GPT Image 2 prompts

Theme: {argument name="theme" default="Rooftop overlooking a mountain town"}

Subject: A vertical full-body photo of a single woman crouching deeply in the center of the frame on a rooftop overlooking a mountain town, with her hands hanging between her knees. Wearing {argument name="clothing" default="a white shirt, black tie, black mini skirt, and pumps"}.

Character/Expression: Face is frontal, dark brown eyes directed at the camera, quiet expression with closed pale lips. Slightly long oval face with a small round chin, narrow almond-shaped eyes, thin parallel eyebrows, thin nose bridge, and a small tip. Natural beige light makeup. Long straight dark brown hair with a center-ish part and thin bangs, falling from shoulders to below the chest.

Clothing/Pose: Very pale bluish-white long-sleeved shirt. Collared with sleeves rolled up to the elbows, a thin black tie hanging loosely and long. Black high-waisted pleated mini skirt and black glossy pointed high-heeled pumps. Both knees opened wide to the sides with heels touching the ground, crouching deeply, arms extended between legs with fingertips pointing down.

Background/Light: {argument name="background" default="Gray rooftop floor and white low wall at the bottom, dense tile-roofed town in the center background, diagonal power lines on the left, green mountain range and pale blue sky in the upper half"}. Hard noon main light from the upper right, casting a deep shadow of the person to the lower left of the floor.

Composition/Camera: 3:4 vertical, full-body portrait captured from a horizontal perspective slightly lower than the crouching figure, head to shoes in frame. Focus on the face, with the town and mountains softly blurred by a shallow depth of field.

Texture/Style: Photorealistic natural light photography. High-definition rendering of shirt fabric, pleats, glossy shoes, concrete, and tile roofs, based on calm black, white, and green tones.

Negative: Do not change deep crouching to standing; do not omit the loose black tie.

Subject: 
{argument name="theme" default="Sea-colored peace sign"}

Main Subject:
A vertical above-the-thigh photograph of a single woman sitting at the edge of a sunny shore in the center of the screen, making a peace sign with her left hand near her eyes. A {argument name="swimsuit" default="coral and light blue floral print bikini"}, a white shirt falling from her shoulders, and the blue-green sea are visible.

Person/Expression:
Facing almost front, with large gray-brown eyes towards the camera, a playful expression with pink lips slightly pursed. Small oval face with a round chin, large rounded almond-shaped eyes, thin gentle eyebrows, small bridge of the nose and a round tip, plump and glossy lips. Long eyelashes, pale pink cheeks, smooth bright skin. {argument name="hairstyle" default="Dark brown long hair"} with thin bangs and thin stray hairs on both sides remaining, the top part lightly braided and falling straight down to the back and chest.

Clothing/Pose:
A triangle bikini swimsuit based on coral color with light blue edges and a faint floral pattern. Fine neck ties, a small knot in the center of the chest, and light blue ribbon ties on both hips are visible. A white long-sleeved shirt with transparency and woven patterns is worn open and dropped from both shoulders. A thin flower-shaped pendant. Sitting with the left knee open towards the lower left of the screen, raising her own right hand (left side of the screen) next to her face, opening two fingers horizontally.

Background/Light:
The bottom of the screen is a sandy beach mixed with pebbles and a concrete edge, the center back is white waves and a transparent blue-green sea, and low mountains and seaside buildings in the left and right far distance. Hard direct light hits from the sun above, making the skin, hair, white shirt, and sea surface shine strongly.

Composition/Camera:
A 3:4 vertical, close-up portrait captured from head to above the thighs from a slightly looking-down bird's-eye view. The person occupies a large portion of the center, with the peace-sign fingers overlapping near the right eye in the upper left foreground. Focus on the face and eyes, with the sea and mountains softly blurred with a medium depth of field.

Texture/Style:
A bright photorealistic live-action photograph. High-definition depiction of swimsuit fabric, transparent shirt, skin, hair, and sparkling seawater, vividly contrasting coral and turquoise.

Negative:
Do not omit the horizontal peace sign; do not make the floral pattern and light blue edges a solid color.

Subject: Hand reaching toward the summer sky

Main: {argument name="scene" default="A vertical full-body-leaning photo of a single woman in a sunny park, reaching her left hand out greatly toward the camera"}. {argument name="clothing" default="A short peach-colored dress with a small floral pattern"}, with the palm in the immediate foreground, blue sky, and trees as main elements.

Character/Expression: Leaning forward slightly, face facing forward, brown eyes looking at the camera with lips slightly parted. Small oval face with a thin chin, slightly slanted almond eyes, thin parallel brows, straight nose, and glossy pale pink lips. Bright skin with pink blush and thin eyeliner lifting the corners of the eyes. Long brown hair with thin bangs and large waves flowing to the right side of the chest.

