GPT Image 2 prompts

Create a soft pastel anime-style illustration of three young women sitting close together in a dreamy pink bedroom, framed as a waist-up group portrait in a 3:2 horizontal canvas. The scene has warm morning backlight from a window on the left, sheer curtains, a vase with pink tulips on the windowsill, a bed or sofa with lavender bedding, and a softly blurred wall with small cute framed pictures and posters on the right. Show exactly three girls, all with very short hair variations: the left girl is completely bald with a smooth shaved head, the center girl has an ultra-short brown buzz cut with visible stubble texture, and the right girl has a short dark pixie/buzz cut with spiky texture. Their faces are covered by flat pale pink rectangular censor blocks, one rectangle per girl, matching the soft color palette. Outfit details: the left girl wears a pale pink ribbed cardigan over a white lace camisole with small pink bows, a muted plaid skirt, and dangling flower earrings while touching her ear; the center girl wears a cream cable-knit cardigan with buttons and tiny bows over a white frilled blouse tied with a pale pink ribbon; the right girl wears a white blouse with a frilled collar and black ribbon bow under a pink plaid pinafore dress, with a small dangling earring. Use delicate linework, soft watercolor shading, luminous highlights, blush-pink atmosphere, cozy feminine fashion, high-detail fabric textures, and gentle depth of field. Keep the mood intimate, cute, and serene. Customize the three hair styles as {argument name="hair styles" default="left bald, center ultra-short brown buzz cut, right short dark pixie buzz cut"}, the setting as {argument name="bedroom setting" default="soft pink bedroom with tulips, sheer curtains, lavender bedding, and cute wall pictures"}, the outfit palette as {argument name="outfit palette" default="pale pink, cream, white, lavender, and muted plaid"}, the face covering as {argument name="face covering" default="flat pale pink rectangular censor blocks over each face"}, and the lighting as {argument name="lighting" default="warm hazy morning backlight from the left"}.

Goal: Create a vertical pixel-art fantasy RPG character sheet for {argument name="character name" default="ウルフくん"}, a slim dark-tanned wolf-eared traveling boy in a tropical southern-island setting, with Japanese game UI panels and a bright 16-bit/32-bit JRPG aesthetic.

Canvas: Tall 9:16 mobile poster, approximately 675×1200. Use crisp high-detail pixel art with glowing UI borders, saturated tropical colors, warm sunset light, and a beach-resort background with palm trees, blue sky, ocean, wooden deck, fruit, and a blue drink on a table.

Main character: Center-right full-body anime boy, slender and lightly toned, not muscular, tan skin, long messy {argument name="hair color" default="dark navy blue"} hair, large gray-brown wolf ears, golden eyes, wolf tail, playful confident pose. His face is mostly covered by a centered opaque dark blurred square censor block. Outfit is a blue-and-white open tropical dancer/traveler costume with a bare torso, gold bangles, layered necklaces, blue gemstone pendant, waist chains, sash, short skirt-like wraps, shell ornaments, ankle wraps, and small white rabbit charms. One hand reaches toward the viewer; the other holds a round gold-framed tambourine with a white rabbit silhouette.

Layout: Top-left status card, four stacked portrait cards down the left side, large main character occupying the right and center, two bottom information panels, and two small mascot/chibi figures along the bottom.

Top-left status panel text: Use a translucent navy glass RPG window with icy glowing corners. Include exactly these lines in Japanese: 「【南国ゆるふわ旅芸人】」, large name 「{argument name="character name" default="ウルフくん"}」, 「ジョブ：旅芸人」, 「Lv：27」, 「HP：248／248」 with a green HP bar, and 「MP：25／25」 with a blue MP bar.

Left portrait column: Include exactly 4 square expression portraits of the same wolf-eared boy, each with a glowing blue frame: 1) calm golden-eyed neutral smile, 2) eyes closed laughing joyfully, 3) half-lidded slightly smug expression, 4) gentle confident smile.

Bottom-left belongings panel: Title it 「❖ そうび・もちもの ❖」. Include exactly 5 item rows with small pixel icons and Japanese labels/descriptions: 1) 「うさタムリン」, a gold tambourine with rabbit silhouette; 2) 「うさみみペンダント」, a white rabbit-head pendant; 3) 「うさぴょんブレスレット」, a bracelet with bunny charm and sparkles; 4) 「うさぎポシェット」, a beige rabbit-shaped pouch; 5) 「ミニオトモ：モフうさ」, a fluffy white rabbit companion icon.

Bottom-right skills panel: Title it 「❖ とくぎ（スキル）❖」. Include exactly 4 skill rows with icons and Japanese labels/descriptions: 1) musical note icon, 「ゆるふわビート」; 2) bunny icon, 「うさぴょんダンス」; 3) yellow sparkle icon, 「南国スマイル」; 4) ZZZ sleep icon, 「サマー・ゴロ〜ン」.

Bottom mascots: Include exactly 2 small figures at the bottom center/right: 1) a chibi version of the wolf boy holding the tambourine, with wolf ears and tail; 2) a cute fluffy white rabbit companion wearing a blue flower collar and small gold bell.

Visual style and constraints: Japanese fantasy gacha-game character profile, pixel-art sprite detail, luminous icy-blue UI frames, tropical resort lighting, small pink flower petals drifting, no extra characters, no extra panels, no modern UI outside the RPG windows. Keep the boy slim and youthful; avoid a bodybuilder or bulky muscular build.

Goal: Create a bright square Japanese social media campaign banner advertising an MNP switching promotion for {argument name="campaign type" default="MNP limited campaign"}.

Canvas: 1:1 square banner, 768×768 style, high-resolution, clean commercial ad design. Use a sky-blue radial burst background with pale rays from the center, dotted halftone texture near the edges, small sparkle stars, and about 28 colorful confetti pieces in pink, yellow, and blue around the border.

Layout: Arrange the design in 5 stacked zones. Zone 1 is a dark royal-blue ribbon at the top with rounded rectangular center and pointed ribbon tails. Zone 2 is a large headline area with bold Japanese text. Zone 3 is the main offer amount, oversized and centered. Zone 4 is a white rounded information card with icons. Zone 5 is a large blue rounded call-to-action button at the bottom.

Text content: Top ribbon text: {argument name="top ribbon text" default="MNP限定キャンペーン"}. Main headline: {argument name="headline text" default="乗り換えで"}. Main offer text must read {argument name="offer amount text" default="最大2万円還元"}, with the number and currency in bright yellow and the surrounding characters in deep blue. Information card text: {argument name="detail text" default="対象プラン申込で ポイント還元"}. CTA button text: "今すぐチェック". Small disclaimer centered below the button: "※表示はダミーです".

