GPT Image 2 prompts

2x2 grid, 16:9 do this for {argument name="number of potters" default="4"} famous master potters   input:   [{argument name="potter data" default="four master potters & their glaze chemistry charts"}]   system:   render each potter as a “raw clay becoming vessel.” each panel contains one central artisan emerging from a giant, wood-fired kiln and plaster bat. the lower half is made from raw grog, unfired clay, wooden ribs, ash glaze recipes, and kiln wadding. the upper half is a hyper-realistic cinematic potter covered in slip, radiant heat distortion, leather aprons, and sweat.   transformation logic:   the chemistry of the kiln becomes physical texture. glaze crawls and crackles become physical skin textures or fabric patterns, throwing rings become ribbed, structural armor, and kiln wadding becomes heavy, structural supports or jewelry.   environment:   the kiln expands into the roaring, blinding interior of a wood-fired anagama kiln or a hyper-detailed, dusty pottery studio, with massive thermal heat distortion and flying embers.   data layer:   add refined ceramic callouts: - “cone 10 reduction" - “thermal shock" - “silica flux ratio" - “atmospheric carbon" - “vitrification point" - “glaze crawl"   visual style:   photorealistic industrial and craft photography, intense warm lighting (oranges, reds, and deep earth browns), heavy smoke and heat distortion textures, macro details of raw earth and glassy glazes.   composition:   strict 2x2 grid, each panel like a premium craft and material science magazine feature.

Vintage rustic farmers market poster featuring {argument name="fruit or product" default="FRUIT OR PRODUCT"}, overflowing wicker baskets filled with {argument name="main subject" default="MAIN SUBJECT"}, arranged on a weathered wooden table with {argument name="props" default="PROPS"}, glass jars, preserves, wildflowers, burlap cloth and country market details, aged parchment background, ornate Victorian typography reading "{argument name="title text" default="TITLE TEXT"}", distressed paper edges, warm harvest color palette, rich earthy tones, highly detailed painterly illustration, antique food advertisement, nostalgic farmhouse aesthetic, dense still-life composition, vintage packaging artwork, commercial print design, ultra detailed, masterpiece.

Ultra-detailed 2x2 portrait grid featuring four different young men, each poking their head through a torn, irregular hole ripped into bright yellow seamless backdrop paper. The cast has a 90s indie-rock vibe with varied hair textures, think wild curls and locs. Studio lighting is soft and flat, like a pro magazine shoot. The vibrant yellow paper fills the frame while a neutral beige background peeks through the jagged tears. Editorial photography feel, crisp focus, 8k resolution. How to dial it in: Framing: 2x2 quad-grid layout, each subject peeking through a punched hole. Hero Prop: ripped yellow construction paper, rough torn edges, seamless paper BG. Talent Look: Gen Z streetwear, indie-band energy, 90s alt-lite vibe, expressive young faces. Capture Specs: even flat lighting, shallow depth of field, clean studio look, punchy color saturation.

A realistic smartphone mirror selfie of a {argument name="subject" default="woman in her early-to-mid 20s"} seated on a {argument name="prop" default="black gym bench"} inside a {argument name="location" default="contemporary fitness center"}. She sits slightly right of center in the vertical frame, leaning back casually, holding her phone in front of her face to capture the reflection. She wears a dark gray forward-facing baseball cap with long straight black hair flowing out beneath it. Her expression is detached and cool, eyes directed at the phone or mirror, lips lightly closed and partially obscured by the device — no smile, as if caught mid-rest after training. Her top is a black bandeau or single-shoulder strappy athletic piece, fitted with slight creasing and matte stretch texture, leaving collarbones, shoulders, and the tops of her arms bare. Her bottom is a black high-waisted mini skirt or skort, snug through the hips, with natural fabric folds near the thighs from the seated position. She wears tall brown boots. One leg crosses into the foreground while the other extends diagonally toward the lower-left corner, producing a strong diagonal line and subtle wide-angle distortion that elongates the legs. Skin tone is fair and warm ivory, naturally lit by cool overhead fluorescents but not plastic or retouched. Exposed areas include the neck, shoulders, collarbones, upper chest, both arms, hands, a sliver of waist, and significant portions of both thighs, knees, and lower legs. Background shows a modern gym: gray walls, dark gray tile flooring, black benching equipment, weight bars, a nearby workout bench to the left, and a large rounded-rectangular glass window behind her reflecting cool blue-white fluorescent light. Primary lighting is from wide overhead LED panels. Retain mirror scratches, glass glare streaks, mild exposure blooming, and slight image compression artifacts. Visual feel should mimic a real small-sensor smartphone wide-angle mirror selfie: vertical 2:3 or 3:4 frame, mid-height mirror viewpoint, subject shifted slightly right, legs reaching lower-left. Deep focus keeps both the figure and background gym equipment readable. Avoid polished studio aesthetics — preserve casual social framing, minor imperfections, background equipment clutter, accidental reflections, and authentic indoor digital grain.

