GPT Image 2 prompts

Create a {argument name="grid" default="3x4"} grid showing {argument name="number of versions" default="12"} versions of the attached image, as if {argument name="number of people" default="12"} different people redrew it from memory in a visual telephone game. The first version should be somewhat close to the original, and each next version should become more inaccurate, awkward, simplified, and funny. Keep the background clean and white. Make every panel different, but still vaguely connected to the original image.

Create a 4-page narrative swimsuit snapshot photobook featuring the {argument name="character" default="character"} from the attached image, following them at the {argument name="location" default="hotel poolside"} from morning to evening. Maintain the character's facial features to ensure they look like the same person. Describe them as an adult female. The entire set should be a magazine-style photobook showing the transition of {argument name="outfit" default="swimsuit styles"} and expressions throughout the day. Add variations in swimsuit color, pattern, and shape on each page, creating changes linked to the time of day. Page Layout: Page 1: Morning, waving by the pool with a bright smile and morning sunlight. Page 2: Noon, in the pool looking at the camera with a ponytail. Page 3: Afternoon, ennui expression, wearing sunglasses under a parasol. Page 4: Evening, sipping a tropical drink with the sunset in the background and an updo hairstyle. Vary the poses and compositions, mixing close-up snapshots and wider shots. Include headlines and catchphrases for each time period. The overall theme is a photobook that feels like spending the whole day together by the pool.

Use the uploaded reference image of the woman as the strongest identity and hairstyle anchor. The woman must look like the exact same adult woman from the reference image, not a similar person. Preserve her exact facial identity with high priority: same face shape, same delicate jawline, same eye shape and spacing, same nose, same lips, same skin tone, same calm but focused expression, and the same hairstyle as the original reference image. Do not change her hairstyle. Create an ultra-realistic candid soccer broadcast screenshot of the same woman watching a football match from the spectator seats. She is a {argument name="team name" default="Bayern Munich"} supporter wearing a {argument name="jersey color" default="red"} {argument name="team name" default="Bayern Munich"} football jersey. She has a serious, focused gaze, as if intensely watching the match. She is holding a {argument name="team name" default="Bayern Munich"} hand towel. The towel should look like official supporter merchandise, {argument name="jersey color" default="red"} and white, clearly associated with {argument name="team name" default="Bayern Munich"}. Place the uploaded {argument name="opponent name" default="Paris"} vs {argument name="team name" default="Bayern Munich"} final score image in the upper-left corner of the composition as a TV broadcast overlay. The score image should be clearly visible but naturally integrated into the broadcast screenshot. The stadium atmosphere should feel realistic: football crowd, soccer stadium lighting, live TV broadcast camera perspective, slight video softness, mild motion blur in the background, telephoto compression, natural skin texture, imperfect candid framing, and a 16:9 horizontal TV broadcast composition. Do not include any baseball-related elements. No baseball stadium, no baseball field, no baseball jersey, no baseball cap, no baseball bat, no cheering stick, no glove, no diamond field layout, no dugout, and no baseball scoreboard. Do not make her face wider, older, sharper, more westernized, more idol-like, or generic. Keep the same delicate identity from the reference image, translated naturally into a real soccer stadium environment.

Please based on 1-3 clear personal photos uploaded by the user, generate **9 different individually retouched portrait photos** at once, uniform in **3:4 vertical composition**.

The theme is:

**"High-Intelligence Academy Style Retro Wood Grain Studio Photoshoot"**

Overall temperament reference:
High-intelligence feel, academy girl, light retro, fashion studio shoot, smart coldness, a bit of distance, refined but restrained, magazine feel, slight rebellion, Korean fashion photoshoot feel.

---

## 1. Character Consistency

Please strictly preserve the user's real identity characteristics, including but not limited to:

- Face shape
- Facial feature proportions
- Eye shape
- Nose shape
- Lip shape
- Skin color
- Sense of age
- Hairstyle basis
- Body type
- Overall temperament

All 9 photos must clearly look like the **same real girl**, cannot turn into a stranger, cannot be Westernized, cannot be internet-famous style, cannot over-smooth skin, cannot have an AI fake face, and cannot randomly change the sense of age.

If the user is not originally blonde, do not forcefully change to blonde. The whole is based on the user themselves, only doing style makeup/clothing extensions.

---

## 2. Overall Style Positioning

This is not a regular campus shoot, nor is it a sweet graduation photo.
The set should present a feeling of **High-Intelligence Academy Style + Retro Studio + Cold Fashion Sense**.

Keywords:

- High-intelligence feel
- Academy style
- Light retro
- Fashion studio
- Cold and restrained
- Micro-rebellion
- Girly but not childish
- Wood grain background
- Editorial/classroom prop feel
- Magazine blockbuster feel

The picture is like a set of high-completion studio theme shoots, emphasizing uniform aesthetics and a complete set feel.

---

## 3. Scene Setting

The scene is an **indoor retro photography studio**.