Counted visual elements: Include exactly 1 top ribbon banner, 1 main headline, 1 oversized offer amount block, 1 white rounded information card, 1 CTA button, 1 smartphone icon on the lower left, 1 yellow circular gift icon near the center of the card, 1 group of gold coins on the right of the card, 1 white play-arrow circle inside the CTA button, and 2 yellow attention-burst marks flanking the CTA button. In the coin group, show exactly 3 stacked coin columns plus 1 large front coin marked with a capital "P".

Visual style: Bold Japanese advertising typography, thick white outlines, blue drop shadows, glossy gradients, high contrast, celebratory telecom-promotion look. Use deep navy blue, royal blue, cyan, white, and vivid yellow as the primary palette. The main "2万円" should dominate the composition, with heavy outline layering: yellow fill, blue stroke, thin white outer glow, and soft shadow.

Subject details: The smartphone should be tilted slightly left, with a cyan screen and a simple white two-person user icon. The gift icon should be a dark blue present symbol inside a yellow circle with tiny yellow sparkle marks above it. The CTA button should be a glossy rounded rectangle with a white border and soft shadow, containing a white circular play icon on the right.

Constraints: Keep all text crisp and readable, centered, and in Japanese. Do not add logos, brand names, watermarks, QR codes, or extra copy. Maintain a polished Japanese mobile-carrier campaign banner aesthetic.

Goal: Create a bright anime cheerleader pose reference sheet for a sports-stadium performance, showing {argument name="character name" default="a silver-haired anime cheerleader"} demonstrating exactly 15 cheer motions in separate numbered panels.

Canvas: Wide 16:9 contact-sheet layout, 1280x720 style, divided into a clean 5-column by 3-row grid. Each panel has a thin white border, a blue circular number badge in the top-left, a white slanted title banner across the top with bold blue uppercase English text, and a dark translucent caption bar along the bottom with Japanese explanatory text.

Setting and visual style: Sunny outdoor football stadium with green field, running track, curved grandstands, vivid blue sky, fluffy clouds, energetic idol-cheer atmosphere. High-quality anime illustration, glossy promotional look, dynamic wide-angle camera, bright rim light, saturated blue-and-white palette, crisp details, motion in hair and skirt. The character wears a blue-and-white cheerleader outfit: cropped sleeveless top with star motif, pleated mini skirt, blue ribbons, thigh garter accents, ankle shoes, and large blue-and-white pom-poms. Long flowing {argument name="hair color" default="silver-white"} twin-tail hair with blue bows, cute expressive idol style.

Layout and exact panel list: Include exactly 15 panels, ordered left to right, top to bottom:
1. HIGH V — both arms raised high in a V shape with pom-poms.
2. LOW V — both arms extended diagonally downward in a low V stance.
3. T MOTION — arms stretched straight sideways at shoulder height.
4. TOUCHDOWN — one arm raised high and the other extended diagonally downward.
5. HIGH TOUCH — one hand reaches straight upward as if touching high, one leg lifted playfully.
6. LOW TOUCH — leaning forward with one arm reaching low toward the ground.
7. HALF TIME — hands placed on hips, ready stance.
8. CLAP — pom-poms brought together in front, clapping rhythmically.
9. PUNCH — one arm thrust powerfully upward with pom-pom.
10. SPIN — spinning in place, skirt and hair flaring outward.
11. KICK — high leg kick with energetic balance.
12. LUNGE — one knee bent deeply in a strong lunge pose.
13. JUMP — jumping upward with pom-pom raised and legs tucked.
14. BOW — polite forward bow with pom-poms gathered in front.
15. FINISH POSE — final celebratory pose with one arm raised and smile.

Text content: Use the English motion titles exactly as listed. Add Japanese captions similar to a cheer instruction guide under each panel, including these meanings: basic high V position, calm low V stance, synchronized shoulder-height team motion, one hand high and one diagonal, touch high to appeal height, low touch with sparkle, hands on waist for preparation, clap pom-poms to make rhythm, punch upward with strength, spin for華やか movement, dynamic high kick, strong lunge, jump to express energy, graceful bow of thanks, and smiling finish pose. Main title is not needed.

Constraints: Keep the grid perfectly readable and evenly spaced. Use exactly 15 numbered panels and no extra panels. Maintain consistent character design, outfit, colors, and stadium background across all panels. Make the typography bold and legible. Avoid logos, watermarks, and unrelated objects.

Goal: Create an isometric character pose sheet showing the same faceless female presenter in many business presentation situations, suitable for a clean illustration library.

Canvas: Wide horizontal 16:9 canvas on a very light warm gray background, with generous spacing and soft oval gray shadows under each character. No borders, no title, no labels, no watermark.

Subject details: The repeated character is {argument name="character name" default="a young female business presenter"} with shoulder-length wavy {argument name="hair color" default="dark brown hair"}, no facial features, soft peach skin tone, wearing a beige oversized blazer, white inner top, loose cream wide-leg trousers, and white sneakers. Keep the same outfit and proportions across every pose. Use a soft modern isometric vector style, rounded shapes, muted pastel colors, minimal detail, clean edges, subtle depth, and a friendly corporate presentation mood.

Layout: Arrange exactly 14 separate full-body character poses in two loose rows of 7 across the canvas, each figure isolated and evenly spaced, all viewed from a consistent three-quarter isometric angle.

Visible poses, exactly 14:
1. Standing front-facing with both arms open as if explaining to an audience.
2. Standing with back turned, pointing at a floating presentation board filled with small pastel color swatches and image tiles.
3. Side view, placing colorful sticky notes onto a floating board.
4. Standing front view holding a large mood-board sheet with color blocks, image thumbnails, and a leaf motif.
5. Walking to the right while holding a black coffee cup.
6. Walking quickly to the right while carrying a stack of pastel papers or sticky notes.
7. Standing and gesturing conversationally with one hand while the other hand rests near the waist.
8. Standing with one index finger raised as if making an important point.
9. Standing with both arms raised in celebration, with small yellow emphasis marks above the head.
10. Standing in a thinking pose with one hand at the chin and the other arm folded.
11. Sitting cross-legged on the floor, writing or sketching in an open notebook or clipboard.
12. Crouching and arranging exactly 8 pastel sticky notes on the floor, with a small stack of notes in one hand.
13. Standing in a leaning presentation pose, one hand extended as if inviting discussion.
14. Standing with a black marker or pen raised in one hand and the other hand open in a presenting gesture.

Presentation props: Include only minimal presentation materials: floating white boards, pastel sticky notes, a notebook, coffee cup, pen, and paper sheets. Use {argument name="accent colors" default="soft yellow, pink, mint green, teal, beige, and muted brown"} for sticky notes and board graphics.

