GPT Image 2 prompts

Create a whimsical square anime-style illustration of a cute giant custard pudding creature sitting calmly on a flat mossy rock in the middle of a tropical rainforest stream during heavy rain. The character is a glossy golden flan shaped like a squat frog, with closed relaxed eyes, a tiny puckered mouth, small rounded front limbs and hind feet, and translucent caramel sauce dripping down from the top. On its head place exactly 2 dessert toppings: one swirl of whipped cream and one bright red cherry with a stem. The creature should look serene and friendly, like an original mascot character, with soft highlights, wet reflections, raindrops sliding down its body, and tiny splashes on the caramel surface. Surround it with dense jungle foliage, large wet green leaves, one cluster of red tropical leaves on the left, misty depth in the background, and an orange-brown rain-filled stream flowing around the rock. Add visible diagonal rain streaks throughout the scene and water droplets on leaves and stone. In the upper right, add the Japanese greeting text {argument name="greeting text" default="こんにちわ"} in rounded playful lettering, warm orange-brown fill with a thick white outline, slightly tilted. Use a highly polished digital painting style with soft cinematic lighting, saturated greens and warm custard yellows, shallow atmospheric depth, adorable fantasy creature design, no humans, no extra characters, no watermark.

Goal: Create a clean studio character reference sheet for {argument name="character name" default="a young East Asian woman"}, designed for consistent AI image generation across shot distances and angles.

Canvas: Landscape 4:3 image, 1024×768 style, divided by thick white grid lines into four equal rectangular panels. Use a neutral medium-gray seamless photography background in every panel, soft even studio lighting, realistic photo quality, no text, no logos, no watermark.

Layout: Use exactly 4 panels. Top-left panel: a close-up front-facing head-and-shoulders portrait with the face clearly visible, centered, calm neutral expression. Top-right panel: a three-view full-body character turnaround on the gray background, showing exactly 3 standing figures side by side: front view with the face intentionally replaced by a smooth gray oval/blank mask, right side profile with the face blanked, and back view with hair and outfit visible. Bottom-left panel: a close-up left-facing side-profile portrait with the face clearly visible, head and shoulders composition. Bottom-right panel: a single full-body front view standing pose, centered far away with the head cropped off or omitted above the neck, emphasizing outfit and body proportions.

Subject details: The character is a slim petite young woman with fair skin, shoulder-length dark brown softly layered hair, wispy straight bangs, slightly curled-out ends, natural makeup, large brown eyes in the visible portraits, and a composed neutral expression. Outfit: {argument name="outfit" default="a pale cream short-sleeve ribbed polo knit top tucked into a matching high-waisted pleated mini skirt"}, white mid-calf socks, and chunky white lace-up shoes. Keep the same hairstyle, body shape, clothing, socks, and shoes consistent across all panels and all views.

Visual style: Minimal fashion model character sheet, realistic studio photography, soft shadows on the gray floor/backdrop, muted cream-and-gray palette, accurate human proportions, no stylization, no manga or illustration look.

Constraints: Exactly 6 depictions of the character are visible across the sheet: 1 front close-up portrait, 3 full-body turnaround views, 1 side-profile close-up portrait, and 1 headless full-body front view. Keep the face visible only in the two close-up portrait panels; in the top-right turnaround, simplify or hide the face with a gray oval; in the bottom-right full-body panel, omit or crop the head. Do not add extra poses, labels, props, accessories, or background objects.

Goal: Create a soft watercolor-style vertical 4-panel manga explainer about an AI manga creation workflow, featuring a cute chibi child with {argument name="hair color" default="bright pink"} hair and a blue sweatshirt demonstrating how to improve image generation prompts before final output.

Canvas: Portrait page, 4:5 aspect ratio, off-white paper texture, hand-drawn pencil panel borders, gentle pastel peach background washes, small pink cherry blossom doodles and sparkle accents scattered between panels. Use a warm, charming Japanese children’s-book manga style with light ink outlines, watercolor shading, and clean readable lettering.

Layout: Exactly 4 horizontal comic panels stacked vertically, separated by thin rough brown hand-drawn borders. The recurring main character appears once in each panel. Include exactly 4 speech bubbles, one per panel, with Japanese handwritten text.

Character details: A small chibi boy/girl mascot, toddler-like proportions, large amber-brown eyes, round face, short side-swept pink hair, wearing a loose sky-blue sweater. Expressions by panel: panel 1 curious and cheerful, panel 2 focused while inspecting, panel 3 serious and concentrated while correcting instructions, panel 4 delighted with eyes closed and big smile.

Panel 1: The character stands on the right, holding up a printed manga page with small grayscale/sepia comic panels showing a young woman. On the left is a large speech bubble containing: {argument name="panel 1 speech" default="まずは\n安いモデルで\n方向確認！"}. Add a tiny flower near the top center.