Background/Light: Green grass at the bottom, dark green trees on the sides and center-back, and a vivid blue sky with few clouds in the upper half. Hard direct sunlight from the upper left creates starburst rays and lens flares, brightly rimming the hair and shoulders.

Composition/Camera: 3:4 portrait, low-angle view looking up from near the ground. The subject occupies most of the center, showing from head to below the thighs. The extended left hand is extremely large in the lower-left foreground, creating a wide-angle perspective where the face and torso appear smaller in the distance. Focus on the face, with the hand slightly blurred.

Texture/Style: Vivid photorealistic live-action photo. High-definition expression of skin, thin floral fabric, lace, translucent nails, grass, and sunlight filtering through trees.

Negative: Do not make the hand in front small; do not change the shape of the lace-up dress.

Subject: Bouquet and Bicep Flex

Main: {argument name="scene" default="A vertical waist-up photo of a single woman in a sunny green promenade, holding a pink bouquet in her left arm and flexing her right bicep"}. {argument name="clothing" default="A white ribbed sleeveless top, V-neck, with small white buttons down the front"}. A white high-waisted flare skirt with a pale blue floral pattern.

Character/Expression: Face forward, large brown eyes looking at the camera, bright smile showing teeth. Rounded oval face with a small chin, round almond eyes, thin parallel brows, thin bridge and small tip of the nose, pale coral lips. Bright skin with pink blush and natural eye makeup. Dark brown long hair with thin bangs and loose waves, flowing together toward the right side of the chest.

Background/Light: A bright tiled promenade extends from the bottom to the back right, with low hedges and fresh green trees on the sides, a green hill in the center-back, and a white building on the far right. Soft midday main light from above brightly illuminates the skin and flowers. The bouquet consists of pale pink roses, small pink and white flowers, and green leaves, wrapped in pink paper and transparent mesh.

Composition/Camera: 3:4 portrait, eye-level shot from head to below the waist. The subject occupies the center, the bouquet is in the lower-left foreground, and the bent arm extends to the upper right. Focus on the face and bouquet, with the background brightly blurred due to shallow depth of field.

Texture/Style: High-key photorealistic live-action photo. Soft, high-definition depiction of petals, wrapping paper, ribbed fabric, skin, and hair, refreshing with white, pink, pale blue, and fresh green.

Negative: Do not make the flexed right arm thin; do not make the bouquet small.

Subject: Night game pitching pose

Main: {argument name="scene" default="A vertical full-body photo of a single woman on a baseball mound at night, raising her left knee high before pitching and holding a black glove in front of her chest"}. {argument name="clothing" default="An off-white short-sleeved baseball shirt and a dark navy high-waisted pleated mini skirt"}. Spectator stands and two strong lighting towers are in the background.

Character/Expression: Body turned to the left, looking back at the camera over the shoulder with brown eyes staring straight and smiling modestly. Small rounded oval face with a round chin, large almond eyes, thin parallel brows, small well-shaped nose, and pale coral lips. Light pink on the cheeks, natural eye makeup. Dark brown short bob with bangs touching the eyebrows and ends flipping outward.

Clothing/Pose: Off-white short-sleeved baseball shirt. Dark navy round neck and sleeve piping, with a small dark mark on the chest. Dark navy high-waisted pleated mini skirt, white ribbed socks, and white spikes with dark navy lines. Hands together inside a black leather glove, raising her left knee higher than the waist, standing on her right leg.

Background/Light: Reddish-brown mound dirt and white lines at the bottom, green fence and stadium in the middle ground, blurred spectator stands in the background. White lighting towers shine strongly in the upper left and right. Hard main light from the upper left and right evenly illuminates the hair, face, clothes, and legs; the night sky is deep navy-black.

Composition/Camera: 3:4 portrait, near-eye-level sports portrait from the head to below the shin of the pivot leg. The subject occupies the center, with the raised left leg extending to the left. Focus on the face and glove, with the spectator stands strongly blurred due to shallow depth of field.

Texture/Style: Sharp photorealistic live-action photo. High-definition depiction of leather glove, uniform fabric, dirt, and skin, utilizing the contrast between strong stadium lighting and the dark night.

Negative: Do not lower the raised left leg; do not hold the glove with one hand.

Subject:
{argument name="subject" default="Arms spread at a waterfall basin"}

Main Subject:
A vertical full-body photo of a single woman in the center of the frame, walking through shallow water at a waterfall basin in a deep forest with her back to the camera, spreading both arms to the sides. The hem of her {argument name="clothing" default="black small floral print dress"} swirls and spreads.

