GPT Image 2 prompts

Subject:
{argument name="subject" default="Summer picnic by the river"}

Main Figure:
A vertical portrait featuring a woman in her 20s sitting on a reddish-brown and off-white checkered picnic mat in the shade of a tree by a river in midsummer. The subject occupies a large portion of the center to the right of the frame, with both knees pulled up near her chest, legs crossed diagonally, and holding a silver potato chip bag in front of her knees with both hands. The bag enters largely from the bottom left, made of silver glossy material, with the bold black text "{argument name="brand" default="Calbee"}", "Potato Chips", white text "{argument name="flavor" default="Consomme Punch"}", a small "5-vegetable package" in the top left, and "55g" at the bottom clearly legible from the front. In the bottom right, there is a shallow rattan basket containing a clear plastic bottle of unsweetened tea, a small silver and black film camera, a blue and white cloth, and yellowish-green grapes. The scene is depicted as a complete moment of a peaceful summer picnic, including the person, snack bag, drink, basket, and small items.

Person/Expression:
A woman in her 20s with fair, smooth skin and a slender, healthy build. Her soft ash-brown long hair reaches below her chest in loose waves with a fine clumped texture, with thin bangs falling naturally over her forehead, and delicate outlines of sunlight filtering through the crown and tips. Her face is slightly tilted toward her own right while looking directly at the camera, with rounded dark brown eyes, natural parallel eyebrows, pale coral pink cheeks, and glossy pink-beige lips slightly parted, showing a soft expression as if quietly speaking. In her right ear, she wears a small silver hoop and a swinging earring with four white flowers, a thin silver chain with a four-white-flower pendant around her neck, and a delicate bracelet with the same flowers on her left wrist. The skin is not excessively washed out, leaving a natural complexion and fine texture on the cheeks, shoulders, arms, and legs.

Clothing/Pose:
Wears an ivory-colored thin-strapped ribbed knit camisole and short pants of the same color. The camisole has a deep round neck and naturally follows the body, while the shorts are made of soft fabric with thin frills at the hem and a tie string on her left side. She sits with her upper body almost facing forward, her left knee raised high in the center, and her right leg folded in front of it to keep both legs compact. Her arms extend naturally in front of her knees, lightly wrapping around the bottom edges of the bag. Long fingers and pale pink nails are clearly shown, with the joints and number of fingers drawn accurately. The exposure is natural for summer casual wear, with the posture, clothing boundaries, and leg overlap anatomically correct.

Background/Light:
The background consists of a bright green lawn, followed by a calm blue river surface, a low concrete bridge, and distant grayish-blue urban architecture. A thick tree trunk and branches enter from the top right, spreading green leaves at the top and creating cool shade around the person. On a clear afternoon, strong natural light from the top left brightly illuminates the background grass and water, while sunlight filtering through the leaves creates small, soft spots on the hair, shoulders, arms, knees, and mat. The person's face is shown uniformly in the diffused light of the shade, while maintaining catchlights in the eyes and fine highlights on the bridge of the nose, lips, and hair. The river surface in the background has fine sparkling reflections, the grass has gradations from yellowish-green to deep green, and the shade adds cool blue-green nuances.

Composition/Camera:
Vertical 3:4, a slightly wide seated portrait capturing from the top of the head to near the ankles with a small amount of leaves and space above. The camera is slightly higher than the person's eye level, looking down lightly from a close distance. A stable triangular arrangement with the person's face in the top center, raised knees in the center, the silver snack bag in the bottom left, and the rattan basket and tea bottle in the bottom right. Equivalent to 50mm, around F2.8, resolving the face and foreground bag sufficiently while blurring the river, bridge, and buildings with a smooth shallow depth of field. Maintain natural perspective, not distorting the face or limbs with a wide angle. The top of the frame is filled with leaf bokeh and a pale blue sky, while the bottom is dense with details of the checkered cloth and small items.

Texture/Style:
High-definition realistic fashion photo, in the style of a clean and transparent Japanese summer magazine gravure. Avoid excessive skin correction or artificial facial features, and realistically depict fine downy hair, individual strands of hair, the weave of the ribbed knit, the seams of the shorts, the folds and reflections of the silver packaging, the weave of the rattan, the pale amber color inside the clear bottle, the translucency of the grapes, and the fibers of the checkered cloth. The color scheme is based on ivory, bright yellowish-green grass, pale blue river, and silver, balancing the warmth of natural light with the coolness of the shade. Moderate fine contrast, soft highlight roll-off, low noise, natural dynamic range, and a sophisticated film-like finish.