Constraints: Keep all characters faceless, do not add text, do not add extra people, do not change outfits between poses, keep the background clean and uncluttered, and maintain a cohesive isometric illustration set.

Goal: Create a clean isometric character pose sheet showing {argument name="character role" default="a professional female presenter"} in multiple presentation and business-meeting situations.

Canvas: Wide horizontal 16:9 canvas with a light gray-white background, no borders, no title text, no watermark. Use generous spacing so each vignette reads as a separate pose.

Visual style: Minimal modern isometric vector illustration, soft shadows under every figure and object, muted corporate palette, smooth flat shading, subtle gradients, crisp edges. The character should have fair skin, dark brown hair in a low bun, simplified faceless features, and wear {argument name="outfit" default="a white skirt suit with white high heels"}. Props should be light gray and white with small blue-gray chart accents.

Layout: Arrange exactly 12 discrete isometric vignettes in two rows of six, evenly spaced across the canvas. Keep the scale consistent and show the character from varied isometric angles.

Vignettes, exactly 12:
1. Front-facing woman standing and speaking into a handheld black microphone, one arm gesturing outward.
2. Rear-view woman standing behind a white podium with a gooseneck microphone.
3. Woman pointing at a transparent presentation board showing one pie chart and three vertical bar charts.
4. Woman standing with a dark tablet or clipboard held at chest height.
5. Woman walking while carrying a dark briefcase in one hand.
6. Woman walking quickly while holding a stack of folders and documents.
7. Rear three-quarter woman standing and gesturing to an unseen audience.
8. Woman holding a handheld microphone while presenting with an open-palmed gesture.
9. Woman standing confidently with one arm raised and index finger pointing upward.
10. Woman standing with one hand raised high as if asking or answering a question.
11. Woman seated at or standing behind a podium setup with a chair visible behind, speaking into a gooseneck microphone.
12. Woman leaning over a low table or desk with printed charts, reviewing or arranging presentation papers.

Additional presentation furniture and props: Include exactly 3 separate white podiums with black gooseneck microphones, 1 transparent chart board, 1 low desk/table with papers, 1 chair, 1 handheld microphone in the first vignette, 1 handheld microphone in the eighth vignette, 1 tablet/clipboard, 1 briefcase, and 1 stack of documents. Avoid any readable text or logos.

Goal: Create an isometric vector illustration set of the same {argument name="character type" default="professional businesswoman presenter"} in many presentation and office situations, arranged like a clean pose library on a white background.

Canvas: Wide horizontal 16:9 canvas, bright white or very light gray background, no borders, no title, no captions, no watermark. Use soft oval gray floor shadows under each figure.

Visual style: Polished modern isometric people illustration, semi-flat vector style with subtle gradients and soft shading. The character has a simplified faceless face, light skin, brown hair tied in a low bun, navy business suit, white blouse, cropped trousers, and black high heels. Keep proportions consistent across all poses and use the same outfit throughout.

Layout: Arrange exactly 13 separate full-body figures in two rows with generous spacing, like a presentation asset sheet. Top row contains 7 figures; bottom row contains 6 figures. All figures face slightly different isometric angles and remain isolated with no overlapping.

Pose count and details: Include exactly 13 discrete poses: 1) standing presenter facing forward with one hand open in a welcoming gesture, 2) rear-view presenter pointing with a stick at a wall presentation board showing a pie chart and lines, 3) standing presenter holding a small object or remote and gesturing to the side, 4) standing presenter holding a remote in one hand and gesturing with the other, 5) walking businesswoman carrying a gray folder under one arm, 6) walking businesswoman holding a spiral notebook or report, 7) standing businesswoman gesturing as if speaking to someone, 8) standing figure raising one index finger to make a point, 9) standing figure holding up a spiral report sheet with chart lines, 10) standing figure with arms crossed confidently, 11) seated figure on a gray chair using an open laptop while gesturing, 12) leaning forward and pointing to papers or charts on the floor, 13) standing figure leaning slightly with one hand on hip and the other index finger raised.

Subject customization: Keep the character recognizable as the same person in every pose, but allow the overall character to be changed to {argument name="character appearance" default="light-skinned woman with brown hair in a low bun"}. Use {argument name="outfit" default="dark navy suit, white blouse, black heels"}. Optional props should match a {argument name="presentation theme" default="corporate business presentation"}. Keep the illustration clean, minimal, and presentation-ready.

Create a dark cinematic medieval fantasy battlefield scene titled by mood as a confrontation with the enemy king. Three main figures stand in a tense triangular standoff amid smoke, ash, banners, soldiers, and battlefield debris. On the left stands {argument name="armored knight" default="a blonde female knight commander"}, facing slightly toward the center, wearing ornate polished silver plate armor with engraved patterns, a pale weathered cloak, chainmail details, leather belts, and a sheathed side weapon; she holds one long steel sword angled downward in her right hand. In the center stands {argument name="barbarian warrior" default="a tall muscular shirtless warrior with long wet black hair"}, bare-chested, scarred and muddy, wearing ragged black furs and leather straps around the waist, heavy boots, and holding one blood-darkened sword downward, his posture defiant and exhausted. On the right, seen mostly from behind and in three-quarter view, stands {argument name="enemy king" default="a towering dark king in black and gold armor"}, wearing a spiked golden-black crown, long black hair, imposing black plate armor covered in intricate gold filigree, and a heavy black cape embroidered with gold ornamentation; he faces the two warriors like a sovereign antagonist. Count exactly 3 primary characters: left silver knight, center shirtless warrior, right crowned enemy king. Include exactly 2 clearly visible drawn swords, both held downward by the left and center figures. The setting is a devastated medieval battlefield with ranks of blurred soldiers in armor behind them, tall spear tips, torn black banners bearing faded gold heraldic designs, smoky gray sky, drifting embers, mud, broken shields, helmets, and scattered weapons in the foreground. Use a gritty high-detail realistic style with dramatic low-angle composition, shallow depth of field, volumetric smoke, muted gray-brown color grading, metallic highlights, dark fantasy atmosphere, and epic movie poster lighting. No readable modern text, no clean studio background, no cartoon style.