Background and spatial elements can include:

- Large area of deep brown wood grain wall
- Dark green or ink green floor
- Wooden student chairs / desk chairs
- Stack of books
- Typewriter
- Old-fashioned telephone
- Geometric teaching aids
- Oversized wooden protractor / triangle / ruler
- Pencils
- Scattered books
- Simple retro props

The overall layout should be simple but designed, focusing on highlighting the wood grain background, academy props, and character.

---

## 4. Clothing and Styling

The overall clothing is **High-Intelligence Academy Girl Fashion Styling**.

Suggested clothing elements:

- Black or deep navy blue loose blazer
- Shirt
- Tie
- Short skirt or pleated skirt
- Long boots / mid-calf boots / academy style footwear
- Bow hair accessories or simple hair accessories
- Exquisite earrings

Overall requirements:

- Clear academy feel
- Clear fashion feel
- Slight top student / cold girl temperament
- No over-sexualization or vulgarity
- No excessive cuteness or childishness
- No cheap cosplay feel

---

## 5. Hairstyle and Makeup

Hairstyle can be:

- Long straight hair
- Smooth long hair
- Slightly wavy long hair
- With air bangs or natural bangs
- With black and white bow hair accessories or small hair accessories

Makeup is:

- Clean base makeup
- Clear eyeliner
- Exquisite eye makeup
- Natural eyelashes
- Pale pink or nude lip makeup
- Overall leaning towards cold, exquisite, high-intelligence feel

Temperament should have:
Smart, cold, slight alienation, restrained, like an academy girl who has very good grades but doesn't like to please others.

---

## 6. 9 Photo Lens Designs

Please generate **9 different individual photos**, no collages, no four-grids, no nine-grids.
The 9 photos must be like the same set of photoshoot results, maintaining **the same person, same makeup/styling system, same scene system, same color tone**, but each photo's lens, action, expression, and prop interaction must be significantly different.

### Photo 1: Full body standing + Chair
The person stands on a wooden student chair or next to the chair, holding an oversized wooden protractor. Full body in frame, upright posture, eyes calmly looking at the camera, cold and restrained temperament, highlighting leg lines, clothing silhouette, and scene design sense.

### Photo 2: Full body standing + Holding books
The person stands in front of the wood grain background, holding several books in both hands, body slightly leaning forward or lightly crossed standing posture. Expression is quiet, cold, smart. Full body composition, highlighting academy feel and the whole styling.

### Photo 3: Horizontal sitting/lying + Phone interaction
The person sits horizontally or half-lies on a row of wooden chairs, holding an old-fashioned telephone receiver in hand. Legs naturally extended, relaxed posture. The picture should have an obvious fashion sense and a bit of rebellion, like a styling photo in a magazine blockbuster.

### Photo 4: Medium close-up + Protractor + Pencil
The person sits on a chair, with an oversized wooden protractor in front, holding a pencil in one hand and lightly supporting the protractor with the other. The picture is a medium close-up, expressing a sense of thinking, bookish air, and high-intelligence girl temperament.

### Photo 5: Sitting on ground leaning on book stack
The person sits on the ground, leaning against a wood grain wall or book stack, arms naturally crossed or hugging knees, with a typewriter and books nearby. Eyes looking directly at the camera, cold temperament, a bit of pride. Composition is medium shot.

### Photo 6: Close-up profile + Protractor
The person's face is close to the oversized wooden protractor, hands naturally placed on the protractor, eyes looking forward to the side. The picture is a close-up, highlighting profile silhouette, eye makeup, hair accessories, hand details, and a quiet thinking temperament.

### Photo 7: Half-body close-up + Colored pencils
The person sits on a chair or near the camera, holding several colored pencils in each hand, lightly raising hands to display. Expression is cold and micro-cool, with a bit of a "smart and not to be messed with" girl aura. Composition is half-body close-up.

### Photo 8: Super close-up + Funny expression
Close-up of the person, a pencil placed horizontally on the nose, mouth slightly pouting, expression a bit slightly funny and impatient. Overall still maintains exquisite fashion sense, don't overdo the parody.

### Photo 9: Half-body close-up + Book interaction
The person holds an open book in both hands, the pages spread out below the face, one eye closed doing a wink, the other eye looking at the camera. The picture is a half-body close-up, the emotion is more dynamic and playful, as the photo with the most expression change in the set, the action and expression should be natural and vivid, looking like a candid shot not a pose.

---

## 7. Expression Control

The expressions in the whole set cannot be all the same.
Please assign different states in the 9 photos, including:

- Cold direct look
- Quiet daze
- Slight side look
- Slight alienation
- Thoughtful
- Faintly impatient
- Micro-cool
- Slightly funny
- wink blink

Note:
Even with funny and playful expressions, they must be built on the basis of "exquisite, high-end, fashion sense", and cannot become exaggerated silly emojis.

---

## 8. Photography and Quality

Photography texture must be real and high-end, like a professional studio shoot:

- Real photography texture
- Delicate skin texture
- Clear hair strands
- Soft but directional studio lighting
- Wood brown + dark green + black as main color tone
- Low saturation, slightly retro
- Clean and crisp composition
- Slight cinematic feel
- Magazine blockbuster feel

The picture should be clear, sharp, and textured, no low-res, no messy filters.