Negative:
People who look underage, immature physique, extra arms or legs, fused legs, missing fingers, extra fingers, collapsed hands, reversed accessories, excessive HDR, anime style, 3DCG feel, watermark.

A 4-view character reference sheet of a {argument name="age" default="9"}-year-old boy with {argument name="hair style" default="messy dark brown hair"}, round curious eyes, wearing {argument name="pajamas" default="navy blue dinosaur-print pajamas"}, bare feet, holding a small red flashlight. 3D animated cartoon style, CGI animated

Create a luxury editorial studio portrait of a confident young {argument name="subject description" default="Black man in his early 20s"} seated on minimalist white geometric blocks, centered in a square composition. He wears an immaculate all-white formal outfit: tailored white suit jacket, white waistcoat with subtle buttons, white dress shirt, white tie, slim white trousers, white socks, and glossy white dress shoes, with one discreet silver wristwatch visible. Pose him seated with legs crossed, one hand resting relaxed on his knee and the other elbow propped on a tall white cube, fingertips lightly touching his temple. Use a calm direct gaze, soft closed-mouth smile, polished grooming, and short neat {argument name="hair style" default="black fade haircut"}. The background is an elegant draped {argument name="backdrop color" default="silver-gray satin fabric"} with soft folds and a premium studio feel. Lighting should be high-end portrait lighting: large softbox illumination from the front-left, gentle fill, smooth skin tones, crisp facial detail, subtle shadows under the blocks, and no harsh contrast. The mood is refined, aspirational, clean, and luxurious, like a fashion magazine or personal branding portrait. Use realistic photography, sharp focus on the face, natural proportions, 85mm portrait lens look, shallow-to-medium depth of field, square crop, no text, no watermark, no extra props.

Goal: Create a clean modern Chinese PPT cover slide about {argument name="topic" default="柠檬的妙用"}, using an international Swiss-style modular grid, fresh lemon-yellow accents, soft pastel blocks, and a bright health/lifestyle editorial look.

Canvas: 16:9 widescreen presentation slide, off-white background, airy margins, crisp vector typography, no watermark. Use asymmetrical balance: large text-heavy left column and large rounded photo composition on the right.

Layout: Left side contains the title stack and four feature cards. Right side contains a large rounded-corner lemon photography panel occupying about half the slide width. Bottom right has a black-and-yellow page label bar. Add subtle vertical baseline/grid ticks along the bottom edge.

Text content: Main title in very large bold black Chinese characters: {argument name="headline text" default="柠檬的妙用"}. Below it, add a short neon-yellow underline, then subtitle: {argument name="subtitle text" default="10页现代信息演示"}. Under the subtitle, place a rounded neon-yellow pill containing exactly 4 category words separated by centered dots: {argument name="category text" default="清洁 · 饮食 · 保鲜 · 美容"}. Bottom page label shows a large black “01” on a neon-yellow rounded rectangle segment and “封面” in white on a black rounded rectangle segment.

Feature cards: Include exactly 4 rounded pastel cards in a row under the category pill, each with a simple black line icon, bold Chinese label, and two-line description: 1) spray bottle sparkle icon, label “清洁”, description “天然去污 / 安全环保”; 2) bowl with lemon slice icon, label “饮食”, description “增添风味 / 丰富营养”; 3) refrigerator icon, label “保鲜”, description “延长保鲜 / 减少浪费”; 4) face profile with sparkles icon, label “美容”, description “自然呵护 / 焕亮肌肤”. Card colors from left to right: pale lemon yellow, lavender, pale blue, pale lemon yellow.

Right image composition: Use realistic high-key photography of fresh lemons on a light stone surface: one large whole yellow lemon with green leaves at the upper right, one half lemon with visible juicy radial pulp in the center foreground, one lemon wedge at lower right, and a softly blurred yellow lemon in the background. Overlay modern rounded translucent geometric panels: a vertical lavender rounded rectangle at upper left with a tiny hand-drawn lemon doodle, and a large translucent neon-yellow rounded square/rectangle overlapping the lower-left of the photo with a green leaf crossing it. Keep the photo corners heavily rounded.