Panel 2: Close-up composition. The character is larger, leaning over a desk, holding a magnifying glass in one hand and inspecting a paper titled “YAML” on the right. The YAML sheet has red check marks, underlines, and a circled correction. A small generated manga page printout sits on the left side of the desk. On the right is a speech bubble containing: {argument name="panel 2 speech" default="AIと一緒に\nYAMLを\nチェック…"}.

Panel 3: Side view of the character on the right, writing with a red pen on a large instruction sheet labeled in Japanese “プロンプト指示”. The sheet shows multiple prompt categories and red corrections, including crossed-out marks, circled text, and emphatic red symbols near the top. On the left is a speech bubble containing: {argument name="panel 3 speech" default="ズレを直して\n指示を\n具体的に！"}.

Panel 4: The character stands on the right holding up a finished-looking manga page with three visible comic panels, featuring a young woman working at a desk and clean dialogue bubbles. Add pink flowers and golden sparkle marks around the page to imply success. On the right is a speech bubble containing: {argument name="panel 4 speech" default="仕上げは\n高精度モデルで\n完璧っ！"}.

Constraints: Use exactly 4 stacked panels, exactly 4 speech bubbles, exactly 1 main chibi character per panel, exactly 3 visible paper/page props total types across the comic: generated manga printouts, a YAML inspection sheet, and a prompt instruction sheet. Keep all Japanese text legible and handwritten-looking, avoid photorealism, avoid dark heavy shadows, no watermark, no extra panels.

Using the provided reference image as the character base, transform the flat LINE-sticker-style character into a small cute 3D figurine/toy while preserving its key design details and colors. Place it near the bottom center of a vertical smartphone wallpaper composition, sitting on a dark grassy mountain ridge with rocks in the foreground. Add a vast realistic night landscape behind it: layered mountain silhouettes on the horizon, a deep navy star-filled sky, and one prominent vertical Milky Way band rising through the right side of the sky. Use cinematic low-light photography, subtle depth of field, realistic ground texture, soft moon/starlight, and a calm adventurous mood. Canvas should be {argument name="wallpaper size" default="9:16 vertical smartphone wallpaper"}. Keep the character small relative to the landscape, readable but not dominating the scene. Do not add text, logos, borders, or extra characters.

Goal: Create a watercolor manga page in a vertical comic layout about three anime girls trapped in a gloomy closed room, with Japanese speech text.

Canvas: Portrait page, approximately 2:3 aspect ratio, off-white paper texture, hand-painted watercolor and ink line art, soft washes, visible panel borders, no digital UI, no watermark.

Layout: Use exactly 4 comic panels: (1) top-left small panel, (2) top-right small panel, (3) one large central panel spanning the width, and (4) a narrow bottom reaction strip. Keep the page reading like a Japanese manga, with vertical speech balloons.

Characters: Show exactly 3 girls. Girl A is a confident purple-haired girl with long curled hair, a side braid, green eyes, cat-ear-like black silhouette behind her head, wearing a shiny gold sleeveless turtleneck; she looks angry and sweaty in the top-left panel and appears again as an intense close-up in the bottom strip. Girl B is a cheerful blonde bob-haired girl with purple eyes, wearing a red frilled coat and a dark green scarf; she raises one fist triumphantly in the top-right panel. Girl C is a quiet black-haired girl with very long straight hair and red eyes, wearing a loose green dress; she sits barefoot on the floor in the central panel, knees bent, one hand near her lips, looking thoughtful and worried.

Panel details: Top-left panel has a dark bluish background and Girl A speaking with frustration. Top-right panel has a bright yellow starburst background with sparkles, Girl B smiling confidently, and a red Japanese sound effect near the top. The large middle panel shows Girl C seated in a dim, concrete or stone-walled room with a vertical metal pipe on the left, cold blue-gray shadows, and a small bright window or opening at the right casting pale light. The bottom strip shows only Girl A’s narrowed green eyes and purple hair against a red background, emphasizing suspicion and irritation.

Text content: Include exactly 5 vertical Japanese speech balloons and 1 red sound effect. Top-left panel has two speech balloons: {argument name="top left speech one" default="もう一ドヤって場合じゃないよ！"} and {argument name="top left speech two" default="こんな不気味な所に閉じ込められて……入り口、塞がってる!!"}. Top-right panel has one speech balloon: {argument name="top right speech" default="わははっ！ほら、入れるって言ったっしょ！さあ、手がかりを探そう！"}. Middle panel has one speech balloon: {argument name="middle speech" default="脱出も大事だけど…それより犯人、まだいるかもよ？"}. Bottom strip has one speech balloon: {argument name="bottom speech" default="ったく……いっつも後先考えないんだから……"}. Add the red sound effect 「ドヤ!」 in the top-right panel.

Visual style: Delicate shoujo manga faces, expressive eyes, thin ink outlines, watercolor gradients, slightly uneven hand-painted panel edges, soft paper grain, dramatic color contrast between the cheerful yellow top-right panel, cold blue-gray central room, and red bottom reaction strip.