Create a crisp, clean infographic storyboard poster for {argument name="dish" default="THE PAD THAI COOK"}. Wide 16:9 layout, white background, black borders, bold black typography, premium Pixar 3D stylized rendering, {argument name="colors" default="bright vivid colors — golden noodles, vivid orange tamarind"}

Create a {argument name="style" default="2.5D IDENTITY BOARD"} using image1 as exact reference and motorcycle Image2 in detail. Preserve face, hair, eyes, outfit, and colors with 100% consistency across all panels. Clean studio sheet on soft

Create a high-end editorial character profile spread layout split into two sections.
LEFT SECTION
→ Large bold typography "{argument name="character name" default="JACE ROMAN"}" in heavy black uppercase font
→ Subtitle: "{argument name="subtitle" default="THE CAGE IS HIS CANVAS"}"
→ Full body photorealistic image of a real

The image must be extremely clean. Do not include any photographic grain, digital noise, or blurry artifacts. Keep the skin and background perfectly smooth and pure.

highly detailed {argument name="vehicle" default="F1 car"} design sheet,
concept showcase layout,
technical illustration,
character + vehicle design sheet,
previsualization board,
photorealistic rendering,
multiple angles and views,
production-quality presentation,

Create a cinematic full-body fashion portrait of {argument name="character description" default="a smiling young woman with long dark brown hair"} standing in the foreground on the left side of the frame, using a realistic face-reference style if a face is provided. She poses confidently with both hands raised near her shoulders making finger-heart gestures. Outfit: oversized glossy streetwear trench jacket in pink and navy panels, with bold black lettering “PEW BE” on the chest and lower coat, a white cropped tank top printed with “BE LIKE US” and a small star, loose high-waisted navy cargo pants with pink piping and dangling straps, and chunky pink-and-white sneakers. Behind her, place one enormous semi-transparent close-up portrait of the same woman’s face and upper shoulders, filling the right half and top of the composition like a dramatic billboard or double-exposure background. Scene: neon-lit urban street at night with wet reflective pavement, pink neon glow on the left, cool blue storefront lights on the right, dense rolling fog and smoke around the feet and background, high contrast rim lighting, shallow depth of field, glossy reflections, cyberpunk fashion editorial mood. Composition: vertical 3:4 poster frame, foreground figure full body visible from head to shoes, giant background face cropped by the top and right edges, no extra people, no watermark, keep all visible clothing text exactly as specified.

Goal: Create a polished promotional banner announcing {argument name="product name" default="GPT Image 2"} inside Magic Edit, using a soft pastel anime aesthetic and a clean SaaS-style layout.

Canvas: Wide horizontal 16:9 social media banner, approximately 1200×675 px. Use a dreamy pink, lavender, and white gradient background with soft bokeh blur, faint sparkles, and subtle depth shadows.

Layout: Split composition with 2 main panels. Left panel is a frosted-glass rounded rectangle card occupying about 40% of the width, positioned left-center with a soft drop shadow. Right panel is a large rounded-square anime illustration card occupying about 48% of the width, positioned right-center, slightly taller than the text card, with a subtle shadow and rounded corners.

Left text card: White translucent glassmorphism surface with very rounded corners. Set all typography in dark purple, modern rounded sans-serif, bold for the main announcement. Include exactly 5 visible text lines/groups: “New in”, “Magic Edit”, “{argument name="headline product text" default="GPT Image 2"}”, “is here”, and a smaller line “Choose from Light / Standard / Premium”. At the bottom of the card place exactly 1 pill-shaped white button with pink-purple text reading “{argument name="price badge text" default="Light: 10 diamonds"}”.

Right illustration card: Pastel anime girl portrait in a highly detailed glossy illustration style. She has {argument name="hair color" default="lavender-purple"} twin-bun or side-bun hair with loose flowing curls, white flower hair accessories, red-pink ribbon bows, a pink beret-like hat, and a pink cardigan over a frilly white blouse with small pink buttons. She is surrounded by large soft pink flowers, petals, sparkles, and glowing sunlight. Her pose is gentle and cute, with one hand near her face. Place a solid opaque lavender rectangular censor/edit mask over the center of her face, covering the eyes, nose, and mouth area, with sharp square corners; the mask should be clearly visible as an editing overlay.

Visual style: Pastel kawaii, glossy anime rendering, soft bloom, high-key lighting, spring floral atmosphere, gentle gradients, clean app-promo design, premium but friendly. Use rounded corners throughout and subtle shadows for floating depth.

Constraints: Keep the composition uncluttered with no extra logos or watermarks. Use exactly 2 main cards, exactly 1 button, and exactly 5 text lines/groups on the left card. Ensure the product announcement text is large and readable, and the anime image remains clearly secondary to the promotional message.