Decorative details: At top left, add a small 3×3 dot motif, exactly 9 dots, using alternating lemon yellow, lavender, and pale blue. Keep all icons monoline black, typography bold and modern, with strong alignment and generous whitespace. Avoid extra text or additional cards.

Create a candid ultra-realistic night street portrait of a doll-like young woman, {argument name="character style" default="cute neighborhood milk-tea shop girl with a delicate vulnerable expression"}, standing at an outdoor food or drink counter on a city sidewalk. She is shown from the waist up in a vertical smartphone photo, slightly turned sideways while looking directly at the camera with large dark eyes, pale smooth skin, soft red lips, and a quiet melancholic expression. She has {argument name="hair color and style" default="messy short black layered hair with long wispy bangs partly covering her face"}. Her outfit is {argument name="outfit" default="an oversized white button-up shirt under dark denim overall straps"}. She holds a smartphone or menu card in both hands near the counter, with glossy pale nails visible. The background is a busy urban night scene with warm restaurant lighting, a menu board on the right, counter equipment, blurred passing cars, traffic lights, street signs, trees, and blue-black evening shadows. Use realistic handheld phone photography, close perspective, shallow depth of field, slight motion blur and grain, natural low-light exposure, warm highlights on her face and clothing, no studio lighting, no posed fashion shoot, no text overlay, no watermark.

Create a cute kawaii sticker illustration of a sitting chibi cat on a plain white background. The cat is mostly white with {argument name="patch color" default="black"} markings: one large rounded patch covering the viewer-right side of the head and ear, small dark patches on both sides of the body, and a dark curled tail on the viewer-right. Give it oversized triangular ears with soft pink inner ears, a tiny black oval nose, a small pouty cat mouth, three whiskers on each cheek, and closed horizontal eyes shaped like {argument name="eye shape" default="ーー"}. Add two pink blush ovals on the cheeks, each with three darker blush strokes. Use thick smooth black outlines, rounded simplified paws with visible toe lines, a white sticker die-cut border around the character, and a very subtle light gray shadow/outline outside the sticker edge. Center the cat, front-facing, symmetrical, minimal, clean vector-like anime mascot style, no text, no background objects.

Create an ultra-realistic close-up portrait photograph of a young {argument name="subject ethnicity" default="East Asian"} woman smiling softly at the camera on a bright city street. Frame her face very tightly, slightly off-center, with long dark brown-black hair blowing across her forehead, eyes, and cheeks in loose strands. She has luminous dewy skin, glossy natural pink lips, subtle makeup, warm brown eyes, and a gentle candid expression. The scene is backlit by strong low afternoon sunlight coming from behind her head, creating a bright starburst lens flare near the upper right side of her hair, horizontal flare streaks across the image, glowing rim light on individual hair strands, and a hazy film-like bloom. Use a shallow depth of field so the urban background is heavily blurred: indistinct buildings, street railings, road, and parked cars in cool blue-gray tones. The left edge should include a soft out-of-focus vertical foreground blur as if a hand, hair, or camera strap partially covers the lens. Color grade with warm golden highlights, soft contrast, slight overexposure, natural skin texture, and nostalgic 35mm film photography aesthetics. Square 1:1 composition, intimate candid snapshot, realistic optics, no text, no watermark, no accessories, no extra people.

Create a square close-up anime portrait sticker of {argument name="character name" default="a cute virtual idol maid girl"} facing forward against a very pale lavender background. Show only the head, shoulders, and upper chest, centered and symmetrical, with an oversized chibi-like head and soft delicate line art. The character has long twin-tails with {argument name="hair color" default="cotton-candy pink hair with bright cyan-blue streaks and blue tips"}, straight blunt bangs, two braided side sections, and large white animal ears rising behind the hair. Add a white frilly maid headband with a small sailor-style cap at the top, displaying a bold black anchor emblem. Include exactly two large side hair accessories: black-and-white frilled rectangular maid clips with pale gray bows and dangling ribbons. Her expression is sad and teary: half-lidded purple-pink sparkling eyes, visible tears at both lower eyelids, flushed cheeks, tiny downturned mouth, and a vulnerable pout. Dress her in a white frilly maid outfit with a high ruffled collar, off-shoulder sleeves barely visible, black trim on the bodice, and one prominent pale lavender bow at the neck. Use a soft pastel VTuber fan-art style, glossy eyes, gentle blush gradients, clean cel shading, subtle highlights in the hair, no text, no watermark, no extra characters, and keep the composition tightly cropped like a digital sticker or emote.