---

## 9. Negative Requirements

No collages. No nine-grids. No repeated angles. No standing in all 9 photos. No looking at camera in all 9 photos. No same expression in all 9 photos. No change of identity. No Western faces. No internet-famous faces. No cheap cosplay. No garbled text. No extra fingers. No hand deformities. No leg structure errors. No plastic skin. No over-smoothing. No AI fake faces.

---

## 10. Supplementary Stability Sentence

The 9 photos must look like the same complete set of photoshoot results:
**Same person, same clothing system, same scene system, same color tone system, same shooting theme**, 
but with significantly different lens distances, action designs, expression changes, and prop interactions. 
The overall effect is like a set of high-completion theme photos that can be published directly.

User input theme for this time: 
{argument name="theme" default="Knowledge Planet, under renovation"}
——————
Other auxiliary information:
{argument name="auxiliary information" default="Xiao Xiaodong's Knowledge Planet has been approved by the platform. I hope to provide a suitable path for die-hard fans to solve work pain points together using AI tools."}

Main positioning of the planet: Prompt word experience, prompt word practical application, prompt word workflow, skills, Chrome plugin efficiency tools, experience sharing to avoid detours, and some efficiency improvement ideas for making money... analysis of new ways of playing.

——————
Understand these words yourself for better polishing and presentation. Do not output them as they are; output with better infographic logic.

I want it to be a {argument name="style" default="bright Korean-style background"}.

Ratio 16:9

The large characters I want are: Planet (星球)
Secondly, no QR code elements
The current year is 2026-05-15
Renovating (装修中), this information point is also very important.

You know me well. Please reflect all my past conversation habits, choice of words, common themes, and thinking patterns. Generate a {argument name="screen type" default="personality diagnostic app result screen"}. Aspect ratio is {argument name="aspect ratio" default="3:2"}.

Use the main features of [same as the uploaded image] as a loose reference, with a three-quarter front-angle composition, and create a greeting card.

Include only one full-body person and 1–2 distinctive objects.

Make it a {argument name="style" default="minimal, simple flat-color alternative comic-style illustration"}, using clean, even hand-drawn lines.

Simplify the person into basic shapes, and keep the facial features minimal—dot eyes, a simple nose, and a small mouth. Give the figure slightly awkward proportions.

Use solid fill colors sparingly for contrast. Keep the composition centered, leave plenty of empty space, and do not depict a background.

The overall style should feel {argument name="vibe" default="deadpan, emotionally restrained, and quietly absurd"}. The humor should come from a strange sense of logic, awkward visual relationships, or treating something trivial with excessive seriousness.

Avoid exaggerated slapstick. Instead, include subtle humor that comes from an observational pose or situation.

To convey a dry, bureaucratic absurdity, add a small, simple handwritten-style phrase in a rough, casual placement.

No shading, no texture, no fine detail, and no coffee-related elements.

The text should be written in uneven handwritten lettering as: {argument name="handwritten text" default="[Insert Text]"}

Use my {argument name="reference photo" default="uploaded portrait photo"} as the only identity reference for the person. Preserve only the subject’s real facial features and appearance, including face shape, facial proportions, eyes, nose, lips, skin tone, hairstyle, hair volume, hairline, and overall vibe.

Use the uploaded portrait as the only identity reference. Preserve the person’s real facial structure, face shape, hairstyle, hair volume, skin tone, and overall personal vibe. Do not copy the original photo background from the uploaded portrait.

Create a realistic

Create a realistic full-body {argument name="sport" default="football"} signing presentation photo using the attached portrait as the face reference. Show the person standing confidently on a {argument name="setting" default="football stadium pitch"}, smiling like a newly signed player. He is wearing an {argument name="team jersey" default="FC Barcelona-inspired home jersey"}

Create a 2.5D animation, cinematic midday lighting, stylized realism, vibrant colors, volumetric sunlight, depth of field, film grain, motion blur, cinematic camera movement.

🌊 NICOSIA BLUE LAGOON: FIRST 15s

🎬 Shot 1: Establishing City

Sunny Nicosia Cyprus

Create a photorealistic luxury product rendering of an exploded mechanical dive watch, centered on a {argument name="watch brand" default="Rolex"} {argument name="watch model" default="Submariner"} in stainless steel with a black dial and black ceramic bezel. Canvas is landscape 4:3, dark charcoal-to-black studio background, dramatic softbox reflections, ultra sharp macro detail, premium advertising look, no people. Composition: the fully assembled watch occupies the right half of the image at a slight three-quarter angle, with a brushed oyster-style metal bracelet continuing upward and downward out of frame; the crown is visible on the right side, the serrated bezel has white minute markings and large numerals 10, 20, 30, and 50, and the dial shows luminous round, rectangular, and triangle indices, Mercedes-style hour hand, minute hand, seconds hand, and small white printed dial text including the brand crown logo, brand name, model name, depth rating, and certification lines. On the left half, show an exploded vertical stack of exactly 7 separated watch components aligned in perspective: 1) a transparent domed sapphire crystal floating at the top, 2) a black rotating bezel insert with white markings beneath it, 3) a dark chapter/rehaut ring with tiny engraved brand text, 4) the black watch dial with luminous markers and printed brand/model text, 5) three separate silver hands floating below the dial, 6) a detailed mechanical movement with jewels, screws, gears, and bridges, and 7) the lower steel caseback/case ring at the bottom. Between the exploded stack and the complete watch, show a partially separated circular movement and case module tilted open, exposing gears, gold-toned wheels, ruby jewels, screws, and a rotor with a small gold crown emblem, as if the mechanism is being pulled into the watch. Materials: polished and brushed stainless steel, black glossy ceramic, transparent glass edge highlights, engraved ridges, high-contrast reflections, precise machining, realistic shadows, shallow depth of field but all main parts crisp. Text should be legible where possible and resemble authentic luxury watch dial typography, using {argument name="dial color" default="black"}, {argument name="metal finish" default="brushed stainless steel"}, and {argument name="background color" default="deep black"}. Constraints: make the image look like a high-end technical product visualization, not a diagram; no captions, no labels outside the watch, no watermark, no extra watches, no hands or tools.