Person/Expression:
A profile view with the back to the camera and the face turned to the left of the screen. A quiet expression with downcast eyes, a slender nose, and closed lips visible in the silhouette. Fair skin. Her {argument name="hair color" default="dark brown long hair"} is partially tied back at the crown, with the rest flowing down her back, and the ends and fine loose strands fluttering in the wind to the right.

Clothing/Pose:
A short one-piece dress with thin shoulder straps featuring small white and yellow flowers on a black background. The back is deeply V-shaped with a thin strap across the shoulder blades and a layered sheer frill hem. Both arms are extended to the sides slightly lower than the shoulders, with relaxed fingertips. Feet together, wading forward in ankle-deep water.

Background/Lighting:
The bottom of the screen shows a transparent blue-green shallow area with fine reflections, and the entire background is a dark mossy rock wall. A thin waterfall descends in the right middle ground, with bright green leaves at the top right. Soft sunlight filters in diagonally from the top right, edging the hair, arms, legs, and swirling hem in gold.

Composition/Camera:
A 9:16 very vertical full-body portrait from behind, captured from a low horizontal viewpoint near the water surface, including head to toe and reflections. The person occupies a large part of the lower center, leaving the dark rock wall as negative space in the upper half. Focus on the person and the hem, with a shallow depth of field softly blurring the waterfall and rock wall.

Texture/Style:
Cinematic photorealistic live-action photography. High-definition depiction of translucent wet fabric, skin, water surface, splashes, and mossy rocks, contrasting deep green with warm backlight.

Negative:
Do not flip the back-facing posture; do not thin the spreading floral hem.

A photorealistic, high-resolution lifestyle photograph of the same {argument name="subject" default="young woman with brown hair styled in a high ponytail"}, wearing a light cream-colored sweater, dark leggings, thick white socks, and white sneakers, sitting on a high rocky mountain peak overlooking a {argument name="background" default="vast, breathtaking alpine valley with snow-capped peaks"} in the background. In the composition, she sits in profile with her knees tucked up, leaning her head on her hand with her eyes closed in a peaceful, serene expression as she enjoys the fresh mountain breeze, accompanied by a vintage canvas backpack with leather straps resting on the rock beside her. The lighting is bright and natural daylight with soft clouds drifting below, and the shot is captured from a slightly elevated angle with a shallow depth of field that keeps the traveler sharply in focus while softly blurring the magnificent distant mountain ranges.

Cinematic candid photograph of a {argument name="young couple" default="young couple"} sitting together on the seat of a {argument name="public transport" default="city bus/subway"} at {argument name="time of day" default="night"}. The young man has tousled dark hair, wearing a simple black jacket and black shirt, gazing quietly out the window with a thoughtful, distant expression. The young woman sits close beside him, resting her head gently on his shoulder with her eyes closed, looking peaceful and sleepy. She wears a dark oversized jacket and carries a simple shoulder bag.
Shot naturally through the bus interior, with yellow vertical handrails in the foreground creating soft framing and partial occlusion. Dark windows behind them reflect blurred city lights and passing buildings. Warm tungsten interior lighting mixed with cool, dim nighttime ambient light outside. Subtle reflections on the glass, soft shadows, realistic skin texture, slightly messy natural hair, understated expressions, intimate but completely natural body language.
35mm film photography aesthetic, low-light cinematic realism, muted olive, amber and deep-black color palette, fine film grain, slight halation, soft focus falloff, shallow depth of field, imperfect candid composition, gentle motion blur from the moving vehicle, authentic urban atmosphere, nostalgic indie-film mood, quiet romantic storytelling, natural exposure, no posing, no glamour retouching.
Vertical 4:5 composition, medium shot, realistic proportions, highly detailed, photorealistic, editorial street photography, subtle analog texture.

Mid-century theatrical cartoon still promotional Hollywood movie trailer banner featuring a {argument name="adventurer" default="daring adventurer with tousled brown hair"} and a confident smile, wearing an open safari shirt and leather jacket, and a {argument name="heroine" default="spirited heroine with wind-swept dark hair"} and bright eyes, wearing a practical yet stylish khaki outfit, seated together on a weathered crate inside a {argument name="setting" default="exotic jungle camp"}. Stage them in a lighthearted, adventurous, good-humored interaction, with both characters fully visible, relaxed, and clearly reacting to each other with shared excitement. Design the figures with exaggerated early-1950s animation caricature: clean black contour lines, sharply simplified anatomy, elegant angular silhouettes, expressive eyebrows, stylized facial shapes, graceful hand gestures, and cheerful readable expressions. Render the entire image in a unified traditional cel-animation aesthetic, with the background following the same stylized visual language as the characters, using flat opaque colors, minimal hard-edged shading, simplified furniture and décor, subtle hand-painted texture, controlled perspective, and crisp theatrical staging. Use a bold adventure palette of warm khaki, deep green, soft gold, and cream and maintain a witty, charming, visually snappy mood.