Create a whimsical children's picture-book illustration in a rough hand-painted gouache/oil-pastel style on a 4:3 horizontal canvas. Show a large {argument name="animal character" default="cheetah"} in the left foreground, sitting on a gray-brown rock ledge and facing right, with a rounded cartoon body, golden-yellow fur, many black spots, a striped tail curled around the rock, a white muzzle, and very thick uneven black outlines like crayon or marker. The cheetah is looking across a sunny African savanna toward exactly 4 zebras in the mid-distance on the right, each small, black-and-white striped, walking in a loose line. Add exactly 3 white sky clouds with heavy black outlines: one medium cloud near the top center, one large long cloud near the top right, and one small cloud near the horizon. Above the cheetah's head place a large white cloud-shaped thought bubble with black outline, connected by exactly 2 small thought circles, containing bold black Japanese text: {argument name="thought bubble text" default="前菜、主菜、副菜、デザート…"}. The landscape should have a bright cyan-blue textured sky, a golden-yellow grassy plain with scattered scrub brush, one flat-topped acacia tree on the right horizon, dark green foliage and a tilted tree trunk at the far left edge, and spiky green grass in the lower right foreground. Use visible brush texture, thick impasto-like strokes, playful imperfect shapes, high saturation, simple composition, and a storybook mood, as if it could be a page from a picture book. Avoid photorealism, gradients, thin clean vector lines, extra animals, or extra text.

Create a cinematic ultra-realistic alpine landscape at sunrise: a single young woman with {argument name="hair color" default="brown"} hair in a braid sits quietly on a flat rock at the edge of a perfectly still mountain lake, viewed from behind, wearing a simple {argument name="dress color" default="blue"} knee-length dress. She is the only human figure, small in the composition, centered slightly right of the lake’s vanishing point, gazing across the water toward towering snow-dusted mountains. The scene contains one calm reflective lake, one visible sunburst breaking over the left mountain ridge, and one set of circular ripples spreading from the rock into the foreground water. Surround the lake with rugged gray boulders, grasses, purple wildflowers, dark evergreen trees, and golden autumn larches along the right shoreline. The water should mirror the mountains, sky, trees, sun flare, and the seated figure with crisp but natural reflections. Use a wide panoramic 16:9 composition, dramatic depth, cool blue morning mist in the valley, warm golden sunlight touching the peaks and trees, soft volumetric rays, high dynamic range, rich natural textures, serene contemplative mood, sharp foreground rocks, atmospheric distant mountains, no text, no buildings, no extra people, no boats.

Goal: Create a vintage Chinese idiom visualization poster for {argument name="idiom title" default="掩耳盗铃"}, showing the literal scene of a man covering his ears while trying to steal a ringing bell.

Canvas: Vertical poster, 3:4 aspect ratio, aged beige rice-paper background with subtle stains, worn ink texture, and an antique educational print feel.

Layout: Large black brush-calligraphy title centered at the top. Directly below it, place the English subtitle in small serif capitals: “COVERING EARS DOES NOT STOP SOUND”. Under the subtitle, add a small red outlined label reading {argument name="story label" default="成语事故现场 02"}. The main illustration fills the lower two-thirds of the poster.

Scene: An ancient Chinese courtyard rendered in monochrome black ink with muted sepia shading. On the left, a massive bronze temple bell hangs from a rough wooden beam and iron chains. The bell is tilted slightly, vibrating, with engraved patterns, rivets, and a scalloped lower rim. On the right foreground, a thief in dark traditional clothing crouches low and pulls a rope attached to the bell while pressing both hands over his ears; his posture is tense and foolish, one knee bent forward and one leg stretched back. Behind him are courtyard paving stones, an old wall, a traditional tiled roof gate on the right, and delicate tree branches above the wall.

Annotated elements: Include exactly 3 red numbered callouts: “01” near the hanging bell and sound source, “02” near the thief covering his ears, and “03” near the stone ground in the lower left foreground. Draw thin red leader lines from each number to its target.