Constraints: Count and show exactly 4 panels, exactly 3 recurring characters, exactly 5 speech balloons, and exactly 1 sound effect. Keep all dialogue vertical Japanese text, but do not add extra captions or additional text.

Create a square anime-style mascot portrait on a dark navy blue background. Center a confident anthropomorphic wolf character from the waist up, facing forward with a slight three-quarter angle, smirking and pointing directly at the viewer with one gloved hand. The wolf has spiky gray fur, white cheek and muzzle fur, tall pointed ears, sharp blue eyes, a black nose, and a mischievous expression. Put red-lensed steampunk goggles with a black strap on top of the head. Dress the character in a white futuristic racing jacket with red stripes and gray panel accents, plus a deep red scarf wrapped around the neck, trailing to the right with a small gold tassel fringe at the end. Use polished digital anime rendering with clean line art, soft cel shading, high contrast highlights, and a slightly glossy game-avatar look. At the bottom, add large bold Korean text reading {argument name="bottom text" default="엘베스"} in bright cyan, blocky sans-serif lettering with a subtle shadow, centered beneath the character. Keep the composition simple: exactly 1 wolf mascot, 1 pair of goggles, 1 red scarf, 1 pointing hand gesture, and 1 bottom text label. No extra characters, no watermark, no border.

Create a vertical 3:4 high-detail anime fantasy key visual featuring two royal mages in a dramatic magical city at sunrise. In the foreground, place {argument name="female character" default="a young violet-haired princess mage"} facing forward-right with an intense expression, luminous purple eyes, long flowing deep purple hair, and an ornate gold filigree tiara set with purple and green jewels. Dress her in a luxurious white, black, and violet robe with heavy gold embroidery, layered necklaces, dangling earrings, gemstone ornaments, and sparkling trim. She grips a long black-and-gold magic staff diagonally across the lower half of the image; the staff has a large faceted emerald orb and ornate golden crescent details, emitting bright neon-green circular spell rings, spiraling ribbons of energy, and glittering particles. Behind her on the left, show {argument name="male character" default="a black-haired prince sorcerer"} in three-quarter profile looking over his shoulder with sharp eyes, tousled black hair, and a high-collared crimson cloak covered in gold floral embroidery and glowing red sparks. His raised hand casts dark red-black magical tendrils that curl across the upper left sky like ink and fire. Background: a vast fantasy kingdom with dense rooftops and a tall radiant white-gold castle on the right, mountains in the distance, and a brilliant blue sky with warm sunrise clouds and lens-flare sunlight. Use an ultra-polished Japanese light novel cover style, cinematic composition, vivid saturated colors, intricate jewelry and fabric detail, dynamic wind-swept hair and robes, sparkling magical particles everywhere, dramatic rim lighting, high contrast, sharp focus, and epic romantic fantasy atmosphere. Keep exactly two main characters, one ornate staff, one castle, green magic around the staff, and red-black magic around the male sorcerer. No text, no logo, no watermark.

Goal: Create a clean 3D character design sheet for a cute stylized toddler girl named {argument name="character name" default="AIKO"}, presented like a professional toy/animation reference board.

Canvas: Wide horizontal character sheet on a warm off-white background with faint grey guide lines, thin section dividers, and a subtle hand-drawn annotation feel. Use soft studio lighting, high-quality 3D render style, rounded childlike proportions, plush toy/Pixar-inspired realism, and gentle pastel colors.

Header: Top left text reads “{argument name="character name" default="AIKO"} | 3D CHARACTER SHEET”. Top right include exactly 5 small rounded color swatches labeled: Cream, Blue, Red, Yellow, Pink.

Layout: Show exactly 5 character views across the sheet: 1 side view on the far left, 1 back view left-center, 2 front-area full-body views in the middle (one straight front pose and one 3/4 pose), and 1 close-up face view on the far right. Above the full-body columns place neat uppercase labels “SIDE VIEW”, “BACK VIEW”, and “FRONT VIEW”. Along the bottom, use handwritten-style captions: “SIDE VIEW”, “BACK VIEW”, “FRONT VIEW”, “3/4 VIEW”, and “CLOSE-UP FACE”.

Subject details: The character is a shy, sweet young girl with oversized golden-brown eyes, rosy cheeks, tiny smile, small nose, rounded face, and black hair gathered into a messy high bun with side bangs. Hair accessories: pink star clips. Outfit: cream ribbed long-sleeve sweater under a light blue denim pinafore dress with a small front pocket and embroidered patch detail, cherry-red tights, yellow rain boots, and a quilted pink backpack. The far-left side view shows her profile and backpack straps; the back view clearly shows the quilted backpack and rear of the bun; the front view shows hands clasped shyly; the 3/4 view shows the backpack and outfit volume; the close-up face is a large floating head cropped at the neck.

Annotations: Add thin curved callout arrows and handwritten labels pointing to exactly 9 details: “Black Hair Bun”, “Pink Star Clip”, “Quilted Backpack”, “Denim Pinafore”, “Cherry-Red Tights”, “Yellow Rain Boots”, “Shy Posture”, “Oversized Golden Eyes”, and “Rosy Cheeks”. Keep labels legible, black or dark brown, placed around the figures without clutter.