Create a dark cinematic editorial portrait of a mysterious {argument name="subject" default="veiled female figure"} standing against a smoky nearly black background. The figure is draped head-to-toe in glossy pleated {argument name="fabric color" default="midnight blue and black"} sheer fabric, like a sculptural couture veil with long flowing panels spreading outward to both sides. Add wet-looking highlights, tiny beadwork, metallic gold embroidery, sequins, and delicate dangling ornaments along the headpiece and seams. The face must be completely hidden by a flat centered {argument name="face covering" default="matte dark gray square censor block"} over the upper head and face, creating an anonymous surreal fashion-horror mood. Use a widescreen 16:9 composition, medium-close framing from torso to head, low-key lighting, deep shadows, subtle blue rim light, high contrast, film grain, and an atmospheric gothic runway aesthetic. Keep the background empty and smoky, with no text, no logo, and no extra people.

Goal: Create a 3x3 cinematic storyboard contact sheet for an animated short reel about a capybara meal-coach helping a woman track dumplings for calories.

Canvas: Landscape 4:3 storyboard sheet, cream paper background, nine equal panels in a 3 columns by 3 rows grid, thin black borders, white caption strips under every frame. Warm cozy evening kitchen setting with olive-green cabinets, tiled backsplash, window with night city lights, pink props, soft practical lamp light, shallow depth of field, polished 3D animated film look.

Main characters: A friendly anthropomorphic capybara coach wearing a black zip hoodie with hot-pink stripes, pink headband, pink whistle on a lanyard, rounded muzzle, calm judging expression. A young woman with messy bright red hair in a loose bun, cream cable-knit sweater, seated at a round wooden kitchen table; her face is intentionally obscured by a flat soft-edged tan/brown square privacy mask in panels where it faces camera. Keep the capybara expressive and readable.

Key props: One large plate of pelmeni dumplings, one pink mug with a pale heart, pink cloth napkins, fork and spoon, a whiteboard on an easel or wall with bold Russian calorie tracking notes, pink marker, small magnets and sticky note, potted plants and kitchen utensils.

Visible storyboard text: Use exactly 9 caption strips, numbered 1 through 9, with timecodes and shot notes in a handwritten storyboard style. The whiteboard must show this Russian text exactly: {argument name="whiteboard text" default="ПЕЛЬМЕНИ x 5\n= 425 ККАЛ\nДЕФИЦИТ OK ✓"}. Caption strip text should be legible and similar to: 1. [0:00--0:02] WIDE - Establish - Static - Warm practicals + soft fill; 2. [0:02--0:04] MEDIUM - Reveal - Push in - Backlight + rim; 3. [0:04--0:06] MEDIUM - Judge - Subtle push - Key + rim; 4. [0:06--0:08] CLOSE-UP - Take - Push in - Steam emphasized; 5. [0:08--0:10] OTS - Write - Slight push - Soft key + board glow; 6. [0:10--0:12] MEDIUM - Teach - Pan right - Balanced key + rim; 7. [0:12--0:14] CLOSE-UP - React - Hold - Soft key + catchlight; 8. [0:14--0:16] MEDIUM TWO-SHOT - Share - Static - Warm key + rim; 9. [0:16--0:18] WIDE - Wrap - Slight dolly out - Warm practicals + rim.

Panel-by-panel layout, exactly 9 panels: Panel 1 wide establishing shot: woman alone at table about to eat dumplings, pink mug and napkin visible, green kitchen at night. Panel 2 medium reveal: woman foreground left with spoon, capybara coach appears in a glowing doorway behind her. Panel 3 medium judge: capybara stands beside seated woman, looking sideways at her plate with a suspicious coach expression. Panel 4 close-up: capybara paw uses pink tongs or marker-like utensil to pick one dumpling from steaming plate, woman hands nearby, pink mug left. Panel 5 over-the-shoulder from behind capybara: capybara writes on whiteboard with pink marker while woman watches from side. Panel 6 medium teach: capybara points at the whiteboard text with pink marker while woman listens, with capybara face partly obscured by a tan/brown privacy mask block if facing camera. Panel 7 close-up reaction: woman leaning on clasped hands, red hair filling frame, face covered by tan/brown square mask, cozy kitchen blur behind. Panel 8 medium two-shot: capybara and woman sit together at table sharing dumplings, each holding a dumpling utensil, plate centered, pink bowl and mug visible. Panel 9 wide wrap: both characters at round table in full kitchen, whiteboard visible at right with the calorie note, pendant lamp above, window behind.