Create a vertical cinematic fantasy street scene from an extremely low ground-level viewpoint, centered on the lower legs of a colossal or giant woman standing in shallow reflective water. The main subject wears tall glossy black leather gothic boots reaching above the knees, with platform soles, pointed rounded toes, tight lacing up the front, and many metal buckles and straps on both boots; the visible outfit continues as tight black pants above the boots. The boots dominate the frame, cropped at the upper thighs, with the feet planted in ankle-deep water and a strong mirror reflection extending downward across the lower half of the image. Surround her with a grand steampunk / dark fantasy city street: ornate Victorian buildings on both sides, towers, balconies, pipes, cables and overhead wires, distant haze, and small pedestrians walking in the background to emphasize scale. Include exactly 10 visible pedestrians: 5 on the left side of the street, 1 central figure walking toward the giant boots, and 4 on the right side, all dressed in dark long coats or fantasy urban clothing. Add natural water ripples around the soles and across the reflection, as if the street is flooded after rain, with teal-blue water, shimmering highlights, and subtle distortion of the boot and architecture reflections. Lighting should be dramatic backlit sunrise or sunset, bright glow from the upper center-left, cool blue shadows, warm rim light on leather and buildings, cinematic depth, high detail, ultra-realistic anime fantasy illustration style, sharp foreground boots, atmospheric background, no text, no watermark. Use {argument name="boot color" default="glossy black"}, {argument name="city style" default="steampunk Victorian dark fantasy"}, {argument name="water effect" default="shallow reflective water with natural ripples"}, {argument name="time of day" default="backlit sunrise"}, and {argument name="main subject scale" default="colossal giant woman"}.

Goal: Create a vertical cinematic anime emotion-fragment poster for a dark xianxia swordsman named {argument name="character name" default="Han Li"}, using a moody shattered-glass collage style.

Canvas: Tall 9:16 poster, approximately 675 x 1200 px, black and deep green color palette with cold blue highlights, high contrast, dramatic rim lighting, realistic anime-fantasy rendering.

Layout: The poster is split by jagged cracked-glass shards with thin glowing gold fracture lines. Use exactly 5 main visual fragments: 1 large central portrait fragment showing a long-haired male cultivator in layered dark robes reaching toward the viewer with a spell-casting hand gesture; 1 upper-right close-up shard showing one intense eye, pale skin, loose black hair, and a small blood scratch; 1 mid-right shard showing an open hand surrounded by lightning; 1 lower-left shard showing folded dark robe fabric and bamboo leaves; 1 lower-right shard showing a black sword hilt or scabbard engraved with small Chinese characters and crossed by lightning cracks. Add a dark opaque rectangular block over the center of the face, as if the face is intentionally obscured.

Subject details: The character has long windblown {argument name="hair color" default="black"} hair tied partly back with a small silver ornament, a sharp mysterious expression, pale skin, dark layered ancient Chinese robes with subtle embroidered patterns, and hands performing a cultivation spell. His front hand is closest to camera, fingers extended dynamically with shallow depth of field. Surround him with bamboo stalks and leaves, smoky mist, sparks, and thin arcs of golden-white lightning.

Text content: Include exactly 2 visible Chinese text elements on the left side. The first is a huge vertical brush-calligraphy title reading {argument name="main Chinese title" default="韩立"}, off-white with rough ink texture, placed over a dark green brushstroke. The second is a smaller vertical green calligraphy subtitle reading {argument name="subtitle text" default="念头通达"}, placed below the title between two thin vertical divider lines.

Visual style: Ultra-detailed dark fantasy anime poster, cinematic lighting, fractured mirror composition, textured calligraphy, ink-brush energy, wuxia/xianxia atmosphere, dramatic depth, sharp glass edges, black background, bamboo forest hints, premium character-poster finish.

Constraints: Keep the composition vertical and dense, with no extra characters, no modern objects, no watermark, no English text on the poster, and preserve the exact count of 5 visual fragments and 2 text elements.

Create a moody, cinematic square portrait photograph of {argument name="character description" default="a Japanese high school girl"} standing on a narrow wet residential street on a rainy day, viewed from the waist up in side profile facing left. She wears a crisp white short-sleeve school shirt, a dark navy tie or ribbon, a dark pleated skirt, and a black backpack slung behind her. Her {argument name="hair style" default="straight short black bob haircut"} is slightly damp. She holds the handle of a transparent vinyl umbrella close to her chest with both hands; the umbrella canopy is open above and behind her, covered in many fine raindrops, with visible pale ribs, a curved clear edge, and a slightly imperfect asymmetrical shape. Her face is intentionally obscured by a flat, centered, soft-edged rectangular censor block in muted brown-gray. The background is softly blurred with shallow depth of field: a rain-slick asphalt road curving into the distance on the left with a white lane line, dark green trees and walls in the distance, and a dense hydrangea bush on the right containing exactly 9 visible blue-to-purple flower clusters among glossy green leaves. Use cool blue-green rainy tones, natural overcast light, film-grain texture, bokeh, realistic wet reflections, and an intimate 35mm street-photography look. Keep the composition square, place the subject slightly right of center, and make the umbrella dominate the upper half of the frame. No text, no watermark, no extra people.

Goal: Create a dark tactical character reference sheet for a fictional TRPG antagonist, styled like a classified dossier and fashion design board. The character is {argument name="character name" default="Talamonsu"}, a hostile target / eternal godmother figure from an occult modern-fantasy setting.