Constraints: No extra characters, no scenery, no watermark, no logo, no photorealistic adult proportions. Maintain a tidy concept-art sheet composition with consistent scale between full-body views and a larger close-up face.

Create a photorealistic ultra-wide mountain landscape at golden hour: a rugged dirt hiking trail begins in the lower right foreground and curves down through an alpine slope, bordered by mossy grass, low shrubs, wild vegetation, and large rounded gray granite boulders. Tall dark evergreen pine trees stand on both sides, with a dense conifer forest filling the valley below. In the middle distance, show a calm reflective mountain lake with a smaller connected inlet, nestled between forested ridges. In the background, depict three sharp rocky mountain peaks with striated gray stone faces and small patches of lingering snow, the tallest peak slightly right of center. The sun is low near the left horizon, partly above a distant ridge, casting strong warm orange light, long shadows, lens flare rays, and glowing highlights across the trail, rocks, trees, and mountain faces. The sky has dramatic soft clouds with orange, peach, and blue tones. Use cinematic HDR lighting, high detail, realistic depth, crisp foreground texture, wide-angle composition, natural colors, no people, no buildings, no text, no watermark.

Create a candid, flash-lit documentary photograph in a dim apartment at night, framed horizontally in a 4:3 aspect ratio. The subject is {argument name="subject" default="a pale adult woman with dark messy hair"} sitting at a dining table and staring directly into the camera with a tired, blank, slightly haunted expression. She wears {argument name="outfit" default="a pale pink satin robe with cream lace trim and a tied front"}; her makeup is smudgy and dramatic with dark eyeliner, flushed cheeks, glossy lips, and shiny skin from the harsh camera flash. Compose her slightly right of center, cropped from mid-torso upward, leaning near the table edge. In the foreground show a round dining table covered with a light floral tablecloth, with exactly 2 visible objects: a white ceramic mug with a small floral design on the left, and a small dark mobile phone or remote near the center front. Behind her include exactly 5 main background elements: a cane-back wooden dining chair, a dark open hallway doorway, a narrow interior door, a partially visible kitchen with brown wooden cabinets and a white dishwasher, and a wall calendar on the far right. Use muted beige walls, deep shadows, underexposed background, direct on-camera flash, slight red-eye-style glare, imperfect snapshot framing, and a gritty early-2000s compact-camera realism. Avoid glamour lighting, polished fashion styling, text overlays, or extra people.

Goal: Create a black-and-white sci-fi manga page showing a ruined futuristic city after a catastrophe, with cinematic wide shots and empty dialogue bubbles for later lettering.

Canvas: Vertical manga page, portrait orientation, clean white page border, high-contrast monochrome ink and grayscale screentone, detailed debris textures, dramatic cloudy sky, no color.

Layout: Use exactly 4 horizontal panels stacked from top to bottom, separated by thick white gutters.
- Panel 1: A very wide establishing shot filling the top third of the page. Show a devastated megacity with collapsed skyscrapers, broken elevated roads, rubble, dust, and a stormy sky split by bright light. Include floating chunks of wreckage in the sky. Add exactly 2 blank white manga speech bubbles: one tall rounded polygon bubble at the upper left and one larger rounded polygon bubble at the upper right.
- Panel 2: A narrow panoramic panel beneath it. Low-angle view of a towering damaged building wall and a dark sky with a levitating rock or debris chunk. Add exactly 1 blank white speech bubble on the right side.
- Panel 3: A narrow action-detail panel. Close-up of one black combat boot stepping onto cracked ground, with small stones and debris floating around it. Add exactly 1 small blank white speech bubble on the left. Leave a large empty white rectangular gutter/space on the far right of this row, as if the panel does not extend across the full page.
- Panel 4: A large final panel occupying the bottom portion. Rear view of a young girl and a small robot facing the ruined city and apocalyptic sky. The girl stands on the left, wearing a dark jacket, short shorts, thigh-high black socks, and sturdy boots, with short dark bobbed hair. The robot stands to her right, squat and cylindrical with a domed head, small side arms, treads, vents, and mechanical details. The background shows broken buildings framing the scene, rubble in the foreground, floating debris in the sky, and a bright crack of light through dense clouds.

Subject details: The main human character is {argument name="character name" default="a silent young girl"}, with {argument name="hair style" default="short dark bobbed hair"}, wearing {argument name="outfit" default="dark jacket, short shorts, thigh-high socks, and black boots"}. The companion is {argument name="robot companion" default="a small cylindrical treaded robot with a domed head"}. The setting is {argument name="ruined city setting" default="a collapsed futuristic megacity under a stormy sky"}.

Visual style: Serious seinen sci-fi manga, painterly ink wash, dense architectural detail, gritty rubble, dramatic perspective, atmospheric haze, heavy blacks, textured grayscale, cinematic composition, no sound effects, no visible lettering inside the bubbles.