Create one complete high-detail professional storyboard sheet image for a 30-second 3D animated short film.

STORY TITLE: “{argument name="story title" default="The Tiny Crew and the Last Photo"}”
SHEET: PART 1 — {argument name="timeframe" default="0:00 to 0:15"} https://t.co/PMPoGt3vTq

Create a premium, modern high-fashion CHARACTER CONCEPT ART SHEET in a 16:9 widescreen layout on a pure white background. 

THE ATTACHED REFERENCE IMAGE DEFINES THE ART STYLE — treat it as a strict style bible and reproduce its exact rendering

Create a single vertical anime animation development board for an original emotional short film titled "{argument name="film title" default="The Boy and the Umbrella"}".
The output must be ONE combined image containing an anime character design sheet and a cinematic storyboard page.

Create a playful full-body studio composite of {argument name="character name" default="a stylish teenage boy"} being followed by a tiny chibi version of himself. The main character is a realistic young man with voluminous dark brown wavy hair, clean face, relaxed slightly aloof expression, looking back over his shoulder while walking to the right. He wears a burgundy bomber jacket over a black hoodie, oversized faded blue cargo jeans with large side pockets, and gray-and-white sneakers; his right hand holds a clear plastic cup of iced tea or coffee. Behind him at lower left, include exactly 1 miniature chibi duplicate, about one third his height, with the same hair and outfit, exaggerated angry face, furrowed eyebrows, clenched teeth, and both hands gripping and pulling the loose fabric of the main character’s left pant leg backward, stretching the denim dramatically. Use a clean light gray seamless studio background with soft floor shadow, high-detail photorealistic rendering for the main teen, glossy 3D cartoon/anime styling for the chibi, and a humorous fashion-editorial mood. Add exactly 10 simple black hand-drawn doodle accents around the figures: 2 sparkle icons near the upper body, 1 sparkle icon to the right, 1 squiggly surprise mark near the main character’s head, 2 small steam-puff frustration clouds near the chibi, 1 anger symbol near the chibi’s head, and 3 short motion-line clusters around the stretched pant leg and feet. Keep the composition vertical, centered, with ample negative space, no text, no watermark, no extra characters.

Create a cinematic, polished, naturally realistic studio portrait of a {argument name="subject" default="handsome young adult man"} facing the camera head-on from mid-chest up. He has {argument name="hair color and style" default="thick tousled dark brown hair with volume"}, warm brown eyes, strong brows, subtle stubble and a faint mustache, a defined jawline, and a calm serious expression with closed lips. Dress him in a {argument name="clothing" default="plain off-white crewneck sweatshirt"}. Use dramatic movie-style lighting with a warm amber key/rim light from the viewer’s left catching the hair, cheek, ear, and shoulder, balanced by a cool cyan-blue rim/fill light from the viewer’s right. Set him against a dark smoky blue-black studio background with a soft teal glow on the right side, shallow depth of field, crisp facial detail, realistic skin texture, natural pores, glossy catchlights, and high-end editorial color grading. Center the subject symmetrically, vertical 4:5 composition, no text, no accessories, no watermark, no extra people.

Create a realistic intimate lifestyle portrait of {argument name="subject" default="a young adult woman"} lounging on her stomach on a dark charcoal sofa in a dim modern apartment at night, holding and looking over a smartphone near the bottom foreground. She has {argument name="hair color and style" default="dark brown shoulder-length straight hair swept to one side"}, fair skin, a calm slightly dreamy expression, glossy natural lips, and thin round metal eyeglasses with subtle green-blue phone reflections on the lenses. She wears {argument name="outfit" default="a cream silk camisole with lace trim and thin straps, with matching light shorts"}. Pose her diagonally across the couch with elbows forward, hands around the phone, shoulders and upper torso closest to the camera, one bent leg visible behind her. Use warm low ambient apartment lighting mixed with cool phone glow on her face and fingers, shallow depth of field, soft cinematic bokeh, realistic skin texture, natural proportions, and a candid indoor photography feel. Background: blurred beige walls, a dark hallway or kitchen entrance on the left, a small wall thermostat/switch on the back wall, and minimal furniture, all unobtrusive. Composition: vertical 2:3 portrait, eye-level slightly above the couch, 35mm lens look, close focus on face and hands, no text, no watermark, tasteful fashion/lifestyle mood, not explicit.