Canvas: Wide horizontal 2.4:1 character sheet, approximately 1200×500 px. Use a distressed black-green background with thin technical blueprint lines, red warning accents, worn paper textures, grid marks, compass circles, and small dossier UI details. The overall mood is grim, elegant, militaristic, and supernatural.

Layout: Split the sheet into three main zones. Left zone: an intelligence dossier panel with large red Chinese title text 「塔拉蒙苏」, Romanized name “TALAMONSU”, smaller line 「原名 / 塔拉蒙苏」, and a table of fields. Center zone: one dominant full-body standing character illustration framed by red guide lines. Right zone: torn-paper equipment detail board showing outfit material closeups. Bottom strip: smaller reference panels for alternate views and facial portrait crops.

Text content: Include the small top-left label “HOSTILE TARGET”. In the left dossier table, show these 6 rows exactly: “CODENAME: 塔拉蒙苏 / TALAMONSU”, “TRUE NAME: 塔拉蒙苏”, “OCCUPATION: 不详”, “RANK: 不详”, “AGE: 不详”, “HEIGHT: 不详”. Add a section label “HOSTILE PROFILE / 敌对档案” with tiny unreadable classified text lines and a red triangular hazard symbol. On the right upper corner, add “DESIGNATION TALAMONSU”. Near the lower center, add small text “MELIANDER ACTING GODMOTHER”. Bottom-right corner: “FILE No. XXXX” and a red barcode.

Subject details: The main character is a tall, slim adult woman with {argument name="hair color" default="dark charcoal-black"} wavy short hair, pale skin, and an elegant severe posture. Her face should be anonymized by a soft square grey blur mask, as if redacted from a classified document. She wears a structured black military-inspired tunic with asymmetrical buttons, high collar, fitted waist, black trousers, black gloves, heeled black boots, a leather belt with metal buckle, and a dramatic long teal-blue cloak with dark quilted lining and shoulder epaulets. Use desaturated teal, black, charcoal, muted brass, and blood-red UI accents.

Counted visible elements: Include exactly 1 large full-body front pose in the center. Include exactly 3 small turnaround reference figures along the bottom-left: front three-quarter view, rear cloak view, and front uniform view without the cloak dominating. Include exactly 4 square portrait panels along the bottom middle-right, each with the same grey face-redaction blur and different head/hand angles: neutral front, hand raised near mouth, side glance, and calm front-facing bust. Include exactly 5 torn-paper equipment/detail swatches on the right: a pendant or ID tag on fabric, a quilted teal fabric rectangle, a folded hood or cloak collar, a black belt with dangling chains, and stacked cuffs/epaulet or folded uniform trim.

Visual style: Semi-realistic anime concept art mixed with graphic-design dossier layout. Clean inked character rendering, muted painterly shading, thin gold and red linework, weathered paper borders, and subtle grunge scratches. Make the clothing design readable and fashion-reference oriented, with precise seams, buckles, buttons, quilting, and cape folds.

Constraints: Keep the image as a single finished character sheet, not a collage of separate unrelated images. Preserve the exact counts of figures, portraits, and equipment swatches. Do not add extra characters. Do not use bright colors or cheerful lighting. Keep all text as small dossier typography except for the large red Chinese name.

Create a cinematic wide 16:9 anime-style nighttime cityscape showing a full “advertising takeover” of a futuristic Tokyo-like entertainment district by one idol character, {argument name="character name" default="Hikari"}. View from a high pedestrian balcony in the foreground, looking down over a rainy neon plaza packed with tiny pedestrians, wet reflective streets, glass buildings, trains, skybridges, and dense skyscrapers under a dark blue evening sky. In the foreground, place the same character seen from behind at the railing: a young anime girl with short brown hair, a dark beret, cream blouse, burgundy jumper dress with ribbon bow and subtle gold embroidery, looking out at the city. Fill the city with exactly 18 visible advertisements featuring the same girl in different poses and crops: 1) giant top-left billboard reading “HIKARI / FASHION,” 2) curved mid-left screen reading “AOZORA / BANK,” 3) lower-left orange cafe ad reading “CAFÉ PHOENIX,” 4) slim vertical travel ad reading “WanderSee / TRAVEL,” 5) large central billboard reading “LUMINOUS / BEAUTY CLINIC,” 6) central appliance billboard reading “BrightHome / LIFE APPLIANCE,” 7) small high billboard reading “SKY LEARN,” 8) large upper-right purple gaming ad reading “Phoenix Gear / GAMING,” 9) right-side blue technology billboard reading “FUTURE VISION / TECHNOLOGY,” 10) center-right transit ad reading “HIKARI LINE,” 11) narrow vertical sign reading “Smile / DENTAL,” 12) small vertical character sign beneath the central screens, 13) small character billboard near the lower center intersection, 14) small character sign on a mid-right building corner, 15) long horizontal station-front banner with repeated Hikari portraits, 16) train-side wrap ad labeled “HIKARI LINE,” 17) another train or platform wrap with repeated portraits in the lower-right transit area, and 18) small rooftop or facade screen near the center-right showing her portrait. Make every ad visually consistent but branded differently, with clean white sans-serif English text, glossy LED glow, and no extra unreadable copy. Use rich cyberpunk-commercial lighting, rain reflections, warm shop windows, blue and pink neon, highly detailed architecture, and a grand sense of scale. Some distant character faces on billboards may be softly blurred by glow and distance, but the overall impression must be clear: one beloved character has taken over the entire city’s advertising landscape. Use {argument name="primary outfit color" default="burgundy"}, {argument name="hair color" default="brown"}, {argument name="city mood" default="rainy neon night"}, and {argument name="main brand theme" default="idol fashion and lifestyle"}.