Constraints: Exactly 4 panels and exactly 4 blank speech bubbles. Keep all speech bubbles empty white with black outlines. Do not add captions, titles, watermarks, logos, page numbers, or extra characters.

Goal: Create a wide two-panel hand-drawn social media illustration explaining {argument name="headline text" default="My API Stack"} on the left and showing a warm parent-child repair scene on the right.

Canvas: Horizontal 2:1 aspect ratio, split exactly down the center into two equal halves. Left half has a pale warm yellow background; right half has a pale sky blue background. Use clean ink outlines, soft pastel fills, subtle paper texture, and a friendly indie webcomic style.

Left panel layout: At the top center, write the large handwritten headline “{argument name="headline text" default="My API Stack"}” in dark ink with a small underline under “API” and tiny blue accent marks on both sides. Below it, create a hub-and-spoke diagram with exactly 6 circular nodes connected by thin black lines: 1 large central circle and 5 smaller outer circles.

Left panel node details, exactly 6 total:
1. Center large node: an illustrated bearded man in a tan baseball cap and gray shirt holding a glowing box full of small gears and a code icon, labeled “{argument name="central agent label" default="Hermes Agent"}”. Add tiny sparkles around him.
2. Upper-left small node: browser window with a magnifying glass over a globe, labeled “XAI Search”.
3. Upper-right small node: framed landscape image icon with a sparkle, labeled “GPT-Image 2”.
4. Lower-right small node: speech bubble with audio waveform, labeled “Whisper TTS/STT”.
5. Bottom small node: filmstrip/video icon with a play button, labeled “Kling 3.0 Video” with small sublabel “FAL.ai”.
6. Lower-left small node: document/page icon with speed lines, labeled “Default Extractor” with tiny subtext “60x faster” and “49x cheaper”.

Right panel subject: Draw an adult bearded man wearing a tan baseball cap, gray T-shirt, dark pants, and light socks, sitting cross-legged and smiling down at a small child. The child has short black bobbed hair and wears a soft yellow long-sleeve outfit, sitting on the floor and holding red-handled pliers. The adult holds a hammer while another hammer rests near his leg. Add exactly 4 loose tools on the floor: 1 blue-handled screwdriver, 1 gray adjustable wrench, 1 gray open-ended wrench, and 1 extra hammer. The scene should feel gentle, instructional, and wholesome.

Text: In the top-right corner of the blue panel, add the handle “{argument name="social handle" default="@dadhalfdev"}” in small dark handwritten text.

Constraints: Keep the split background clean with no border line other than the color change. Maintain the same character design for the central Hermes Agent and the adult on the right: bearded, tan cap, gray shirt. Avoid photorealism, avoid 3D rendering, avoid extra nodes, avoid additional labels, and keep all text legible.

Create a vertical cinematic triptych made of exactly 3 stacked horizontal panels, all showing the same {argument name="character description" default="elegant young East Asian woman with black hair in a low ponytail"} inside a modern airport terminal at golden hour. Panel 1: a wide establishing shot in a vast glass-walled airport concourse with a polished reflective stone floor, high ceiling, flight information boards on the left, check-in counters, distant travelers, large windows on the right, an airplane visible outside, warm sun streaming in from the right, and the woman standing full-body near the center wearing a {argument name="outfit" default="beige trench coat over a black suit and white shirt, slim black trousers, black heels"}. Panel 2: a realistic shoulder-up portrait of the same woman turning back over her shoulder toward the camera, soft shallow depth of field, blurred airport lights behind her, natural skin texture, delicate earrings, loose strands of hair framing her face, calm thoughtful expression. Panel 3: an extreme macro close-up of one of her eyes, showing detailed eyelashes, eyelid texture, tear line, glossy cornea, and a crisp reflection inside the iris of the airport concourse and the small full-body woman standing in it. Use photorealistic editorial fashion photography, high dynamic range, warm natural sunlight, cinematic contrast, accurate perspective, detailed reflections, no captions, no logos, no watermark. Canvas should be a tall vertical image with the three panels evenly stacked edge-to-edge, approximately {argument name="aspect ratio" default="9:16"}.

Ultra-realistic premium kids fashion advertisement featuring an adorable little boy with fair glowing skin, stylish brown hair, wearing a {argument name="jacket" default="beige corduroy jacket"}, {argument name="shirt" default="white T-shirt"}, rolled-up dark blue jeans, and premium {argument name="footwear" default="brown leather lace-up ankle boots"}. Sitting confidently on modern concrete steps with a cheerful smile, clean minimal urban background, focus on stylish footwear, luxury fashion campaign, soft natural daylight, cinematic lighting, editorial composition, DSLR photography, ultra-detailed, masterpiece, 8K,