Goal: Create a dark futuristic widescreen tech poster introducing {argument name="headline text" default="今日分享：5 个 GitHub 爆涨星技能"}, showing five fast-growing GitHub AI tools as large interface cards above a modern developer workstation.

Canvas: 16:9 horizontal social media banner, cinematic dark office background, cool blue lighting, high contrast, sharp UI mockup style. Use a blurred glass-and-metal workspace at night, with a desk across the bottom third.

Layout: Place a large centered Chinese headline at the top; make the number “5” oversized and bright electric blue while the rest of the headline is bold white. Under it, add a thin decorative line and small subtitle text reading {argument name="subtitle text" default="真实工作流场景一图看懂"}. Arrange exactly 5 numbered feature sections in a row across the middle, each with a circular number badge, tool name, short Chinese description, and a large screenshot-style UI card beneath it. The five sections are: 1) GPT-Image2-Skill, description “图像生成与创意设计”; 2) smart-ralph, description “智能开发规划与执行”; 3) OpenAI Skills, description “可复用的 AI 技能目录”; 4) open-saas, description “开箱即用的 SaaS 基础设施”; 5) awesome-copilot, description “Copilot 配置与提示词库”.

Subject details: Section 1 card is a dark prompt-library interface titled “GPT-Image2-Skill,” with a left sidebar of categories and a grid of poster/UI/illustration thumbnails. Section 2 card is a dark coding workflow interface titled “smart-ralph,” showing a task checklist, code editor, checkpoints, and an approval dialog. Section 3 card is a clean light “Skills Directory” interface with category navigation and stacked skill cards such as Code Reviewer, API Documenter, SQL Assistant, and Data Analyst. Section 4 card is an open-saas dashboard mockup with a landing page headline “Build Faster. Launch Sooner.”, login panel, dashboard metrics, payments table, and deploy panel. Section 5 card is a dark awesome-copilot marketplace/library interface with tabs and exactly 6 visible tool rows: Python Best Practices, API Development, Test Generator, Code Refactor, Security Scanner, and Docs Generator.

Desk foreground: Include exactly 4 main desk objects: an open laptop on the left showing a mountain wallpaper, two wide desktop monitors in the center showing code and analytics dashboards, a white keyboard with mouse, and a small potted plant near the right monitor. Keep the desk minimal, premium, and realistic.

Visual style: Premium GitHub/AI tooling aesthetic, dark navy and black palette, electric blue accents, crisp white typography, glassmorphism UI panels, realistic reflections, subtle depth of field, cinematic lighting, polished startup presentation design.

Constraints: Keep all five tool names readable. Do not add extra tools or extra numbered sections. Preserve the Chinese headline and short Chinese descriptions. Avoid watermarks, logos unrelated to the five tools, clutter, distorted text, or cartoon styling.

Goal: Create a cinematic first-person shooter screenshot in Shanghai at night, blending gritty realistic military-game visuals with a surreal giant anime maid looming over the skyline.

Canvas: Wide 16:9 frame, dark rainy night, high-contrast blue-gray color grading, wet lens grime, film grain, subtle scratches, dramatic storm clouds, reflections on the river.

Scene and layout: The viewer stands on a riverside battlefield facing the Pudong skyline across dark water. The Oriental Pearl Tower is clearly visible on the far left, with illuminated skyscrapers and neon signs along the waterfront. In the lower right foreground, show a realistic black scoped assault rifle held in first-person view, rain-speckled and aimed toward the skyline. Add fires, smoke plumes, small explosions, and city lights reflected in the water. A helicopter with a bright searchlight appears in the upper right sky.

Main subject: A colossal semi-transparent anime giantess appears behind and above the skyscrapers, scaled so her head and shoulders dominate the sky. She is {argument name="character name" default="an original anime maid giantess"} with {argument name="hair color" default="pink-purple twin-tail hair with blue streaks"}, large purple eyes, small open mouth, and a mischievous or threatening expression. She wears a frilly maid-style headpiece and sailor-maid outfit with dark blue ribbons. Her left hand reaches forward toward the viewer in extreme foreshortening, with the palm and fingers huge in the foreground, as if trying to grab the city or the player. Make her softly luminous and slightly transparent, like a massive holographic or supernatural apparition in the rain.

HUD and readable overlays: Include exactly 5 distinct game-interface elements: 1 square minimap in the lower left, 1 centered crosshair/weapon status cluster near the bottom, 1 ammo counter panel in the lower right showing large numbers similar to “21 / 84”, 1 small status label above the ammo panel, and 1 row of small icons at the top right. Include exactly 2 centered Chinese subtitle lines near the bottom: {argument name="first subtitle line" default="汉娜：大家快看天上！"} and {argument name="second subtitle line" default="爱尔兰佬：妈的，这是啥玩意！"}. Add a small Chinese objective label near the minimap reading {argument name="objective label" default="抵达撤离点"}.