Graphic effects: Use many thin red concentric sound-wave rings radiating outward from the bell across the whole courtyard, passing behind the thief and architecture. Add small black vibration marks around the bell. Keep the red lines slightly faded and printed like old ink.

Visual style: Detailed hand-drawn Chinese historical engraving, ink wash and woodblock print aesthetic, high line density, muted black-gray sepia palette with red annotation ink, realistic perspective, weathered archival poster design.

Constraints: Use exactly one bell, one thief, three numbered red callouts, and one set of expanding sound-wave rings. Do not add modern objects, bright colors, cartoon styling, extra characters, watermark, or logo.

Create a vertical anime album-cover illustration in a monochrome deep navy and antique ivory ink style. Show {argument name="character name" default="a delicate blond young man"} leaning out of the central open window of an abandoned old mansion at night, resting his cheek on one hand with a faint wistful smile and pale reflective eyes. He wears {argument name="outfit" default="a formal black three-piece suit, white shirt, and tie"}; his messy light hair is softly backlit and slightly tousled. The mansion facade is symmetrical and richly detailed: a large classical pediment roof with carved trim above, weathered columns and molding around the central window, cracked stone and brick below, and exactly two boarded side windows, one on the left and one on the right, each crossed by rough diagonal wooden planks. Inside the dark room behind him, hint at an old chandelier and shadowy antique interior details. The sky is a starry midnight blue with {argument name="moon" default="a bright full moon"} near the upper right partly veiled by thin clouds. Frame the scene with overgrown leaves and branches around the edges, especially upper left, lower left, and lower right, creating a gothic romantic atmosphere. Use fine manga cross-hatching, engraved linework, delicate stippling, worn paper texture, high architectural detail, subtle moonlit highlights, and a melancholic yet serene mood. No text, no logo, no modern objects, no extra characters.

Using REFERENCE_0 as the character base, expand it into a full illustrated outdoor scene while keeping the same bold black outline, hand-painted texture, and recognizable character design. Change the character from a static standing pose into a walking pose, holding exactly 1 red leash in their right hand, with exactly 1 sweat drop on the face and a slightly tired expression. Add exactly 1 small dog, {argument name="dog breed" default="Shiba Inu"}, walking in front of the character on the leash, happy and energetic with its mouth open and curled tail. Place them in {argument name="setting" default="a sunny riverside path"}: blue river behind them, green trees along the banks, sandy ground with scattered dark stones, and a bright blue sky. Add exactly 1 white cartoon cloud in the upper left. Add exactly 1 white thought bubble in the upper right with exactly 2 small connector dots, containing vertical Japanese text {argument name="thought bubble text" default="あつい..."}. Keep the composition landscape-oriented, with the dog on the left and the character on the right, both facing left as if walking together. Maintain a chunky oil-paint/impasto surface texture, vivid colors, simple cartoon shapes, thick black borders, and no photorealism.

Goal: Create a vertical Japanese recipe poster for {argument name="dessert name" default="ルバーブのチーズケーキ"}, combining a polished anime cooking illustration with a food-magazine infographic layout.

Canvas: Tall 9:16 poster, warm cinematic kitchen lighting, dark brown and gold palette, high-detail semi-realistic anime rendering. Use a rustic European kitchen background with brick walls, copper pots, hanging cookware, window light from the upper left, steam, wooden counter, and elegant thin gold ornamental borders.

Main layout: Top left has the large Japanese headline 「ルバーブの チーズケーキ」, beneath it an elegant cursive English subtitle “Rhubarb Cheesecake,” then a short Japanese tagline in small white serif text. Top right shows {argument name="character name" default="Mei-chan"}, a cute young anime maid chef with short chestnut-brown bob hair, green eyes, a white frilled maid headband, brown maid dress, white apron, green ribbon brooch, and soft friendly smile. She stands at a stove stirring a copper pot of bright red rhubarb jam with a wooden spoon; steam rises from the pot.

Center foreground: A large finished cheesecake dominates the poster on a white plate with gold rim. It has exactly 3 visible layers: a crumbly brown biscuit crust, a thick smooth pale cream cheese layer, and a glossy ruby-red rhubarb gelée topping with visible diced rhubarb pieces. Add a small green herb sprig garnish at the center. Make the cake look photorealistic and appetizing, with shiny highlights on the gelée.