Create a clean horizontal circle or software brand logo on a pure white background, centered with generous empty space. The main wordmark reads {argument name="brand name" default="PastelSoft"} in an elegant thin high-contrast serif typeface, with very wide letter spacing, dark navy blue lettering, and a refined premium feel. Place the text across the middle of a wide landscape canvas, approximately 1200 by 515 pixels. Under the wordmark, draw one single delicate flowing wave line in pale cyan-blue, starting very thin at the far left, gently rising and dipping beneath the letters, then tapering toward the right. At the far right above the wave, add exactly one small dolphin silhouette in a soft pastel aqua gradient, angled as if leaping to the right, positioned just after the final letters. Use a minimal airy composition with no border, no shadows, no additional icons, no extra text, and no background pattern. The overall mood should be calm, oceanic, pastel, lightweight, and professional.

Create a full-body anime-style character illustration on a plain white background: a curvy young female idol dressed in a pirate-inspired stage costume, standing front-facing in a confident fashion pose with one hip slightly tilted and arms relaxed at her sides. She has long flowing {argument name="hair color" default="blonde"} hair falling past her shoulders and wears a large turquoise-blue pirate tricorn hat with white trim and a small white skull-and-crossbones emblem centered near the front. Her face should be left blank/obscured or minimally detailed, while the outfit is the focus. The costume consists of exactly 7 main visible clothing elements: 1) oversized turquoise pirate hat with skull emblem, 2) white sleeveless ruffled blouse with high collar and front frills, 3) black ribbon bow tie with a small gold brooch at the neck, 4) turquoise corset waist cincher with black crisscross lacing, 5) layered short turquoise idol skirt with two flouncy tiers, 6) black ruffled underskirt visible beneath the blue skirt, and 7) long turquoise split back tailcoat panels hanging behind the skirt. Add exactly 2 black opera-length gloves with small lace-up ribbon details near the upper arms, and exactly 2 dark brown thigh-high lace-up heeled boots with glossy highlights and pointed heels. Use clean cel-shaded anime rendering, crisp line art, bright turquoise accents, glossy leather shading on gloves and boots, soft skin highlights, and a centered character-sheet composition with no background props, no text, no watermark, and no extra accessories beyond the described pirate idol outfit. Overall mood: glamorous live-performance pirate idol costume design in teal, white, black, and dark brown.

Goal: Create a vertical anime storyboard/comic page in a soft niji-style watercolor look, showing a Japanese high school classroom introduction scene with cinematic framing and hand-drawn ink borders.

Canvas: Tall vertical page, roughly 2:3 aspect ratio, divided into exactly 4 stacked horizontal panels separated by white gutters. Each panel has a thin, slightly imperfect black hand-drawn border. Add small black uppercase labels at the top-left of each panel: “SHOT 1”, “SHOT 2”, “SHOT 3”, and “SHOT 4”.

Visual style: Warm nostalgic anime film still, delicate sketchy line art, muted watercolor shading, painterly sunlight, subtle paper texture, detailed classroom perspective, navy school uniforms, soft shadows, no photorealism. Use a style reminiscent of modern Japanese animation and niji illustration.

Characters and counts: The main character is {argument name="character name" default="a quiet transfer student"}, a teenage Japanese girl with {argument name="hair color" default="long straight black hair"}, wearing a navy sailor school uniform with a red neckerchief in the first three shots. Include exactly 4 panels, exactly 1 main girl, exactly 1 male teacher in shots 1 and 2, exactly 26 seated anonymous gray classmate silhouettes in shot 1, exactly 4 uniformed girls in shot 3, and exactly 3 cropped uniform torsos in shot 4.

Shot 1: Wide establishing view from the back of the classroom. The main girl stands at the front beside a tall male teacher in a dark suit. Behind them is a large green chalkboard, a wall clock, notices, and bright windows on both sides. Rows of wooden desks and chairs fill the room. Show 26 seated classmates as simplified gray silhouettes facing forward. Morning sunlight streams through tall windows, casting warm rectangles of light on desks and floor.

Shot 2: Medium close-up of the main girl standing in front of the chalkboard, centered and large, with backpack straps visible over her shoulders. The male teacher appears smaller in the background at the right. A round classroom clock is visible high on the right wall. The girl’s face should be intentionally soft, indistinct, or blurred rather than detailed, while the uniform, hair, and classroom remain clear.