Create a high-end studio product render of a full-body collectible anime-style figurine of {argument name="character name" default="an adult female school-uniform character"}, standing on a simple round black display base against a seamless warm gray background. The figure has a curvy stylized body, pale skin, and shoulder-length wavy {argument name="hair color" default="deep teal green"} hair with layered curled ends and glossy highlights; the face area is deliberately obscured by a soft square blur, as if anonymized. Pose her upright with legs crossed at the ankles, one hand resting behind her back and the other arm relaxed to the side with delicate fingers extended. The outfit has exactly 6 visible clothing pieces: a fitted light gray blazer with dark piping and one gold button, a white collared shirt, a bright red necktie, a short black pleated skirt with thin white trim stripes, sheer black pantyhose, and glossy dark brown loafers. Emphasize premium PVC figure craftsmanship: smooth painted surfaces, subtle seams, realistic fabric folds, glossy stockings, polished shoes, clean sculpted hair strands, and soft studio lighting. Use a centered front-facing composition, full figure visible from hair to base, vertical 3:4 aspect ratio, shallow shadow under the base, no text, no watermark, no extra props.

Using REFERENCE_0 as the character design sheet, transform the armored mecha girl concept into a realistic {argument name="figure scale" default="1/7 scale"} painted collectible figure product photo. Convert the black-and-white line art into a fully colored, high-quality PVC/anime figure with crisp molded armor panels, glossy and matte material separation, subtle seams, decals, and toy-like sculpted details. Keep the same core character design, dynamic airborne pose, helmet, sword, shield, cape, skirt armor, and mechanical leg armor, but render it as an actual physical figure instead of a concept sheet.

Color and finish: Use the reference sheet’s color notes as the basis: {argument name="main armor color" default="white and navy blue"}, with {argument name="accent colors" default="red, gold, gray, and light brown hair"}. Make the face, hair, armor, cape, shield, and sword look like a premium painted prototype.

Composition: Show a single full-body figure in a three-quarter front view, centered, with the sword raised behind her and the large shield forward on one arm. Add a simple studio setup with a neutral {argument name="background color" default="gray gradient"} background and a clear circular display base beneath the figure.

Style constraints: No concept-sheet panels, no annotation boxes, no Japanese labels, no extra turnaround drawings, no text, no watermark. Make it look like a professional product render/photo of a yet-unreleased anime mecha-girl figure.

Create a cinematic anime fantasy landscape illustration about departure and crossing mountains toward a new land. Use a wide 16:9 canvas. In the right foreground, show a lone young traveler viewed from behind and slightly in profile, sitting or crouching on a dark rocky cliff edge, gazing across a vast valley toward a distant crystalline castle city. The traveler wears a dark navy cloak or jacket, trousers, boots, and a wide-brimmed explorer hat; their face is softly obscured by shadow/blur. They carry one large rugged travel backpack with visible straps and buckles, with one rolled sleeping mat tied across the top, and hold one long walking staff topped with a small glowing crystal ornament. The landscape contains exactly three major depth zones: foreground rocky ledge with bright blue crystal clusters and pink alpine flowers, middle valley with a winding river and small waterfalls, and background snow-covered jagged mountains surrounding a luminous palace-like city with many tall spires. Add numerous floating translucent blue-violet crystals in the sky and embedded in the cliffs, sparkling with magical light. The sky is dreamy and expansive, with turquoise-blue upper atmosphere, stars and nebula-like streaks, and large peach-pink sunset clouds glowing behind the mountains. Lighting should be epic, soft, and radiant, with rim light on the traveler, volumetric mist in the valley, iridescent crystal reflections, high detail, painterly anime style, grand adventure mood, no text, no watermark. Optional customizable elements: make the destination a {argument name="destination" default="crystalline castle city"}, set the traveler as {argument name="character name" default="a lone young traveler"}, use {argument name="main color palette" default="sapphire blue, lavender, peach-pink, and icy white"}, include the theme {argument name="theme" default="departure toward a new land"}, and render in {argument name="visual style" default="cinematic high-detail anime fantasy illustration"}.

Create a vibrant tropical party fashion poster in a dynamic low-angle close-up composition. A young woman stands leaning back in front of a rusty corrugated metal shutter, wearing exactly 3 visible outfit pieces: an open turquoise Hawaiian shirt with large hibiscus flowers and green palm leaves, a white strapless bandeau crop top, and frayed white denim short shorts. Her face is intentionally covered by a plain centered square censor block in warm medium brown, while messy dark hair falls around it. She holds exactly 2 clear plastic cups of golden beer: one oversized cup thrust toward the camera in her left hand, strongly blurred by depth of field and motion, with a foamy white head; one smaller sharp cup in her right hand near her hip, overflowing with thick foam dripping down the side. Show a small belly-button piercing and a delicate necklace. Use dramatic flash photography, glossy skin highlights, strong shadows, shallow depth of field, energetic nightlife mood, and saturated summer colors. Surround the photo with a bold grunge graphic border of paint splatters, torn brush strokes, and diagonal shapes in turquoise, orange, black, yellow, cream, and neon green, framing the subject like a mixed-media collage. Add a small handwritten signature and date in the bottom-right corner reading {argument name="date text" default="May 26, 2026"} and {argument name="signature" default="Oyagi"}. Overall theme: {argument name="theme" default="tropical summer memory"}. Keep the image vertical, high-resolution, edgy, photorealistic with poster-art overlays, no extra people, no additional cups, no readable brand logos.

Goal: Create a whimsical cinematic snack advertisement poster for {argument name="brand and product" default="Oishi Potato Fries"}, themed as a magical fantasy amusement park called {argument name="headline text" default="SNACK KINGDOM"}.

Canvas: Vertical 2:3 poster composition, ultra-detailed, high-resolution, warm golden-hour lighting, saturated reds and yellows, energetic commercial photography mixed with fantasy illustration.

Main subject: Center a smiling young woman with glasses wearing a {argument name="hoodie color" default="pink"} hoodie, seated inside an oversized red-and-yellow bucket filled with golden potato fries. Her face should be intentionally covered by a plain rectangular neutral-color placeholder, while her dark hair, ears, hands, and hoodie remain visible. She holds exactly 2 large fries, one upright in each hand.

Product and foreground: The giant bucket dominates the lower half of the image. On the bucket, show the readable logo text “Oishi” above large bold text “POTATO FRIES,” with small crown and star graphics. Fill the bucket with many thick golden fries, and add flying fries around the frame. Include exactly 14 prominent loose fries outside or above the bucket: 2 at the top left, 1 at the top right, 3 on the right side, 3 in the lower foreground, 3 around the left foreground, and 2 partly behind the subject near the bucket rim. Add glowing golden light trails swirling around the bucket and fries.

Background: Build a miniature fantasy theme park snack kingdom behind the subject, with a fairytale castle in the distance, a Ferris wheel on the right, colorful carnival tents and rides, warm lantern lights, tiny crowds, and a visible shop sign reading “SNACK CAFE.” Use shallow depth of field so the subject and bucket are sharp while the amusement park is slightly dreamy and cinematic.