9:16 vertical portrait photo of an adult {argument name="subject" default="Korean pop idol"}, single subject. Use a soft black mist filter look with low contrast, gentle highlight bloom, subtle glow, soft diffusion, and slightly lifted faded blacks. Minimal indoor scene beside a window with white curtains and a clean light toned background. The subject has natural minimal makeup, realistic dewy skin texture, and subtle human imperfections. Outfit is an {argument name="clothing" default="oversized white button up shirt with short bottoms"}, loose and casual, with no revealing styling. {argument name="hair style" default="Long dark hair"}, slightly messy, naturally voluminous, softly flowing. Pose is relaxed standing or a light lean, body subtly angled, one leg a little forward, shoulders relaxed. One hand lightly touches the collar or rests near the neckline while the other stays relaxed. Keep the body curve gentle and not exaggerated. Expression is a soft cute smile with slightly playful eyes, looking directly at the camera or just off camera. Close to mid body framing, eye level, intimate distance, with a slight handheld photography feel. Diffused natural daylight, soft shadows, light wrapping gently around the face and body. Mood is cute, intimate, everyday softness, quiet romantic atmosphere, with an understated dreamy tone. Ultra realistic quality, fine film grain, slight softness near the edges, natural imperfections.

Subject: Morning Rush

Person/Expression: A woman in her 20s. Bright ivory-beige skin with a natural texture and a hint of pink. Soft rose pink blush, champagne beige and pink-brown eye shadow, long thin eyelashes, natural eye bags, and sheer pink-beige lips. Small catchlights in the eyes. Holding toast in her mouth, she is running down the stairs with a hurried but bright expression.

Clothing/Pose: {argument name="clothing" default="White short-sleeved sailor-style blouse"} with elegant navy collar and sleeve lines, and a large, well-shaped navy ribbon. Well-fitted navy pleated skirt with a high waist, white socks, and black loafers. Her right foot steps down while her left foot kicks up slightly behind her. Her left hand grips the bag strap, and her right hand is slightly raised near her hair. The skirt and hair sway naturally with the running momentum.

Background/Light: Pedestrian bridge stairs in front of a station. Metal handrails, covered walkway, blurred station signs and street billboards in the back. Strong morning sunlight comes from the top right, illuminating her face, blouse, legs, and the edge of the stairs. Shadows from the handrail fall diagonally across the stairs, creating a sharp contrast between light and shadow. A slight white glow in the air.

Composition/Camera: Vertical 2:3, full-body composition. Position the subject slightly right of center, using the lines of the stairs and handrails to create depth. Slightly low angle, looking up from the front. Focus on the person, naturally blurring the background. A snapshot that captures a frozen moment of running.

Texture/Style: Realistic Japanese street photography. Bright summer morning, soft highlights, natural skin texture, delicate rendering showing individual hair strands. Hair is {argument name="hair color" default="bright milk tea ash brown"}, thin see-through bangs, layered cut along the face, medium-long below the shoulders. Hair ends curl slightly inward, spreading softly from the running motion. The bag is a dark brown Japanese school bag with several charms: a teddy bear, a small white stuffed toy, ribbons, fluffy keychains, and bead charms.

Create a realistic vertical smartphone selfie taken indoors from a very low floor-level angle, with the camera pointing upward. A {argument name="subject" default="woman"} with fair skin, naturally proportioned facial features, expressive eyes, and shoulder-length dark hair stands in the lower center of the frame wearing an oversized black T-shirt. She looks down into the camera with a warm genuine smile, bright friendly eyes, and a relaxed social-media selfie expression.

At the top center of the frame, an {argument name="pet" default="upside-down fluffy cat"} leans toward the lens, looking directly downward and dominating the upper composition with a cute, funny, and charming expression that matches the woman's friendly and cheerful mood.

Bright modern apartment interior, large windows, soft natural daylight, clean minimal background. Ultra-realistic skin rendering with visible natural pores, fine facial hairs, subtle skin texture, realistic eyelids, natural lip texture, soft under-eye details, authentic complexion variations, slight redness in natural areas of the face, realistic skin translucency, no beauty filter effect, no plastic skin, no excessive retouching, realistic facial depth and micro-details. Matte skin finish with natural light absorption, true-to-life skin tones and highly realistic texture reproduction.

Professional portrait photography quality captured with a Sony A7R V, 85mm f/1.8 portrait lens, f/2.0 aperture, ISO 100, ultra-high-resolution sensor, shallow depth of field, natural optical bokeh, realistic lens rendering, high dynamic range, accurate skin color reproduction, crisp focus on the face and eyes. Soft window light combined with realistic ambient indoor illumination, natural shadow transitions, subtle catchlights in the eyes, cinematic yet realistic lighting.

Highly detailed cat fur, individual hair strands visible, realistic whiskers and facial features. Natural smartphone-photo composition with balanced framing, candid atmosphere, social-media-friendly aesthetic, photorealistic, ultra-detailed, lifelike textures, realistic lighting physics, premium portrait quality, 8K resolution.