Shot 3: Social scene near a window inside the classroom. The main girl sits at a desk by the bright window while 3 other sailor-uniform girls stand around her, leaning in as if greeting or questioning her. Count exactly 4 girls total in this panel: 1 seated main girl and 3 standing classmates. Include bookshelves and desks in the background, leafy green trees visible outside, and warm sunlight on the desk surface.

Shot 4: Cropped close-up of exactly 3 girls from shoulders to waist, showing navy sailor uniforms with pale blue neckerchiefs and collars. Their faces are cut off by the top edge of the frame. Emphasize folds in the dark fabric, ribbon shapes, and warm classroom light. Background should be minimal and softly suggested.

Constraints: Keep all text limited to the four shot labels only. Do not add speech bubbles, captions, logos, or watermarks. Maintain consistent classroom setting, warm daylight, and hand-painted anime storyboard aesthetics throughout.

Goal: Create a wide comparison preview image showing exactly three proposed Japanese cover designs for a service called {argument name="service name" default="note特化版GPT"}, with a clean cream-and-teal aesthetic for promoting a GPT that analyzes note articles and creates thumbnails, diagrams, and illustrations.

Canvas: Horizontal 2.1:1 collage, about 1200×630 px, off-white background. Divide the canvas into exactly three equal vertical panels with thin pale separators and generous margins. Add a bottom label strip under the panels with three centered proposal labels: “A案”, “B案”, and “C案” in teal.

Panel A, left proposal: Minimal elegant cover design on an ivory paper background with soft watercolor teal wash at the bottom right and leafy botanical sprigs on the left and lower right. At top center place a rounded pill label reading “note特化版GPT” with a tiny pen icon to the right. Large Japanese headline centered, mixing dark charcoal and teal emphasis: “あなたの note記事を 分析して、 サムネ・図解・挿し絵を作るGPT”. Add a thin divider line with a small dot beneath the headline. Small subcopy near bottom: “3案で比較できるから、あなたの記事にぴったりの一枚が見つかる”. Refined editorial typography, calm and spacious.

Panel B, middle proposal: Warm lifestyle photo-style cover with a softly lit woman working at a desk near a window; laptop on the right, notebook with thumbnail sketches in the foreground, hand holding a pen, coffee cup and small stationery details. The woman’s face should be intentionally obscured with a simple square blur/mosaic. Overlay Japanese headline on the left in bold dark charcoal and teal: “あなたの note記事を 分析して、 サムネ・図解・挿し絵を作るGPT”. Add handwritten text near top left: “書くことに もっと、集中できる。” Add a circular teal badge near upper right reading “note 特化版 GPT”. Add a teal rounded banner along the bottom with a small icon and white text: “執筆に集中できる、あなたの味方。” Natural beige lighting, cozy creator workspace atmosphere.

Panel C, right proposal: Information-rich infographic cover on a pale mint background with rounded white card. Top pill label: “note特化版GPT”. Large centered headline: “あなたのnote記事を 分析して、 サムネ・図解・挿し絵を作るGPT”. Under it, a line of copy: “3案で比較できるから、理想の一枚が見つかる！” Create exactly four step cards across the middle, connected by small arrows: 1) “記事を分析” with document and magnifying glass icon, subtext “内容や目的を読み解きます”; 2) “3案を提案” with three thumbnail cards labeled A, B, C, subtext “テイストの違う3案を比較”; 3) “選んで使える” with a large checkmark icon, subtext “ぴったりの1枚を選ぶだけ”; 4) “執筆に集中！” with pen writing icon, subtext “デザイン時間をぐっと時短”. Below, create exactly four feature cards: 1) “サムネイル” with landscape thumbnail icon; 2) “図解・まとめ画像” with charts and layout blocks; 3) “挿し絵・イラスト” with simple female portrait; 4) “世界観づくり” with moodboard tiles. Add a teal footer ribbon with sparkle icons and text: “アップグレード追加料金なしで、ずっと使える！” plus a small heart outline at the right.

Visual style: Consistent teal, mint, ivory, charcoal, and beige palette. Japanese typography should look polished and readable, combining modern sans-serif with softer serif accents. Keep the three proposals visually distinct: A is botanical minimalist, B is warm lifestyle/photo, C is structured infographic. No watermarks, no social media UI, no extra panels, no English replacement text.

Goal: Create a clean Japanese presentation-style infographic showing how annual training seminar posters were batch-created with Codex and Image 2.0.

Canvas: 16:9 horizontal slide, off-white background, dark navy and muted gold accent colors, modern corporate healthcare aesthetic, generous margins, subtle shadows, thin divider lines, crisp vector icons.

Header: Large bold navy Japanese headline: {argument name="headline text" default="今年度の研修会ポスターをCodexで一括作成した"}. Add a thin vertical gold line to the left of the headline. Under it, a smaller gray subtitle: {argument name="subtitle text" default="情報をまとめてプロジェクト化し、Image 2.0でデザイン指定。全部まとめて生成。"}. Place a thin horizontal divider below the subtitle.