Typography: At the top, place huge 3D cream-colored block letters spelling “SNACK KINGDOM,” outlined in brown with a red curved underline, and place a gold crown with a red jewel above the title. At the bottom, place a wooden sign with the slogan in bold white and yellow text: “MORE THAN A TREAT. IT’S JOY.”

Style constraints: Make it feel like a viral snack brand campaign poster: playful, magical, high-energy, appetizing, glossy, and richly lit. Keep all main text perfectly legible, avoid extra logos or watermarks, and preserve the exact visible text hierarchy.

Create a vibrant full-body anime-style mixed-media illustration of {argument name="character theme" default="twins"} standing side by side on a vertical poster canvas. The scene features exactly 2 teenage twin girls with short wavy dark-brown bob hair, similar hairstyles, and their faces intentionally covered by soft rectangular beige blur blocks. The left twin stands casually with one hand in her pocket and her other arm raised high making a peace sign; she wears a bright oversized color-block windbreaker in orange, turquoise, yellow, and white over a white shirt, loose white paint-splattered pants, layered necklaces, yellow socks, and orange high-top sneakers. The right twin stands shyly with one finger near her chin while hugging exactly 2 books to her chest; she wears a crisp white blouse, blue denim-style pinafore straps or vest, a navy plaid pleated skirt, teal knee socks, and shiny black loafers. Surround them with an energetic abstract collage background on warm off-white paper: torn-paper shapes, graph paper scraps, thick cobalt-blue paint swaths, pink and yellow triangles, turquoise blocks, orange circles, black ink splatters, hand-drawn scribbles, and painterly brush strokes. Include exactly 5 prominent accessory objects around the twins: 1 yellow skateboard with pink wheels leaning on the lower left, 1 stack of several books on the lower right, 1 glass laboratory flask filled with blue liquid above the books, 1 cluster of small white and blue flowers with green leaves on the upper right, and 1 large blue flower near the lower right. Use saturated pop colors, crisp anime linework, watercolor and gouache textures, fashion-illustration proportions, high detail in clothing folds and splatter paint, and a cheerful youthful mood. Add a small handwritten signature and date {argument name="date text" default="May 26, 2026"} in the bottom-right corner. Keep the composition centered, full-body, poster-like, with no extra characters and no readable typography besides the date and signature.

Goal: Create a dense architectural infographic sheet for a historic Chinese circular temple reconstruction, combining precise blueprint annotations with a realistic cutaway construction model.

Canvas: Portrait 4:5 architectural presentation board on warm off-white paper, high resolution, clean margins, fine gray technical linework, subtle paper grain, no border or watermark.

Main subject: A large central 3D isometric render of the Temple of Heaven-style circular wooden hall under construction, labeled with a vertical hanging plaque reading {argument name="plaque text" default="祈年殿"}. The building has exactly 3 stacked dark blue-gray tiled roofs with upturned eaves, a gold spherical finial on top, red-brown wooden walls and columns, decorative blue-green-gold bands, and a right-side exposed bamboo/wood scaffolding framework showing the construction process. Place the building on a round white stone terrace with exactly 4 visible staircases. Add small construction activity around it: exactly 15 miniature workers in muted gray-brown clothing, 1 ox pulling a cart, 2 small handcarts, 1 wooden wheelbarrow, 1 stack of blue roof tiles, multiple timber beams and stone blocks, all scaled like an architectural model.

Layout: Use a balanced blueprint poster composition. The central 3D temple occupies the lower center and right half. Surround it with exactly 8 technical information panels: 1 title/project overview block at upper left, 1 top plan drawing at top center, 1 front elevation drawing at upper right, 1 total floor plan drawing at mid-left, 1 section drawing at mid-right, 1 structural exploded diagram at lower left, 2 node detail drawings at bottom left, and 1 material list column at right. Keep all diagrams monochrome pale gray like pencil/CAD drawings, with thin dimension lines, arrows, scale notes, and small Chinese annotations.

Text content: Large main title at upper left: {argument name="main title" default="天坛"}. Subtitle below it: {argument name="subtitle" default="建筑营造过程示意图"}. Project overview heading: preserve Chinese text {argument name="overview heading" default="工程概况"}, followed by small Chinese fields for name, location, year, building type, structure, and construction craft. Use scale labels such as “平面图 1:200”, “立面图 1:200”, “剖面图 1:200”, “总平面图 1:600”, “结构分解图”, and “节点大样图”. The right column is headed “主要材料表” and lists categories for wood, roof tiles, stone, and other materials. Keep the Chinese characters crisp, correctly formed, and readable.

Diagram details: The top plan panel shows a circular radial plan with concentric rings and dimension marks including values like 1000, 1900, 2800, and 3600. The total floor plan shows a smaller circular plan with square dimension frame and values around 3000, 5000, and 3080. The elevation and section panels show a three-tier temple silhouette with horizontal height markers around 0.000, 7.680, 15.590, 24.650, and 35.550, plus base width dimensions. The exploded structure panel shows a stacked pagoda-like skeleton with labels for roof, boards, tiles, beams, columns, and platform. The two node detail drawings show close-up roof eave joinery and column bracket structure, each marked 1:10.

Visual style: Hybrid of museum architectural drawing, Chinese heritage conservation blueprint, and realistic miniature construction diorama. Use restrained colors: ivory background, graphite gray drawings, dark slate-blue roof tiles, warm ochre scaffolding, red lacquer wood, gold finial, white stone terrace. Lighting is soft from upper left with gentle shadows under the model. Maintain a professional, information-dense technical poster look.

Constraints: Preserve exactly 8 surrounding technical information panels and exactly 3 roof tiers. Do not add modern machinery, English headings, people in modern safety helmets, logos, or decorative borders. Prioritize legible Chinese typography and clean diagram alignment.

Goal: Create a high-density architectural presentation board in Chinese showing the construction process diagram of an ancient wooden pagoda, centered on a realistic 3D cutaway/isometric model of {argument name="building name" default="应县木塔"} under construction.

Canvas: Portrait poster, off-white drafting-paper background, 4:5 ratio, clean museum-exhibition / architectural blueprint style. Use black Chinese typography, fine grey measurement lines, and sepia-brown timber materials. The image should feel like a technical architecture plate mixed with a miniature diorama.

Layout: Place a huge detailed octagonal timber pagoda model in the center, occupying most of the poster. Show the pagoda from a slightly elevated front angle so the roof structure and interior radial timber frame are visible. Surround the base with many tiny construction workers, scaffolds, timber piles, carts, ladders, cranes, and assembly frames. The surrounding drawings and text must stay aligned like a professional architectural sheet.