Portrait shot on 85mm f/1.4, close-up of a {argument name="ethnicity" default="Japanese"} woman in her mid-20s, short dark wet-looking hair, slightly damp texture, strong cheekbones, angular jawline, eyes with deep direct gaze toward window, expression neutral and contained. She wears a plain white ribbed tank top, bare shoulders visible. Upper body portrait, side profile leaning slightly toward window, arms folded on windowsill, subject occupies 60% of frame, window light as only source. Black and white rendering, high-grain film aesthetic, strong directional side lighting from window, soft falloff on shadow side, reminiscent of 1990s Japanese photography books. Photographic texture and quality: visible film grain especially in shadow, skin texture with natural pores and slight imperfection, stray hairs not perfectly groomed, photographic texture not CG render smoothness. Avoid: color, AI-smoothed skin, overly dramatic shadows. --ar 4:5

Creative composite street-food advertisement for Changsha stinky tofu, 16:9 horizontal aspect ratio, intentionally ugly anti-design meme poster, chaotic low-budget food ad aesthetic, deliberately cluttered and awkward composition. The background looks badly assembled from greasy night-market textures, cheap red-yellow menu scraps, stained plastic table covers, mismatched tiled wall fragments, handwritten price boards, oil splatters, and ugly pasted paper elements. Harsh uneven flash lighting, slightly off focus in places, awkward shadows, bad perspective, and an intentionally unpolished “made in 5 minutes” look.

At the center-right is a real-life photograph of authentic Changsha stinky tofu as the hero product. The tofu is served in a cheap paper bowl or disposable snack tray, featuring deep black crispy cubes with rough porous texture, golden crunchy edges, vivid red chili oil, chopped coriander, minced garlic, pickled long beans, and glossy sauce dripping into the cracks. Add visible steam, oily reflections, crunchy surface detail, spicy aroma feeling, and strong late-night street food realism. The food itself should look hyper-realistic, delicious, hot, and craveable, even though the surrounding design is intentionally messy and ugly.

Awkwardly interacting with the real stinky tofu is a badly hand-drawn black ink cartoon doodle character, scribbly and crude like a bored notebook sketch, with disproportionate limbs, wobbly line art, bad crosshatching, and silly emotional expression. The doodle character is pinching its nose with one hand while being dragged toward the Changsha stinky tofu by exaggerated smell waves, eyes turning into hearts, drooling uncontrollably, one foot lifted off the ground as if the smell is magically pulling it in. The scale is wrong, the perspective is wrong, and the doodle is integrated terribly on purpose.

Add extra ugly doodle elements around the food:
crooked smell waves,
messy arrows,
scribbled flames,
awkward little stars,
sweat drops,
shaky excitement marks,
badly drawn chili peppers,
tiny ghost-like aroma clouds,
sloppy handwritten side notes,
all in rough black notebook-doodle style.

At the top, add a deliberately ugly but highly eye-catching Chinese slogan in badly kerned anti-design typography:

"Smells like a turn-off"
"Eats like a long line"

Add a smaller messy handwritten side note nearby:

"This Changsha bite"
"Is stinky in just the right way"

At the lower center, add a bold distorted main title in Chinese:

"Changsha Stinky Tofu"

Add a fake street-snack sub-brand phrase in tacky anti-design style:

"Black and powerful"
"Fragrant beyond belief"

Optional additional ugly handwritten Chinese copy:
"First bite: frown"
"Second bite: so good"
"Third bite: want to take home"

The {argument name="bakery color" default="Purple"} Bakery on the British Street, a {argument name="building style" default="two-story storefront at the corner of the street"}, sells a variety of breads and attracts a lot of customers. The door is decorated with bright flowers, there are chairs at the door where customers sit and eat bread, there is a purple tree at the door, there is a bicycle parked at the door, and there are puppies at the door. {argument name="art style" default="Simple illustration style"}, hand-drawn irregular lines, lines have a slight shake, cute style --ar 9:16

Create a full-body ultra-realistic studio portrait using the uploaded photo as the exact face reference. Preserve the person’s face Identity with no modifications to facial structure, eyes, nose, lips, cheeks, jawline, skin tone, hairstyle, or bangs. Keep the expression natural and cute. Place the person in a {argument name="studio background" default="clean light gray seamless studio background"}. She is standing in a playful pose with one leg lifted slightly behind her, smiling naturally while holding a {argument name="prop" default="colorful rainbow lollipop"} in . Add a {argument name="accessory" default="pastel blue crossbody bag with a small Winnie the Pooh keychain"} hanging from it. Keep her hair exactly as in the original photo. Beside her, create a cute chibi illustration version of the same person with identical hairstyle and bangs, wearing the same outfit and blue bag. The chibi is playfully pulling the hem of her skirt while smiling mischievously. Add hand-drawn black doodles around both characters, including motion lines, excitement marks, tiny sparkles, and playful accents. The overall style should combine a realistic fashion studio photograph with a kawaii hand-drawn illustration. Bright, soft lighting, clean composition, adorable, cheerful, {argument name="aesthetic" default="modern Korean-inspired"} aesthetic, ultra detailed, high resolution, premium quality.