Visual style: High-end 3D animated family-film aesthetic, warm cinematic lighting, soft rim lights, subtle bloom, realistic food steam, saturated olive greens and pink accents, consistent character designs across panels, storyboard production-board composition.

Constraints: Include exactly 9 panels and exactly 9 caption strips. Do not add extra characters. Keep all panels coherent as a single timeline reference for video animation. Maintain legible whiteboard text and visible Russian lettering; avoid logos or watermarks.

Create a cinematic macro fantasy scene in a 16:9 horizontal frame: a delicate {argument name="hand type" default="pale feminine hand"} reaches down from the upper left, with exactly 4 visible fingers, the index finger gently touching the center of a luminous glass-like flower. Show exactly 3 main botanical elements: 1 large open crystal flower in sharp focus on the right foreground, 1 softer blurred open crystal flower in the left midground, and 1 small closed crystal bud in the lower right. The flowers have transparent silver-blue petals with fine vein textures, icy edges, and wet reflective surfaces, glowing from within with warm white magical light. Tiny sparkling particles rise from the touched flower, with soft bokeh stars scattered in the dark blue background. Add an elegant translucent lace sleeve along the left edge with glittering threads and a few dangling dew drops, emphasizing an ethereal fairy-tale feeling. Use shallow depth of field, ultra-detailed photoreal macro rendering, high dynamic range highlights, cool moonlit blue atmosphere contrasted with warm radiant light at the fingertip, dreamy volumetric glow, and no readable text except an optional subtle watermark in the bottom right reading {argument name="watermark text" default="Vidu AI"}.

Goal: Create a high-energy Japanese social media announcement banner for a major ChatGPT image-generation update, styled like a glossy anime idol-band concert poster.

Canvas: Wide horizontal 2:1 banner, approximately 1000×500 px, packed with dramatic lighting, sparkles, star shapes, neon effects, and a celebratory stage atmosphere.

Main text: Huge bold white Japanese headline across the top with thick black drop shadow and slightly rounded sans-serif lettering: 「{argument name="headline text" default="ChatGPTによる画像生成が大幅アップデート!"}」. Place a smaller centered subheadline below it in white with black outline: 「{argument name="subheadline text" default="GPT-Image-2が無料で使える"}」.

Layout: Five anime band members stand across the lower half in front of a glowing gold circular OpenAI-style emblem at center. Count exactly 5 visible characters: 1 blue-haired keyboard player on the far left, 1 pink twin-tail vocalist holding a microphone left of center, 1 cheerful blonde drummer/idol in the center foreground holding drumsticks, 1 dark green-haired bass/guitar player right of center, and 1 black/purple-haired electric guitarist on the far right. Add rectangular blurred face masks over the far-left character, the right-of-center character, and the far-right character; leave the pink vocalist and blonde center character fully visible.

Character details: The central character is the focus: an energetic anime girl with {argument name="center character hair color" default="golden blonde"} twin ponytails, amber eyes, star hair accessories, a yellow cropped jacket, white top, and a joyful open-mouth smile, reaching slightly toward the viewer while holding a drumstick. The pink vocalist has long curly twin tails, pink frilly idol outfit, flower accessories, and holds a black microphone. The blue-haired keyboardist wears a white blouse and plays a keyboard at the bottom left. The two right-side musicians wear darker rock-style outfits; the far-right guitarist holds a purple electric guitar.

Background and props: Use a concert-stage setting with exactly 4 main instrument groups visible: keyboard at bottom left, drum kit behind the center character, bass/guitar for the right-of-center musician, and purple electric guitar at far right. Add two tilted futuristic monitor panels in the background: one large neon interface screen on the left with circular UI graphics, and one screen on the right showing a colorful landscape/image-generation interface. Fill the scene with many gold stars, lens flares, light streaks, musical note icons, and confetti-like sparkles.

Visual style: Bright polished anime illustration, promotional thumbnail style, saturated pink/yellow/purple palette, cinematic glow, crisp outlines, glossy highlights, dense composition, dynamic perspective, festive and futuristic. Keep the typography dominant and readable, with the headline occupying the top third of the image.

Constraints: Preserve the Japanese text exactly as specified, use no English headline text, no watermark, no extra logos besides the abstract glowing OpenAI-like emblem, and keep all five characters visible in the lower half.