Text content: At top left, large title text: 「应县木塔」. Directly below, subtitle: 「建筑营造过程示意图」. Add two left-column information blocks titled 「工程概况」 and 「营造特点」 with multiple short Chinese lines describing project name, location, construction period, height, structure type, plan form, and craft characteristics. Keep the Chinese legible and not garbled.

Required diagram elements: Include exactly 7 major labeled diagram areas: 1) central 3D pagoda construction model, 2) left-column工程概况 text block, 3) left-column营造特点 text block, 4) lower-left 「平面图」 octagonal plan drawing with dimension marks around 30.27米, 5) upper-right 「立面图 1:400」 elevation drawing with floor labels and vertical dimensions, 6) mid-right 「剖面图 1:400」 section drawing with storey labels, 7) bottom strip of small technical detail panels labeled with categories such as 「斗拱详图」, 「构件解析」, 「结构分析」, and 「节点详图」.

Subject details: The pagoda should be an octagonal, multi-eaved Chinese timber tower with dark grey tiled roofs, exposed brown wooden beams, dense bracket sets, layered eaves, a central finial, and radial roof framing visible at the top. Add construction staging: unfinished upper interior trusses, scaffolding around lower levels, timber fences, and workers carrying beams. Use {argument name="primary material" default="wooden timber frame"} as the dominant construction material and {argument name="drawing scale" default="1:400"} for the elevation and section labels.

Visual style: Highly detailed architectural visualization, realistic miniature model rendering, precise technical annotation, fine ink lines, soft shadows, muted sepia and grey palette, no bright colors. Combine photorealistic 3D construction diorama with crisp 2D blueprint overlays. Make the typography resemble a formal Chinese architectural drawing sheet.

Constraints: Preserve the dense Chinese layout and technical drawing clarity. Do not add logos, watermarks, English labels, or modern buildings. Keep all diagrams spatially organized, with no overlapping text on the main pagoda. Ensure Chinese characters render correctly, especially the title {argument name="main title" default="应县木塔"} and subtitle {argument name="subtitle" default="建筑营造过程示意图"}.

Create a whimsical vertical snack advertisement for {argument name="brand name" default="Oishi"} Potato Fries in a magical fantasy amusement-park kingdom. The central subject is a smiling young woman wearing glasses and a pastel pink hoodie, seated inside an oversized red-and-yellow bucket overflowing with thick golden potato fries; her face should be visible and natural, and she holds one large fry in each hand. The bucket front prominently displays the product logo text: “Oishi POTATO FRIES” with a small crown motif and star accents. Above her, place a huge 3D title reading {argument name="headline text" default="SNACK KINGDOM"} in bold cream letters with brown shadowing, topped by a golden crown with a red jewel, and supported by a red curved banner underline. The background is a miniature fantasy theme park at sunset with a fairytale castle, glowing snack stalls, a visible “SNACK CAFE” sign, carousel-like rooftops, a Ferris wheel on the right, tiny crowds, warm lights, and cinematic depth of field. Add exactly 12 prominent flying or foreground fries around the composition, plus many fries inside the bucket, with golden seasoning texture. Wrap glowing golden light trails around the bucket and foreground for a magical energy effect. At the bottom, add a rustic wooden sign with the slogan {argument name="slogan text" default="MORE THAN A TREAT. IT’S JOY."}, using white text for the first line and bright yellow text for “IT’S JOY.” Use saturated warm colors, glossy commercial food-ad lighting, playful theme-park fantasy atmosphere, sharp product branding, realistic-yet-whimsical compositing, vertical poster aspect ratio 2:3, no watermark, no extra text.

Create a vertical high-octane snack advertisement poster for {argument name="brand name" default="Takis"}. A young woman with messy dark curly hair and glasses stands centered on a neon city street at night in a powerful action pose, wearing a black leather biker jacket, red crop top, ripped black jeans, belt chain, earrings, and layered necklaces. She thrusts one giant rolled spicy chip forward like a weapon, glowing red-orange and wrapped in flames. Surround her with a huge circular vortex of fire and flying ember sparks, plus exactly 9 visible Takis-style rolled chips: 1 giant chip in her hand, 1 large flaming chip at upper right, 1 small chip at upper left, 1 vertical chip left of the logo, 1 ring-shaped chip at left, 1 small chip near bottom left, 1 small chip near her right hip, 1 large diagonal chip at lower right, and 1 blurred oversized chip cropped in the bottom-right foreground. The background shows a gritty nighttime urban street with skyscrapers, purple neon accents, a food stand on the left with signs reading "SPICY FOOD" and "HOT", and a branded food truck or bus on the right displaying the Takis logo. At the top, add massive distressed fiery orange headline text reading {argument name="headline text" default="FIRE CRUNCH"}, with a bold white angled tagline beneath reading {argument name="tagline" default="MORE THAN SPICY. IT'S INTENSE."}. Place the yellow-and-purple Takis logo large under the tagline, centered above the character. Use extreme contrast, cinematic action-advertising lighting, glossy commercial detail, sparks, smoke, flame trails, saturated reds, oranges, purples, and blacks. Keep the composition dramatic and poster-like, with the character in the lower center and the typography dominating the upper third. Avoid extra logos, watermarks, or unrelated text.

Create a cinematic, high-energy vertical advertising poster for {argument name="brand name" default="Cheetos"} titled {argument name="headline text" default="CHEESY CHAOS"}. A young adult woman with a messy high bun and glasses, wearing a vivid orange bomber jacket over a black top and black tactical cargo pants with straps and zippers, is frozen mid-leap toward the camera in an extreme wide-angle action pose. Her nearest arm thrusts forward in strong perspective, holding one large flaming cheese puff between her fingers; the puff glows with sparks and small flames. Keep her face unobscured and expressive, but make the motion and hand perspective the main focus. Behind her, a massive orange cheese-dust explosion erupts down a busy downtown street, sending countless crunchy cheese puffs flying through the air. The city setting should feel like a stylized Times Square diner district at golden hour, with neon diner signs, traffic, pedestrians, storefronts, and a dramatic teal-blue sky. Include exactly 4 prominent readable sign/text areas: the giant top headline “CHEESY CHAOS” made from rough orange cheese-dust texture, a left diner marquee reading “CHEESY” with tiny menu-style text below, a right vertical neon sign reading “DINER” with a small “Cheetos” sign above and “Open” below, and the bottom tagline {argument name="tagline text" default="MORE THAN A SNACK. IT’S A MOOD."}. Add the {argument name="logo text" default="Cheetos"} logo centered at the bottom and a product bag in the lower-right corner. Use saturated orange cheese tones contrasted with deep teal shadows, explosive particles, flying snack debris, rim lighting, cinematic depth of field, motion blur, sharp commercial-photography detail, dramatic perspective, and a glossy blockbuster snack-ad aesthetic. Avoid extra brands, avoid extra slogans, avoid malformed text, and make the poster feel chaotic but professionally composed.