A mountain lifestyle portrait captured after the rain, evoking the atmosphere of a cool evening; a candid shot using flash. Create an image without altering facial features. Strictly maintain 100% likeness to the reference. Preserve the shape of the face, eyes, eyebrows, nose, lips, proportions, hair color, and all unique characteristics exactly as they appear in the reference image. No smoothing, no retouching, and no changes to facial geometry. Natural skin texture with visible pores. Vertical 4:5 format. Shot on a {argument name="camera model" default="Canon G7X Mark III"} with flash. Camera at chest level; a hip-up shot. Slight digital grain, subdued exposure, no filters, no text or watermarks. Do not blur the background. A young woman stands in a {argument name="setting" default="green mountain meadow"}. Her body is turned slightly to the left, while her head is turned to the right. She is smiling softly, gazing into the distance rather than at the camera. Both hands are relaxed in front of her; she holds a pair of narrow, dark brown sunglasses in one hand. Short, square-shaped nails with a French manicure. Hair color and length must strictly match the reference. Very long hair (reaching the waist), styled with a round brush (brush not visible in the frame); a layered, cascading cut with ends neatly curled inward—no waves. Hair is worn loose and slightly tousled by the wind, with a few strands resting softly against her face. She is wearing a {argument name="outfit" default="form-fitting, chocolate-brown outfit consisting of a short-sleeved T-shirt and wide-leg, low-rise trousers"}; the soft fabric accentuates her silhouette. She wears a small Van Cleef & Arpels Alhambra pendant and a Cartier Love bracelet. In the background, there is a spacious green field, a wooden fence, and a grazing horse. Thick fog hangs in the air, and the towering mountains are completely shrouded in heavy, dark-gray rain clouds. It is a cool mountain evening, with the atmosphere of the air just after the rain. A harsh Canon flash brightly illuminates the young woman; her skin is slightly overexposed, creating distinctive highlights on her face and clothing. The background remains naturally dark and atmospheric, yet the details of the field, fence, horse, and mountains are clearly visible. A lively, spontaneous lifestyle shot with maximum photorealism.

Create an intimate macro portrait in a 2:3 vertical aspect ratio: an extreme close-up, slightly high-angle profile view of a young woman in her {argument name="subject age" default="early 20s"} lying sideways as she rests. Her head is supported by a textured gray-brown cushion and her folded arm. Her eyes are closed with visible lashes, her expression peaceful and defenseless, and her lips are slightly parted, relaxed, glossy, and hydrated. Preserve the original facial identity if a reference is provided. She has fair pale skin with subtle warm undertones, dewy and smooth like hydrated silk, with soft specular highlights on the tip of the nose, cheekbone, and lower lip. Use clean Asian beauty-inspired no-makeup makeup: soft pink blush, natural brushed brows, and glossy lip tint. Her {argument name="hair color" default="dark brown"} hair is casually tucked behind one ear, with fine messy stray strands falling over her cheek and forehead for realism. Show her slender neck, delicate facial structure, collarbone, bare shoulder, and the hand or forearm beneath her head. Add a tiny silver star-shaped earring and a very delicate thin chain necklace. The scene should feel like a private afternoon nap in a quiet indoor bedroom or living space. Background: {argument name="background style" default="warm beige or cream walls"}, blurred architectural lines such as a doorframe or wall, and a textured cushion in the foreground. The framing should be slightly tilted and candid, like a stolen private moment rather than a posed shoot. Use soft diffused directional window light with no harsh shadows, ambient warm room light, subtle film grain, organic softness, and extremely shallow depth of field around f/1.4 to f/1.8 so the neck, ear, and background fall completely out of focus. Overall mood: tranquil, intimate, vulnerable, soft minimalism, delicate femininity, quiet moments. Keep the closed eyes, visible lashes, parted glossy lips, stray hairs across the face, dewy skin highlights, shallow depth of field, warm muted interior, and candid skewed framing. Avoid open eyes, looking at the camera, heavy makeup, matte or plastic skin, harsh studio light, flash lighting, harsh shadows, perfectly styled hair, stiff posing, sharp background, overexposure, and any artificial glamour retouching.

Colorful Royal Scene (8K) A majestic king and queen standing together in front of a magnificent golden palace at sunrise, wearing luxurious royal robes embroidered with gold and precious gemstones, elegant crowns sparkling with diamonds, colorful flower gardens, majestic white peacocks, glowing fountains, cinematic lighting, ultra-detailed faces, realistic textures, vibrant colors, fantasy realism, volumetric light, masterpiece composition, HDR, 8K, hyper-realistic, ultra-sharp focus. Dark Royal Scene (8K) A powerful king and queen standing on the balcony of a dark gothic castle during a thunderstorm, black and crimson royal outfits, ancient jeweled crowns, flowing capes, glowing torches, lightning illuminating the sky, ravens flying overhead, dramatic shadows, mysterious atmosphere, cinematic dark fantasy, ultra-realistic facial details, volumetric fog, high contrast, HDR, masterpiece, hyper-detailed, ultra-sharp, 8K quality.