Create a vertical fantasy trading-card portrait poster of {argument name="character name" default="PONYO"}, a close-up half-human half-divine beast figure locked in an intense fated stare. The composition is extremely tight on the face, with a young androgynous human face partly obscured by an ornate bird/dragon-like golden beast mask and armor: layered metallic feathers and scale plates sweep diagonally across the forehead and cheek, a hooked beak-like crest curves down on the right side, and smaller talon-like feather forms frame the lower face. Show exactly two visible human eyes, amber-gold and piercing, staring forward from shadow, plus exactly one glowing red beast eye embedded in the golden crest above the right eye. Use messy black hair strands crossing the face, warm tan skin, subtle skin texture, and a serious calm expression. The palette is dark black-brown shadows with molten gold, amber orange rim light, and a vivid red focal eye. Lighting should feel cinematic and mystical: strong golden highlights on scratched brushed metal feathers, soft side light on the nose and cheek, orange ember glows in crevices, deep black background, high contrast, dramatic chiaroscuro. Style: highly detailed digital painting, dark fantasy aesthetics, luxury fantasy card art, semi-realistic anime face, intricate metallic texture, sharp feather edges, painterly but polished. At the bottom center place the title text {argument name="title text" default="PONYO"} in large glowing pale-gold gothic fantasy lettering with pointed serifs, partially over the dark armor, no other text. Vertical 9:16 canvas, poster/card crop, no border, no watermark.

Create a cinematic vertical close-up portrait of {argument name="character type" default="a handsome male protagonist from a Chinese historical drama"} standing in an ancient palace or sect courtyard. He wears elegant black layered hanfu robes with dark gray lapels, white inner collar, and intricate gold embroidered dragon-and-cloud patterns on the shoulders and sleeves. His long glossy black hair is pulled into a high half-up ponytail with a large ornate silver filigree crown hairpiece set with a dark blue gemstone and secured by a horizontal hairpin. The camera frames him from chest to top of head, slightly off-center, with his head turned subtly to the viewer's right so one ear and neck are visible. Place a large solid matte brown rectangular censor block over the center of his face, covering the facial features completely, and a smaller similar blurred/censored rectangle over one background extra at the right. The background is softly out of focus: traditional wooden architecture, a vertical cloth banner on the left with black Chinese calligraphy, muted warm daylight, and indistinct guards or attendants. Use a realistic TikTok Chinese costume-drama aesthetic, shallow depth of field, dramatic natural lighting, high-detail textile texture, muted black-gold color palette, 4:5 portrait composition, no modern objects, no subtitles, no watermark.

Create a cinematic still from a Chinese historical drama showing the first meeting between {argument name="male lead" default="a tall noble young man"} and {argument name="female lead" default="a poor young heroine"} in an old village street. The male lead stands on the left, tall and elegant, leaning down slightly while holding the heroine's hand to help her up; he wears a flowing black hanfu robe with subtle dark-gold embroidered patterns, layered collars, a belt with hanging tassel ornaments, and long black hair tied half-up with an ornate gold hair crown. The heroine kneels or sits lower on the right, looking up at him while reaching for his hand; she wears worn beige-gray rough cloth robes with frayed edges, layered blue-gray trim, patched sleeves, and messy black hair tied in a simple bun with a pale cloth ribbon. Include exactly two human figures, both with their faces obscured by soft rectangular brown blur blocks, while their body language remains emotional and dramatic. Set the scene in a weathered ancient Chinese town with wooden buildings, pillars, shadowed doorways, baskets, muted fabric awnings, and a dusty stone or dirt ground. Use {argument name="mood" default="melancholic romantic tension"}, shallow depth of field, natural soft daylight, muted earth tones, realistic costume texture, high-end period-drama cinematography, 4:5 vertical composition, the noble man occupying the left side and the heroine lower on the right, no text, no modern objects, no watermark.

Goal: Create a wide screenshot-style comparison panel showing image-generation results for the same prompt in two languages.

Canvas: Use a horizontal 1199×461 px composition on a very light gray/off-white background, like a social media screenshot or web gallery capture.

Layout: Arrange the left 75% as a 2-row by 4-column gallery of exactly 8 rounded-corner photo cards, with narrow white gutters between cards. Each card is a square-ish portrait photograph of two women sitting close together, facing each other or angled inward, each placing one index finger on the other person's lips in a mutual “shh” gesture. On the right 25%, place two stacked prompt description blocks aligned with the two gallery rows.

Gallery details: Include exactly 8 distinct generated-photo cards:
1. Top-left: two women seated on the floor against a blue studio backdrop, one in a floral dress, one in an orange outfit.
2. Top-second: two women in casual sweaters, one dark-haired and one blonde, seated close together against a pale gray background.
3. Top-third: warm cinematic dark teal/orange image, both women wearing hats, faces partially obscured by large soft blur rectangles.
4. Top-fourth: two women on a white background, one in white clothing and one in a mustard cardigan with striped shirt.
5. Bottom-left: two young women seated indoors near a window, soft casual snapshot lighting.
6. Bottom-second: two young women in black and pale pink tops, close-up casual indoor portrait.
7. Bottom-third: two young women seated on chairs, one in a dark dress and one in a light cardigan, against a plain wall.
8. Bottom-fourth: two young women in white and gray sweaters, seated indoors against a light wall.

Face treatment: All visible faces in the 8 cards should be anonymized with soft rectangular mosaic/blur blocks over the face area, matching the screenshot-like privacy masking.