Goal: Create a high-quality cinematic vertical advertisement for {argument name="brand name" default="Ruffles"} ridged potato chips, showing an extreme sports fantasy scene where flavor and surfing collide.

Canvas: Vertical 2:3 poster, glossy commercial photography style, dramatic wide-angle perspective, sharp foreground, sunset tropical beach background, saturated teal-blue water and golden chip tones.

Layout: Top left contains bold stacked slogan text: small line “WAVE OF FLAVOR”, huge white line “RIDE THE”, huge yellow line “CRUNCH”, with a yellow brush-stroke underline. Top right contains one large red-and-white {argument name="brand logo text" default="Ruffles"} logo. Center foreground contains one young woman surfer riding one giant ridged potato chip like a surfboard. Bottom left contains one upright blue product bag labeled “Ruffles ORIGINAL” with ridged chips and dip imagery. Mid-right background contains one glowing red “Ruffles” sign on a beach bar.

Subject details: The main subject is exactly 1 athletic young woman, smiling and energetic, wearing glasses, a green zip-up tracksuit jacket with white sleeve stripes, a white crop top, green shorts, white crew socks with green stripes, and white sneakers. Her face appears softly anonymized/blurred in the final image while her body and outfit remain crisp. She crouches in a surfer stance with one hand extended left and the other hand reaching down right, hair blowing backward, riding a huge curved ridged potato chip board.

Scene details: The wave is one massive curling ocean barrel made of teal water, white foam, spray, and sparkling droplets, wrapping around the surfer. The background shows a warm orange sunset, palm trees, reflective shoreline water, and a casual beach restaurant/bar with lights.

Counted elements: Include exactly 1 giant ridged potato-chip surfboard; exactly 1 blue product bag; exactly 2 prominent brand logos outside the bag, one large top-right logo and one glowing sign in the background; exactly 4 visible text blocks: “WAVE OF FLAVOR”, “RIDE THE”, “CRUNCH”, and “ORIGINAL” on the bag; exactly 32 flying loose ridged chip pieces around the scene: 5 large foreground chips, 7 medium chips across the upper half, 12 small chips around the wave barrel, and 8 tiny golden chip fragments near the surfer and water spray.

Visual style: Premium snack commercial, cinematic lighting, dynamic action composition, hyper-realistic food texture, crisp ridges on every chip, splash photography, colorful summer energy, shallow depth of field, high contrast, polished advertising finish.

Constraints: Keep the poster readable, make the brand logo clean and recognizable, do not add extra slogans, do not add extra people, do not add extra product bags, and keep all visible text in English. Use {argument name="main slogan" default="RIDE THE CRUNCH"} as the dominant headline and {argument name="product flavor" default="ORIGINAL"} on the bag.

Goal: Create a 3-by-3 social media ad storyboard collage for a beauty serum, showing the same young blonde woman demonstrating a glow serum in a warm bedroom/vanity setting. The subject’s face must be intentionally anonymized by a flat skin-tone rectangular blur/cover in every panel, while hair, pose, outfit, and product remain visible.

Canvas: Wide 16:9 composite image, 1200 x 675 style, divided into exactly 9 equal panels with thin black grid lines. Add a small white panel number with black outline in the top-left corner of each panel, numbered exactly 1 through 9.

Subject details: A young adult woman with {argument name="hair color" default="long wavy blonde hair"}, light skin, wearing a black sleeveless crop top. Warm natural indoor light, beige walls, soft bedroom background with white bedding in some panels. Beauty-ad aesthetic, realistic photography, influencer UGC style, shallow depth of field, golden glow lighting.

Product details: A small glass dropper bottle labeled with a minimal skincare brand, showing text like “AI Glow Up” and “Serum.” The bottle contains warm golden liquid and has a white dropper cap. The product should appear clearly in panels 5, 6, and 8.

Layout: Use exactly 9 discrete panels:
1. Close selfie-like shot of the woman in warm morning light, face covered by a skin-tone rectangle, text at bottom: “Morning glow, no filter”.
2. Centered close portrait in the same room, face covered, text at bottom: “Ready for the day”.
3. Wider bedroom shot on white bedding, woman posing confidently and touching her hair, face covered, text at bottom: “Confidence is key”.
4. Side-profile hair and shoulder shot against a beige wall, face covered, text at bottom: “My skin is glowing”.
5. Woman holding the serum bottle beside her face, face covered, text at bottom: “Introducing my secret”.
6. Closer product showcase with woman holding the serum bottle near her face, face covered by a larger rectangle, text at bottom: “Radiance Serum”.
7. Woman spraying a fine mist upward/sideways in warm light, face covered, text at bottom: “Instantly brightens”.
8. Woman holding the serum bottle and pointing toward it with the other hand, face covered, text at bottom: “Get the glow”.
9. Close-up of long wavy blonde hair and shoulder area with the face mostly covered by a large skin-tone rectangle, text at bottom: “Unlock your light”.

Text styling: Each panel has bold white caption text with a black outline/drop shadow, placed near the bottom. In every caption, highlight one key phrase with a bright green rounded rectangle behind it: “glow, no filter” in panel 1, “the day” in panel 2, “key” in panel 3, “glowing” in panel 4, “secret” in panel 5, the entire “Radiance Serum” phrase or the word “Serum” in panel 6, “brightens” in panel 7, “glow” in panel 8, and “light” in panel 9. Use {argument name="product name" default="Radiance Serum"} as the product/caption name in panel 6 if customized.

Constraints: Keep the collage realistic and consistent as if frames from one influencer video. Use exactly 9 panels, exactly 9 panel numbers, and exactly 9 captions. Do not add extra logos, watermarks, UI elements, or additional text beyond the captions and simple product label.

Goal: Create a warm, nostalgic Filipino snack advertisement for {argument name="product name" default="Chicharrón"}, set in a lively public market with golden sunset lighting and a vintage commercial poster feel.

Canvas: Vertical 2:3 portrait advertisement, full-bleed photo-realistic composition with bold typography. Use warm browns, golds, cream, and deep red accents, with shallow depth of field and a busy but softly blurred market background.