Top process section: Show exactly 3 numbered steps in one row, connected left-to-right by two gold arrow lines. Each step has a navy circular number badge, a navy title, a simple line icon, and short Japanese explanatory text.
1. Step 1 title: 「情報をフォルダにまとめる」 with a folder icon. Body says the information for each seminar is gathered into one folder, with three bullet items: date/time/place, seminar title/division, speaker/lecture title; then says this becomes project material.
2. Step 2 title: 「プロジェクト化して指示する」 with a checklist document icon. Body explains turning the folder into a Codex project and instructing it to create seminar posters using the information, including design directions such as colors, layout, and mood.
3. Step 3 title: 「Image 2.0で一括生成」 with a monitor icon containing sparkle marks. Body explains that all seminar posters are generated together in a unified design, no need to make them one by one, and small corrections can be handled with additional instructions.

Bottom poster gallery: Display exactly 4 vertical seminar poster cards in a single row, evenly spaced, each with rounded corners, soft drop shadow, white body, navy header band, gold icons, gold botanical line decoration at the bottom right, and a gold horizontal line near the bottom. Each card has four information rows with icons and labels: 「日時」, 「場所」, 「演者」, 「講演タイトル」.

Poster card 1: Header text 「第1回 医療安全研修会」 with a shield-and-cross icon. Date row: 「2025年6月20日（金） 18:00～19:30」. Place row: 「本館 3階 大会議室」. Speaker row: 「医療安全管理室　山田 太郎」. Lecture title row: 「インシデントを防ぐための実践的アプローチ」.

Poster card 2: Header text 「第2回 感染対策研修会」 with a virus/network icon. Date row: 「2025年7月18日（金） 18:00～19:30」. Place row: 「本館 3階 大会議室」. Speaker row: 「感染管理認定看護師　佐藤 花子」. Lecture title row: 「標準予防策の再確認と実践のポイント」.

Poster card 3: Header text 「第3回 放射線安全研修会」 with a radiation symbol icon. Date row: 「2025年8月22日（金） 18:00～19:30」. Place row: 「本館 3階 大会議室」. Speaker row: 「診療放射線技師　鈴木 一郎」. Lecture title row: 「放射線被ばくの最小化と防護の実践」.

Poster card 4: Header text 「第4回 接遇研修会」 with a handshake icon. Date row: 「2025年9月19日（金） 18:00～19:30」. Place row: 「本館 3階 大会議室」. Speaker row: 「接遇担当　高橋 美咲」. Lecture title row: 「信頼される医療者になるためのコミュニケーションの基本」.

Footer: Center a small gray Japanese summary line below the cards: {argument name="footer text" default="すべて統一デザイン・一括生成・1枚ずつ作る手間がゼロに"}.

Visual style: Use dark navy for headings and card headers, muted gold for accents/icons/arrows, light gray for dividers and secondary text. Keep typography highly legible, Japanese sans-serif, with the headline bold and all body text aligned neatly. Maintain a polished autogenerated-report look with no photos, no people, no watermark, and no extra poster cards or extra steps.

Goal: Create a vertical Japanese romance manga page in glossy AI-anime style about a goth-lolita girl making whipped cream in a luxurious kitchen while her blue-haired boyfriend reminds her it is Lovers' Day and gives her a heart pendant.

Canvas: Tall smartphone manga page, 9:16 aspect ratio, full-bleed art with thin black panel gutters. Use high-detail lighting, cinematic depth, ornate European kitchen background with copper pots, shelves, tall arched sunlit windows, strawberries, metal mixing bowls, and splashes of white whipped cream.

Characters: Female lead is {argument name="female character name" default="a pale goth-lolita girl"}: long ash-blonde hair, black frilled maid/lolita headband, red-pink eyes, black lace choker, black off-shoulder frilly dress, small pink heart earrings, expressive blush. Male lead is {argument name="male character name" default="Alfreed"}: messy vivid blue hair, cool boyfriend vibe, white T-shirt, black leather jacket, dark pants, wearing or holding a silver necklace with a blue heart pendant. Keep faces attractive anime style.

Layout: Use exactly 7 distinct comic panels. Panel 1 is a wide top panel showing the girl vigorously whisking whipped cream in a metal bowl on the left while the boyfriend stands in the sunlit kitchen on the right. Panel 2 is a left middle panel close-up of the girl whisking with intense focus. Panel 3 is a right middle close-up of the boyfriend speaking with his hand near his mouth. Panel 4 is a lower-left romantic close-up of the boyfriend pointing upward with sparkling pink background. Panel 5 is a lower-right close-up of the shocked girl with wide red eyes, flushed cheeks, open mouth, and whipped cream splashing. Panel 6 is a small bottom-left inset showing a blue heart pendant necklace sparkling in close-up. Panel 7 is a large bottom panel showing the boyfriend standing close behind the girl and putting the blue heart pendant necklace around her neck while she blushes shyly in front of the whipped cream bowl.