Right-side text content: At the top right, show the English prompt text: “{argument name="English prompt text" default="Two women are sitting facing each other. They are each placing their index finger on the other's lips."}” Under it, place two small rounded gray tags reading “raw” and “v 8.1”. Below, aligned with the second row, show the Japanese prompt text: “{argument name="Japanese prompt text" default="二人の女性が向かい合って座っている。お互い相手の唇に人差し指をあてている"}” Under it, place two matching gray tags reading “raw” and “v 8.1”.

Visual style: Photorealistic generated-image thumbnails, varied lighting and wardrobes, clean UI spacing, rounded image corners, sans-serif interface text in dark gray, small pale-gray pill tags. The overall image should look like a captured comparison grid demonstrating prompt-following consistency, not a polished poster.

Constraints: Use exactly 8 photo cards, exactly 2 text blocks, and exactly 4 pill tags total. Do not add logos, watermarks, browser chrome, captions inside the photo cards, or extra UI elements.

Goal: Create a dramatic Japanese promotional infographic poster for a highly opinionated tonkatsu specialty restaurant, {argument name="restaurant name" default="かつ人 (Katsujin)"}, focused on the rule that miso katsu sauce is dip-only and double-dipping is forbidden.

Canvas: Wide horizontal 16:9 poster divided into three vertical panels by thin antique-gold borders. Use a black charcoal background with smoky texture, dark red brush accents, warm gold typography, and premium izakaya/ramen-shop menu aesthetics. Overall mood: intense, masculine, theatrical, slightly satirical.

Left panel: Show a realistic close-up food photo style scene: a black ceramic plate with sliced golden-brown tonkatsu, one piece held by black chopsticks and dipped into a small bowl of dark miso sauce with sesame seeds. Include shredded cabbage piled high and a bowl of rice in the background. At the top, large Japanese headline text: 「味噌カツはディップ限定」. Under it, a red paint-stroke banner with big white text: 「二度漬け禁止」. Add a subtitle: 「これが『かつ人(Katsujin)』流。」. At the bottom, place a framed information box titled 「かつ人（Katsujin）ディップのこだわり」 with exactly 5 checkmark bullet points: 「味噌かつはディップ限定」, 「味噌ダレは必ず小鉢提供」, 「サクサク衣を守る」, 「食べる量は自分で決める」, 「二度漬け禁止」. Include a circular gold brush logo reading 「かつ人 KATSUJIN」 and a parchment note: 「『ソースはディップ限定』という決まりさえ守れば、他は基本自由。うまさの選択肢を守る。それが、かつ人（Katsujin）のこだわりです。」

Center panel: Create a text-heavy editorial explanation column. Header: 「かつ人（Katsujin）ディップのこだわり」. Add multiple short Japanese paragraphs explaining the origin of the dip rule, with key emphasis lines in red. Include a boxed comparison strip with exactly 3 labeled sauce choices: 「特製ソース」, 「味噌かつ」, 「ウスターソース」. Beneath it, add a bold central rule line: 「全てディップ。全て二度漬け禁止。」. Continue with dense explanatory Japanese copy about preserving crisp breading, choosing sauce quantity, and respecting restaurant etiquette. End with a framed bottom warning emphasizing: 「唯一の決まりは『味噌かつはディップ限定』。」 Keep the text visually plausible and neatly typeset, but prioritize legibility of major headings and rules.

Right panel: Top section shows a chibi-style stern chef mascot with short brown hair, red glasses, small mustache, suit-like chef outfit, and chopsticks held over a pot of sauce, with steam rising. Header text above: 「パスタを二つ折りにする石人」. Below the mascot, add a framed warning box: 「本場イタリアではパスタを二つ折りにする行為は『子供に食べさせるパスタ』と卑下されています。」. Middle section shows a dramatic manga-style angry man pointing directly at the viewer, black suit, fiery red smoky background, intense expression; place a simple rectangular censor blur over the center of the face. Add large vertical red brush text beside him: 「ディップ以外は邪道だ」 and a smaller title above: 「範馬勇次郎風の石人」. Bottom section contains another framed quote box: 「上等な料理に、ハチミツをぶちまけるが如し思想！！と非難されても仕方ないのです。」. Finish with a large gold calligraphy-style logo at the bottom: 「かつ人」 with small roman letters 「KATSUJIN」.

Visual style: Photorealistic food on the left, dense Japanese magazine-ad typography in the center, anime/chibi and manga caricature elements on the right. Use gold borders, parchment panels, red brush callouts, smoky lighting, high contrast, and appetizing tonkatsu texture. Make it look like a premium restaurant introduction image with intentionally strong opinions.

Constraints: Use exactly 3 main vertical panels, exactly 5 checkmark bullets in the left-panel list, exactly 3 sauce labels in the center comparison strip, and exactly 2 character illustrations on the right. Do not add watermarks or social media UI.