Counted visible elements to include: exactly 1 giant anime maid, exactly 1 oversized reaching hand, exactly 1 helicopter with searchlight, exactly 1 first-person scoped rifle, exactly 1 Oriental Pearl Tower, exactly 2 subtitle lines, and exactly 5 HUD/interface groups. Do not add extra characters in the foreground.

Visual style: Photorealistic AAA military FPS screenshot with realistic rain, wet surfaces, smoky battlefield atmosphere, cinematic lighting, and authentic UI overlays; contrast the realistic war scene with a crisp anime character rendered as a gigantic translucent figure.

Goal: Create a colorful Japanese manga-style food review comic about a chibi character visiting a Kobe craft beer restaurant and enjoying small-size tasting portions.

Canvas: Vertical 3:4 comic page, bold black panel borders, bright saturated colors, clean digital manga line art, halftone dots and speed-line backgrounds, expressive chibi proportions. The main character is {argument name="character name" default="a chibi black-haired boy with rectangular glasses"}, wearing {argument name="outfit" default="a red beret, orange T-shirt, blue jeans, and black sneakers"}. Use thick outlines, exaggerated facial expressions, and readable Japanese speech bubbles.

Layout: Use exactly 5 comic panels.
1. Top-left panel: The character sits at a wooden restaurant table studying a menu and writing with a pen, looking serious and slightly confused. Add one oval speech bubble reading 「種類が多い…」.
2. Top-right panel: The character walks up a narrow exterior stairway beside a standing yellow signboard that reads “KOBE BEER” with small Japanese text beneath it. The character points excitedly. Add two speech bubbles reading 「神戸の地ビールかー」 and 「珍しいな」.
3. Large center panel spanning the width: The character sits at a wooden table eating from a bowl with a spoon, smiling widely. Show exactly 5 food and drink items on the table: a red craft beer glass labeled “Craft Kobe,” a golden craft beer glass labeled “Craft Kobe,” a bowl of soup or stew, a plate of sliced prosciutto, a small salad plate, and a plate of fried chicken karaage. Add two jagged manga burst speech bubbles reading 「いろいろ楽しむぜ！」 and 「Sサイズで…」. Use a warm yellow radial burst background.
4. Bottom-left panel: Close-up of the happy character laughing with a pink speed-line background. Add two speech bubbles, one oval reading 「唐揚ぎもうまい」 and one jagged reading 「うまーい！ぷはぁ〜」.
5. Bottom-right panel: The character nervously holds a large golden beer glass with foam, sweating slightly, seated at the table with a dark blue uneasy background and wavy stress lines. No extra dialogue except small sweat/emotion marks.

Text content: Preserve all Japanese text exactly as written, including 「種類が多い…」, 「神戸の地ビールかー」, 「珍しいな」, 「いろいろ楽しむぜ！」, 「Sサイズで…」, 「唐揚ぎもうまい」, and 「うまーい！ぷはぁ〜」. The beer sign must say {argument name="beer sign text" default="KOBE BEER"}. The glass labels must say {argument name="glass label text" default="Craft Kobe"}.

Visual style: Playful shonen manga restaurant report, crisp inked outlines, white speech balloons with black Japanese lettering, energetic panel composition, appetizing food illustration, cozy wooden tabletop details, lively but not photorealistic.

Constraints: Use exactly 5 panels, exactly 7 Japanese speech bubbles across the comic, and the listed visible food/drink arrangement. Do not add extra menu writing beyond small unreadable marks, do not add additional characters, and avoid photorealism.

Create a photorealistic 3D studio product render showing exactly three upright infinity-loop sculptures shaped like the number 8, arranged side by side on a glossy reflective gray floor. The left sculpture is transparent clear glass with thick rounded edges, visible refraction, caustic highlights, and subtle internal reflections. The center sculpture is polished mirror chrome metal, highly reflective with dark and bright warped studio reflections across its smooth twisted surface. The right sculpture is white marble with soft gray veining, matte-satin stone texture, and rounded carved edges. All three objects have identical continuous Möbius-like figure-eight forms with an open oval hole at the top and bottom, standing vertically and casting soft shadows plus faint floor reflections. Use a minimalist warm-gray studio background with blurred rectangular softbox light reflections on the left and right walls, diffuse natural lighting, shallow depth of field, centered symmetrical composition, high-end luxury material study, no text, no people, no watermark. Use {argument name="object count" default="three"} sculptures, {argument name="left material" default="transparent clear glass"}, {argument name="center material" default="polished mirror chrome metal"}, {argument name="right material" default="white marble with gray veining"}, and {argument name="background color" default="warm gray"}.