Text content: Include exactly 7 speech bubbles with Japanese text: 1) top panel, boyfriend bubble: {argument name="calling text" default="アルフリード〜♪"}; 2) middle-right panel: {argument name="question text" default="今日は何の日か知ってる?"}; 3) middle-left panel, girl bubble: "知らないわよ？ゲームの発売日とかぁ？"; 4) lower-left panel, boyfriend bubble: {argument name="lovers day line" default="ざんねーん！恋人の日なんだってさー♡"}; 5) lower-right panel, large jagged reaction bubble: "はぁ!?"; 6) bottom panel, boyfriend bubble: "僕にはアルフリードが居るから恋人は要らないけどね♪と言うわけで、感謝の気持ち♪"; 7) bottom-left/bottom panel small bubble: "………".

Sound effects: Add energetic Japanese manga SFX around the action: whisking sounds near the bowl such as "ブィーン", "ポワッ", "ガシガシガシ", cream splash sounds like "ブゥウウウッ" and "ブッガッ!!", sparkle around the pendant "キラッ", heartbeat text near the bottom "ドキッ…". Use hand-drawn black or pink manga lettering integrated into the panels.

Visual style: Highly polished modern Japanese web manga, dramatic lighting, sharp linework, saturated eyes, soft bloom, reflective metal, detailed whipped cream foam and splashes. Romantic comedy tone with exaggerated expressions and sparkly shoujo effects. Preserve clean panel readability despite dynamic action.

Constraints: Do not add extra panels, extra characters, watermarks, logos, or English dialogue. Keep the necklace pendant visibly blue and heart-shaped. Keep the kitchen background consistent across panels.

Create a richly detailed vertical 3:4 fantasy-realistic landscape painting of {argument name="location" default="Cappadocia, Turkey"} at twilight, seen from a high panoramic viewpoint over a vast valley of pale volcanic rock formations and cave dwellings. The lower half is filled with fairy-chimney stone towers, carved rock houses, arched doorways, terraces, stairways, small trees, and warm golden window lights beginning to glow throughout the town; the left foreground has larger cave-house facades and balconies, while the middle and right distance recede into a maze of illuminated rock villages, ridges, and distant mountains. The upper half is a dramatic sunset sky with deep cobalt blue at the top, scattered textured clouds, and glowing gold, peach, pink, and violet light near the horizon. Across the sky, include exactly 10 flying geese in a loose diagonal migration formation: 9 gray-brown geese with dark wings and 1 bright white goose at the lower right of the flock; add a tiny hidden small figure beside the white goose, subtle but discoverable. Use {argument name="sky color palette" default="deep blue, violet, peach, and golden sunset"} and {argument name="town lighting" default="warm amber lantern and window lights"}. Style should be highly detailed, painterly yet luminous, cinematic, whimsical, serene, and awe-inspiring, with crisp architecture, atmospheric depth, no text, no border, no watermark.

Using REFERENCE_0 as the character base, transform it into a vertical smartphone screenshot of a Japanese AI manga viewer page. Keep the same pale-haired researcher character and ornate white-and-gold outfit design, but redraw him in a softer serialized manga style and place him into a multi-panel comic layout rather than a single illustration.

Canvas and framing: Create a tall mobile-screen composition with black app UI bars at the top and bottom, as if viewed on a phone. Add a status bar showing the time {argument name="phone time" default="17:53"}, signal/Wi‑Fi/battery icons, and a manga viewer header with a back arrow, the title {argument name="manga title" default="内気な研究者⑫(終)(1/1)"}, a menu icon, and a refresh icon. Keep the comic page beneath the header.

Comic layout: Use exactly 4 visible manga panels:
1. A large dominant close-up panel of the pale-haired researcher, adapted from the reference, with his face partially obscured/blanked as in the reference and a bright dreamy bokeh laboratory background.
2. A small diagonal inset panel at the upper right showing only the lower half of the researcher’s smiling face and collar.
3. A middle-right diagonal panel set at a bar or laboratory counter, adding a brown-haired ponytail woman in a blouse and dark vest, speaking to the researcher.
4. A wide bottom panel at the bar showing the researcher seated beside the same woman, with warm shelves, bottles, and glassware behind them.

Text and dialogue: Add exactly 4 text elements inside the comic page: one vertical narration box reading {argument name="narration text" default="ひそかに気になってしまうラグネスなのであった"}, one speech bubble in the middle panel reading {argument name="woman line" default="いや〜それにしても…"}, and two speech bubbles in the bottom panel reading {argument name="bottom dialogue" default="あのトキカゲとはドキドキしたわね / その話、詳しく"}. Preserve the Japanese text orientation and manga bubble style.

Style: Polished webtoon/manga page, warm amber laboratory-bar lighting, diagonal panel borders, soft glow, sparkles and bokeh, delicate linework, cinematic romance-drama tone. Add two circular navigation buttons with left and right chevrons near the lower center over the comic page. Do not add extra panels, extra characters, watermarks, or additional text.