large bright eyes, and a friendly expression. Outfit color theme is {argument name="outfit color theme" default="pastel pink"}. Include many kawaii accessories: layered pearl necklaces, bracelets, bows, lace, floral charms, decorative belt chains, and platform boots.

Visible text: Exactly 2 large bottom labels: left label 「ノーマル生成」 and right label 「Photoshop補正互換で生成」. Keep the labels bold, black, horizontal, and placed over the pavement area without outlines.

Constraints: Make both panels look like variations of the same fashion shoot, not a collage of unrelated characters. Use realistic anatomy, full-body framing, no extra logos except the boutique sign, no watermark, no additional captions, no borders other than the central split.

Goal: Create a wide comparison graphic showing how four prompt-writing AI model tiers produce the same landscape scene in GPT Image 2, with a dark editorial UI frame and Japanese headline text.

Canvas: Extra-wide horizontal banner, approximately 2:1 aspect ratio, black/dark charcoal background, clean social-media comparison layout. Use thin black dividers between image columns and a compact top title area.

Top text: Large white Japanese headline: {argument name="headline text" default="同じ構図・同じ画像AI（GPT Image 2）。プロンプトを“書いたAI”だけが違う4枚"}. Under it, add a smaller gray subtitle: {argument name="subtitle text" default="変えたのは「プロンプトを書くClaudeのモデル階層」だけ（左＝最も安い → 右＝最上位）"}.

Layout: Four equal vertical landscape panels placed side by side, each with a small dark label bar at the top. Use exactly 4 panels, ordered left to right:
1. Label: “haiku ｜ 最も安い” — a dramatic semi-realistic anime landscape at golden sunset, high mountain range in the distance, layered misty valleys, wildflower-covered rocky foreground, a tiny solitary person standing on a ridge, warm orange-purple clouds, slightly over-embellished cinematic atmosphere.
2. Label: “sonnet” — cleaner and more elegant sunset mountain valley, broad sky with soft glowing cumulus clouds, golden light spilling across rolling hills, a small lone figure on a ridge path, balanced composition, polished anime background style.
3. Label: “opus” — high-quality but restrained sunset vista, deep blue sky fading to orange near the horizon, large illuminated cloud formations, grassy hillside foreground, distant lake or reflective water, tiny figure near the bottom center, premium but conventional scenic framing.
4. Label: “fable ｜ 最上位” — emotionally directed bright morning or late-afternoon landscape, turquoise-blue sky, towering luminous white clouds, sunbeams over a vast green valley, a small person in a white dress and hat standing in meadow grass at lower right, more narrative and poetic mood.

Visual style: Refined Japanese anime background art blended with painterly cinematic lighting, highly detailed skies and atmospheric perspective. Keep all four panels similar in composition and subject matter so the comparison feels controlled, but vary mood and art direction per label. Use crisp white text, dark label bars, no logos, no watermark.

Constraints: Preserve exactly 4 image panels and exactly these 4 visible panel labels. Make the title and subtitle legible. Keep the overall image looking like a posted comparison chart rather than a standalone landscape collage.

Create a photorealistic 3-by-4 comparison grid of exactly 12 fictional Gen Z influencer portraits, all shot as close-up bust portraits in the same neon-lit gaming bedroom or streamer room environment. Use a 16:9 horizontal canvas with thin white gutters separating the tiles, soft shallow depth of field, cinematic RGB monitor glow, pastel pink and cyan bokeh, realistic skin texture, high-fashion cosplay styling, and a polished social-media portrait aesthetic. Each person faces the camera or slightly angles toward it, with moody studio lighting and no text or logos.

Layout: 12 equal rectangular portrait tiles arranged in exactly 3 rows and 4 columns.

Subjects, counted left to right, top to bottom:
1. Pale young woman with short white bob hair covering one eye, black headband, black choker, white crisscross halter top with black straps, doll-like expression.
2. Young woman with black twin tails fading to dark red, straight bangs, pink heart earrings, black blouse with a large pale pink bow, soft sad expression.
3. Androgynous young person with messy dark navy-blue short hair, blue eyes, black techwear jacket, ear cuff jewelry, cool serious expression.
4. Sweet pastel girl with long wavy pink-blonde hair decorated with many tiny butterfly clips, rainbow pastel cardigan, holding a small plush toy, cheerful smile.
5. Silver-gray-haired young woman in a loose white knit sweater, soft melancholic gaze, hand near her chest.
6. Dark-haired young woman with blunt black bob and straight bangs, round glasses, oversized dark hoodie, large black headphones around her neck, introverted expression.
7. Blonde curly lolita-style girl with big pale bow hair accessory, white frilled blouse, black pinafore straps, glossy pink lips, delicate doll makeup.
8. Purple bob-haired gothic girl with straight bangs, black lace high-neck top, small silver pendant necklace, intense serious stare.
9. Young man with asymmetrical brown undercut hair, black shirt, large gaming headset with microphone, seated in a gaming chair, confident neutral expression.
10. Pastel lavender bob-haired girl with soft bangs, blue hair clips, baby-blue sweater with small cloud decorations, dreamy ethereal look.
11. Red-haired punk girl with layered shaggy hair and a small braid, black choker, dark lipstick, black top, rebellious expression.
12. Ice-blonde elegant girl with very long straight hair, hair clips near both temples, white sleeveless turtleneck with black straps, pale icy makeup, calm piercing gaze.

Overall style: ultra-realistic influencer variation sheet, Japanese kawaii and cyber-goth fashion mix, anime-inspired styling but real human faces, high-end portrait photography, 35mm lens look, shallow focus, detailed eyes, soft facial retouching, atmospheric LED lighting. Customize the overall theme as {argument name="influencer theme" default="fictional Japanese cyber-kawaii gaming influencers"}, the room setting as {argument name="background setting" default="neon-lit gaming bedroom"}, the lighting palette as {argument name="lighting colors" default="pastel pink and cyan RGB glow"}, the portrait realism as {argument name="visual style" default="photorealistic high-fashion cosplay portrait photography"}, and the grid format as {argument name="grid format" default="3 rows by 4 columns with thin white gutters"}.

Create a square sepia-toned storybook illustration of {argument name="main character" default="a fluffy cream-colored hamster"} standing in the center of an elegant old European music salon, eyes gently closed and head bowed as if finishing a performance. The hamster is round, soft, and very detailed with fine ink hatching, tiny paws, small ears, and a calm solemn expression, holding a delicate violin bow across its chest with a tiny dark violin or instrument tucked near its paws. The setting is a richly decorated classical room with polished reflective wooden floors, tall paneled walls covered in ornate rococo carvings, a bright window with sheer curtains on the left, and antique furniture. Include exactly 3 prominent background objects: an ornate upholstered chair on the left, a large cello leaning upright on the right, and a vase of pale flowers on a small carved table behind the cello. Use warm monochrome beige, brown, and cream tones, like an antique pencil-and-ink wash drawing with soft backlighting, subtle floor reflections, intricate linework, nostalgic atmosphere, and no visible text or modern objects.