Goal: Create a colorful Japanese manga page about a chibi character enjoying Kobe beer and food, in a bold anime comic style with thick black outlines, expressive speed lines, halftone textures, and clean digital coloring.

Canvas: Vertical 3:4 manga page, approximately 768 x 1024, divided by thick black gutters into exactly 5 panels: 2 small panels across the top row, 1 large full-width centerpiece panel in the middle, and 2 panels across the bottom row.

Character details: Use one recurring chibi manga character in all 5 panels: a small youthful person with spiky black hair, rectangular black glasses, a bright red beret tilted backward, an orange T-shirt, expressive eyebrows, and exaggerated open-mouth reactions. Keep the character consistent across panels.

Panel count and layout:
1. Top-left panel: the character sits at a wooden table studying a menu or checklist with a pen, looking serious and slightly overwhelmed. Add a small sigh puff near the face and one oval speech bubble containing {argument name="top left bubble text" default="種類が多い…"}.
2. Top-right panel: the character runs excitedly through a narrow street or alley with stairs in the background. Place a yellow sidewalk sign reading “KOBE BEER” in large black letters. Include two speech bubbles: one vertical bubble containing {argument name="top right large bubble text" default="神戸の地ビールかー"} and one smaller bubble containing “珍しいな”.
3. Center large panel: the character sits at a table joyfully eating ramen with chopsticks, pulling up long noodles from a steaming bowl, smiling wide with sweat drops of excitement. Around the bowl are exactly 4 visible food/drink groups: a red craft beer glass labeled “Craft Kobe”, a golden craft beer glass labeled “Craft Kobe”, a tall dark beer glass, and three food plates consisting of sliced meat, a small salad, and fried chicken pieces. Use a radiant yellow burst background. Add two jagged speech balloons: left balloon contains {argument name="center left bubble text" default="いろいろ楽しむぜ！"}; right balloon contains {argument name="center right bubble text" default="Sサイズで…"}.
4. Bottom-left panel: close-up of the character grinning with mouth wide open, satisfied and excited, with pink speed lines in the background. Add two speech bubbles: one oval bubble containing “唐揚ぎもうまい” and one jagged bubble containing {argument name="bottom left jagged bubble text" default="うまーい！ぷはぁ〜"}.
5. Bottom-right panel: the character drinks a large foamy beer from a tilted glass with both hands, cheeks slightly puffed, sweat drop on the face, blue vertical speed-line background, wooden tabletop at the bottom. Show splashes or foam droplets near the glass.

Visual style: Cute chibi Japanese comic, high-energy food manga expression, bold black panel borders, crisp line art, saturated colors, playful exaggerated typography in speech bubbles, no realism, no watermark, no extra panels, no extra characters.

Using REFERENCE_0 as the source material, create a dark-mode Japanese evaluation report slide that compares attempts to generate a character sheet while preserving the same rough watercolor/ink illustration style.

Goal: Turn the reference into a structured experiment-summary dashboard titled 「M2　ステップ2—ムードボードを見本に GPT でキャラクターシート」. Keep the subject as the small armored child character from the reference, but present the result as an analysis screen rather than a single illustration.

Canvas and style: Use a wide 16:9 dark UI layout with charcoal background, thin dividers, rounded cards, small Japanese UI text, orange/green accent labels, and embedded white image panels. Make it look like a screenshot of an internal comparison report.

Top section: Create exactly 3 comparison cards across the top. Each card contains a white character-sheet panel showing the same character in multiple views and a small detached head/bust crop. Use these card titles and edge scores:
1. 「経路A：GPT ムードボード経由」 with 「エッジ 57.6」
2. 「経路B：grok ムードボード経由」 with 「エッジ 52.6」
3. 「参考：grok キャラ直接（B-5）」 with 「エッジ 48.8」
Under each card add one short Japanese note line; the third card should mention that it is the softest and inherits the sample’s green cloth.

Middle section: Add a comparison table with exactly 4 columns labeled 「条件」, 「エッジ強度」, 「コントラスト」, 「見本の欠陥」. Include exactly 7 data rows:
1. 「grok（目標）」 — 47.2 — 82.4 — 「—」
2. 「1) 共通のみ」 — 53.6 — 90.8 — 「—」
3. 「2) + 筆致（言葉）」 — 52.0 — 88.4 — 「なし」
4. 「3) + grok キャラ見本」 — 「48.8 ← 最良」 — 85.4 — 「緑の布を継承」
5. 「4) + GPT ムードボード」 — 「57.6 ← 最悪」 — 84.5 — 「なし」
6. 「5) + grok ムードボード」 — 52.6 — 87.3 — 「なし」
7. 「素材 image3.png（参考）」 — 58.8 — 69.1 — 「—」

Bottom section: Add several short Japanese conclusion lines beginning with 「結論1」 through 「結論5」, explaining that moodboard routes are not very effective, direct character reference works better, and the practical workaround is to avoid moodboards and use the Grok output with cropping. Finish with a collapsible dark bar labeled 「プロンプト（ステップ2の冒頭：ムードボードの役割指定）」.