Main subject: Center a young Filipino woman standing in a traditional outdoor market aisle. She wears a beige short-sleeve polo shirt tucked into high-waisted denim shorts, with long wavy dark brown hair. Her face is covered by a simple rectangular skin-tone privacy block, and one small background face is also covered by a smaller skin-tone block. Her pose is friendly and promotional: one hand holds a gold snack bag upright at chest height, the other hand presents a white plate toward the viewer.

Product and food details: Show exactly 2 featured snack items: 1 gold foil bag labeled “Chicharrón” with visible smaller text “CRISPY & DELICIOUS” and a red “100%” badge, and 1 white round serving plate piled with many crispy pork-rind curls. In the center of the plate place exactly 1 small dipping bowl filled with dark vinegar sauce with chopped chili or herbs. Make the pork rinds crunchy, airy, golden, and appetizing.

Text content: At the top, add a huge bold vintage headline in distressed cream block letters: “{argument name="headline text" default="CRUNCH"}”. Directly underneath, add a slanted yellow subheadline with black shadow: “{argument name="subheadline text" default="THAT BRINGS YOU HOME!"}”. Along the bottom, add a deep red banner with thin gold trim and decorative burst marks at both ends, containing distressed cream text: “{argument name="bottom banner text" default="AUTHENTIC PINOY SNACKS - SINCE 1993."}”.

Layout: The headline occupies the top quarter of the image, the woman fills the center, the snack bag is on the viewer’s left, the plate is on the viewer’s right, and the bottom banner spans the full width. Keep the bag label readable and the plate prominent.

Visual style: Photorealistic advertising photography blended with retro snack-poster design, warm backlight, cinematic market atmosphere, appetizing food styling, subtle film grain, high contrast, and crisp typography.

Constraints: Include exactly 3 main text areas: top headline, yellow subheadline, and bottom banner. Include exactly 1 snack bag, exactly 1 serving plate, and exactly 1 dipping bowl. Do not add extra logos, watermarks, QR codes, or additional promotional copy.

Goal: Create an ultra-high-quality, detailed anime-style storyboard sheet for a 15-second cooking video about {argument name="dish theme" default="golden fried rice"}, showing the full cooking process in chronological order.

Canvas: One wide 16:9 storyboard sheet, divided into exactly 12 panels arranged in a clean 4 columns by 3 rows grid. Use thin black panel borders, cinematic close-ups, warm restaurant-kitchen lighting, dramatic steam, oil shimmer, flying rice grains, and appetizing golden colors. No captions, no speech bubbles, no panel numbers, no watermark.

Visual style: Premium Japanese anime food commercial look, highly detailed semi-realistic cooking illustration, glossy highlights, shallow depth of field, dynamic action framing, orange firelight and golden egg tones, crisp textures on rice, scallions, diced meat, wok metal, and wooden cutting board.

Storyboard panels, exactly 12 in this order:
1. Top-down ingredient prep shot on a wooden board: a large glass bowl of cooked white rice, 4 brown eggs in a small bowl, long green onions/scallions, a bowl of cubed pork or ham, and 3 small seasoning dishes containing white salt, black pepper, and dark soy sauce.
2. Close-up of hands chopping scallions into small green rings with a large chef’s knife on a wooden cutting board.
3. Close-up of hands cracking one egg over a clear glass mixing bowl, yolk and egg white falling with sparkling highlights.
4. Close-up of beaten eggs being vigorously whisked with chopsticks in a glass bowl, swirling bright orange-yellow liquid.
5. Hot black wok over open flames, oil being poured in a thin golden stream, steam rising.
6. Pour beaten egg into the sizzling wok; the egg begins to scramble and bubble into soft golden curds.
7. Toss rice, diced meat, scallions, and egg together in the wok; ingredients fly upward in a dramatic midair arc.
8. High-energy wok-tossing shot surrounded by intense orange flames and sparks, golden fried rice scattering in the air.
9. Extreme appetizing close-up of finished fried rice with glossy individual grains, yellow egg pieces, green scallions, and browned meat cubes.
10. Wider kitchen action shot: a cook’s hand sprinkles seasoning from above into the steaming wok while the fried rice cooks over flame.
11. Hero plating shot: a neat dome mound of golden fried rice on a white plate with a subtle patterned rim, steam rising, warm kitchen background.
12. Final close-up spoonful shot: a metal spoon lifts glossy golden fried rice with egg, scallions, and a large cube of meat, with the plated rice blurred behind it.

Constraints: Keep the grid layout exact with 12 distinct panels, chronological left-to-right and top-to-bottom storytelling, no visible text, emphasize {argument name="main color palette" default="gold, amber, orange firelight, deep black wok, fresh green scallions"}, make the dish look rich, hot, and delicious.

{
  "objective": "Create a candid, smartphone-style street photo of an {argument name="subject" default="extremely beautiful slender Korean woman"} walking confidently in a {argument name="location" default="narrow urban alley at night"}.",

  "aspect_ratio": "3:4",

  "composition": {
    "style": "single candid shot",
    "camera_type": "smartphone",
    "angle": "slightly low angle, centered subject walking toward camera",
    "framing": "full body, natural imperfect framing",
    "aesthetic": "raw, unpolished, street photography"
  },

  "character": {
    "ethnicity": "Korean",
    "appearance": "extremely beautiful, slender figure, soft facial features",
    "hair": "short to medium length dark hair, slightly messy, natural movement",
    "expression": "confident, slightly playful, relaxed",
    "pose": "walking forward mid-step, natural stride, arms slightly swinging"
  },

  "wardrobe_and_accessories": {
    "outfit": "{argument name="outfit" default="low-neck fitted mini dress (light color, slightly tight)"}",
    "shoes": "knee-high boots",
    "accessories": [
      "stylish sunglasses",
      "handbag held loosely",
      "subtle jewelry"
    ]
  },

  "environment": {
    "setting": "narrow urban alley at night",
    "details": [
      "grungy walls with graffiti",
      "metal doors and vents",
      "dim warm lighting from above",
      "deep shadows and contrast",
      "slightly wet or textured ground"
    ]
  },

  "lighting": {
    "type": "harsh direct smartphone flash",
    "effect": "strong highlights on subject, dark background falloff, high contrast",
    "realism": "slightly overexposed skin highlights typical of phone flash"
  },

  "style": {
    "rendering": "photorealistic",
    "texture": "visible grain, noise, slight blur in motion areas",
    "imperfections": [
      "flash glare",
      "slight overexposure",
      "uneven lighting",
      "minor motion blur"
    ],
    "color_grading": "natural with slight warm tint"
  },
  "negative_prompt": [
    "studio lighting",
    "perfect composition",
    "overly smooth skin",
    "high-end fashion shoot",
    "cinematic look",
    "over-polished",
    "symmetrical perfection"
  ]
}