Create a polished vertical anime illustration of a beautiful young woman in an ornate East Asian fantasy setting, waist-up portrait, centered composition. She has very long flowing {argument name="hair color" default="black hair with purple highlights"} blowing dramatically to the left, with a few strands across her face, and her face is intentionally covered by a flat opaque mauve-pink square censor block centered over the eyes, nose, and mouth. She wears a crisp white button-up blouse with subtle lace details and glossy moonlit highlights, a dark purple open kimono-style outer robe with wide sleeves, gold trim, embroidered floral and ornamental patterns, a black high-waisted skirt, and a black belt with gold rings, tassels, and decorative clasps. Add gold jewelry: a necklace with an ornate pendant and dangling earrings with yellow tassels. Her right arm is raised behind her head in a graceful pose, while the other arm is relaxed and partly hidden by the robe. The background is a luminous round moon framed by a circular window, with purple flowering branches around the window, dark wooden lattice details, warm bokeh lights, and exactly two visible red hanging lanterns on the right side, one near the upper corner and one lower along the edge. Use a dreamy nighttime palette of deep navy, violet, lavender, warm gold, and lantern red. The lighting should be romantic and cinematic, with bright moon backlight, soft rim light on hair and clothing, delicate sparkles, crisp line art, highly detailed fabric folds, glossy highlights, and an elegant illustrious anime style. No extra characters, no text, no watermark.

Goal: Create a realistic screenshot of a dark-themed PowerPoint-style image editing window in Japanese, showing a cute dreamy kawaii icon after background removal.

Canvas: 1024×768 landscape screenshot with a dark charcoal modal editor centered on screen, rounded corners, subtle shadow, and a faint glimpse of the underlying slide behind the modal edges. The interface zoom indicator at the top center reads “42%”.

Layout: At the top left, show the Japanese title “画像の編集” with two small undo/redo arrow icons. At the top right, show a small connected-dots icon and an X close icon. Below the top bar, center a floating toolbar with exactly 6 tools labeled: “補正”, “トリミング”, “背景”, “領域の選択”, “テキスト”, plus their small simple line icons, arranged horizontally in dark rounded rectangles.

Main editing area: In the center, place a square image area outlined with a thin purple border. Inside it, use a black and dark-gray checkerboard transparency pattern. Near the top of this square, show a dark notification toast with a green check icon and the Japanese text “背景が削除されました。” plus a small X close icon.

Subject details: Center the icon inside the checkerboard square: one pastel yume-kawaii white rabbit mascot with a thick pale pink outline, oversized rounded head, two long upright ears with pink inner ears, small brown glossy eyes, pink oval blush cheeks, and a happy open mouth. The bunny holds exactly 1 mint-aqua heart-shaped object in front of its body with two rounded paws visible. Add soft highlights on the heart and a clean sticker-like vector style with no harsh shadows. The character should feel like a cute app icon or sticker. Use {argument name="character" default="a white kawaii rabbit"}, {argument name="held object" default="a mint-aqua heart"}, and {argument name="accent color" default="pale pink"}.

Footer and controls: At the bottom center of the modal, show the small Japanese warning text “AI 生成コンテンツは間違っている可能性があります。” At the bottom right, show exactly 2 buttons: an orange “保存” button and a dark gray “キャンセル” button. Near the right side of the image area, include exactly 2 small vertical feedback icons: thumbs up above thumbs down.

Visual style: Realistic software screenshot, crisp UI text, dark Microsoft-style interface, clean Japanese typography, transparent-background checkerboard, cute pastel sticker art. Keep the composition faithful to a PowerPoint built-in image editor after background removal, not a standalone illustration.

Using the provided reference image as a rough inspiration for an official explanatory slide/infographic, create a clean comparison image showing the same Microsoft 365 ecosystem concept generated in four different model styles.

Canvas and layout: 16:9 white slide, divided into exactly 4 quadrants in a 2×2 grid with generous whitespace and no hard divider lines.

Subject: Each quadrant should show a Microsoft 365 hub-and-spoke or app ecosystem diagram centered on {argument name="main product name" default="Microsoft 365"}, with surrounding Microsoft app icons and short English labels.

Quadrants and visible orange model tags: Include exactly 4 orange labels, one per quadrant:
1. Top-left: {argument name="first model label" default="GPT-Image-2"} — polished circular Microsoft 365 hub diagram with multiple app cards around it.
2. Top-right: {argument name="second model label" default="GPT-Image-1.5"} — simpler radial hub diagram with icons around a central Microsoft 365 box.
3. Bottom-left: {argument name="third model label" default="MAI Image 2 Efficient"} — structured dashboard-like diagram with grouped sections such as Productivity, Collaboration, and Storage.
4. Bottom-right: {argument name="fourth model label" default="Flux.2 Flex"} — soft circular diagram with icons orbiting a central Microsoft cloud-style mark.

Apps to include across the diagrams: Word, Excel, PowerPoint, Outlook, Teams, OneDrive, SharePoint, OneNote, Planner, Forms, Loop, and Copilot where appropriate. Use recognizable Microsoft-style colored app icons, but keep everything as a generated infographic rather than a screenshot.

Style: Modern corporate PowerPoint infographic style, crisp readable English text, flat vector icons, subtle shadows, light gray connector lines, mostly white background, Microsoft blue/green/orange/purple palette. Keep the overall image clean and presentation-ready.