Constraints: Preserve the character identity and rough painted look from REFERENCE_0 in the embedded panels, but do not recreate the full original reference screenshot. Do not add unrelated characters. Use crisp UI typography and readable Japanese labels. The main title topic may be customized as {argument name="report title" default="M2　ステップ2—ムードボードを見本に GPT でキャラクターシート"}; the three route labels may be customized as {argument name="route labels" default="経路A：GPT ムードボード経由 / 経路B：grok ムードボード経由 / 参考：grok キャラ直接（B-5）"}; the character-sheet subject may be customized as {argument name="character subject" default="the small armored child knight from the reference"}; the table metric emphasis may be customized as {argument name="best and worst edge notes" default="48.8 ← 最良, 57.6 ← 最悪"}.

Create a dynamic high-detail anime battle illustration showing {argument name="attacker name" default="Origin-sama"} violently slamming {argument name="victim name" default="a pink-haired magical girl"} headfirst into a cracked stone floor. The attacker is a fierce young woman with split black-and-white hair accented by vivid red streaks, long thick twin braids whipping outward like ropes, small round hair buns, sharp teal eyes, and a black star-pattern mask across her eyes. She wears a black tailored jacket over a white ruffled blouse, a short black skirt with red lining and white frills, black shoes, and a dramatic combat pose with one knee raised while her hand grips the victim’s head. The victim has very long flowing pastel pink hair, a frilly white magical-girl dress with orange trim and ribbons, white gloves, and is shown upside down at the moment of impact, one hand braced on the rubble. Emphasize the hit-reaction scene rather than a clean attack pose: the victim’s hair explodes across the foreground, the ground shatters into a crater, rocks and dust fly in all directions, and bright pink-red energy ribbons spiral around both characters. Use an intense low-angle vertical composition, cinematic foreshortening, motion blur, debris arcs, dramatic lighting, dark smoky background, glowing highlights, and polished modern anime rendering with extremely detailed fabric, hair strands, impact particles, and swirling shockwave effects. No text, no speech bubbles, no watermark.

Create a cinematic photorealistic fine-dining scene in a warm modern kitchen at blue hour: an Asian male chef in a dark apron stands on the left side of the frame, partially cropped, concentrating as he uses tweezers and chopsticks to place tiny herb garnishes on an elegant plated dish. The dish sits centered on a large dark stone kitchen island in a shallow wide ceramic bowl, featuring a glossy seared fish fillet with golden char marks, small yellow vegetable batons, red and green accents, edible flowers, shiso-like leaves, and a small mound of black rice or caviar-like garnish. The kitchen has dark wood cabinetry, black stone backsplash, open shelves with bowls and plates, hanging towels, cooking utensils, bottles, and warm under-cabinet lighting. On the right, include a large panoramic window showing a twilight coastal forest cliff and deep blue ocean/sky outside, with three visible black pendant lights glowing warmly near the window. Use dramatic cinematic lighting: amber interior practical lights contrasting with cool blue dusk outside, soft shadows, realistic reflections on the countertop, shallow depth of field, high dynamic range, rich textures, natural skin detail, and an artful editorial food-photography composition. No text, no watermark, no extra people.

Create a photorealistic outdoor portrait of a {argument name="subject" default="young adult East Asian woman"} crouching on a sunlit rooftop terrace in a quiet hillside town. She has long straight dark brown hair with soft bangs, fair skin, natural makeup, and a calm neutral expression while looking directly into the camera. She wears a slightly loose pale blue-white button-up dress shirt with rolled sleeves, a loosened black necktie, a black skirt, and glossy black pointed high heels. Pose her low to the ground with knees bent apart, elbows relaxed forward, hands hanging naturally between her legs, balanced on her heels. The setting is a white concrete rooftop parapet with a gray tiled floor; behind her is a softly blurred village of traditional rooftops, utility wires on the left, green forested mountains, and a bright pale blue sky. Use natural midday sunlight, crisp highlights on the face and clothing, soft shadows on the floor, shallow depth of field, realistic skin texture, high-resolution DSLR look, 50mm portrait lens feel, centered composition, vertical 3:4 framing. Avoid text, logos, watermarks, extra people, exaggerated anatomy, or overly stylized rendering.