Left recipe area: Add one ornate parchment ingredient card with gold frame. It contains exactly 2 ingredient sections: “Crust” and “Rhubarb gelée,” written mostly in Japanese with small measurements. Include the visible Japanese heading 「材料」 and compact recipe-style typography. Nearby, add one small dark translucent note box with a decorative border describing simmering the rhubarb into a glossy jam.

Additional text callouts: Add exactly 1 vertical dark translucent callout box along the left side of the cake with Japanese tasting notes about the rich smooth cheesecake and sweet-tart rhubarb. Keep it elegant and readable but small.

Bottom tutorial strip: Add exactly 6 numbered step cards across the bottom, each with a small rectangular process photo/illustration and Japanese instructions below. The six steps must be: 1) diced rhubarb and sugar in a pot, 2) red rhubarb jam simmering, 3) whisking a pale cream cheese mixture, 4) pouring filling over the crust, 5) set cheesecake in a round pan, 6) spooning glossy red rhubarb gelée on top. Each card has a gold border, black background, a white number label in the top-left corner, and tiny Japanese explanatory text.

Footer: Below the step cards, add a “Chef’s Tip” area with gold cursive English title, Japanese tip text, exactly 1 chibi portrait of the maid character making a peace sign on the left, and exactly 1 bunch of fresh rhubarb stalks with green leaves on the right.

Text style: Use high-contrast white Japanese serif lettering for headings, gold cursive for English accents, thin decorative dividers, vintage cookbook styling. Preserve the main headline text as {argument name="headline text" default="ルバーブのチーズケーキ"} and the English subtitle as {argument name="subtitle text" default="Rhubarb Cheesecake"}.

Constraints: Do not add logos, watermarks, QR codes, or social media handles. Keep all text integrated into the design. Maintain exactly 6 bottom steps, exactly 1 ingredient card, exactly 1 vertical cake callout, exactly 1 chibi portrait, exactly 1 rhubarb bunch, and exactly 3 cake layers. Make the overall image feel like an elegant anime recipe poster for {argument name="featured ingredient" default="rhubarb"}.

Create a vertical anime-style summer illustration of a slender young man with {argument name="hair color" default="messy silver-white hair"} and pale skin standing in a lush traditional Japanese garden beside a wooden house. He wears a white yukata with small blue cross patterns, a dark blue obi, red-and-black geta sandals, and a sheer translucent black haori draped over his shoulders. Pose him full body in a shallow sunlit stream or wet stone path, holding a wooden bucket in one hand and a bamboo water ladle in the other, while raising his opposite hand to shade his eyes and gaze upward toward the bright sky. The setting is late July midsummer: intense blue sky, towering white cumulonimbus cloud, strong sunlight from the upper left with lens flare, sparkling water reflections, humid glowing air, and tiny drifting light specks. Surround him with dense green foliage, a few hydrangea-like flowers, stones, and exactly two prominent sunflowers: one large blurred sunflower in the lower left foreground and one smaller sunflower on the right near the veranda. Include a traditional Japanese wooden veranda and dark tiled roof on the right side, with open sliding doors and warm shadows. Use highly detailed painterly anime rendering, elegant character design, delicate facial features, golden eyes, soft rim light, transparent fabric texture, cinematic depth of field, vibrant summer colors, vertical 9:16 composition, no text, no watermark.

Using the provided reference image as a manga storyboard/layout base, turn the rough YAML-style name into a polished full-color manga page. Preserve the overall page structure, page number placement, and the Japanese dialogue content, but replace the blue placeholder boxes with finished character illustrations and redraw the panels cleanly.

Goal: Create a finished manga page titled {argument name="scene title" default="夜食"}, page {argument name="page number" default="2/4"}, based on the reference storyboard.

Layout: Keep exactly 4 visible panels: 1 wide top panel, 2 angled middle panels side by side, and 1 wide bottom panel. Use clean black manga panel borders, with the middle two panels divided by a slanted gutter. Add the artist handle centered at the bottom: {argument name="artist handle" default="@_sagyo"}.

Character replacements:
- Replace the “あちゃち” placeholder with a beautiful young woman with long dark hair, expressive shoujo-style eyes, and a brown plaid shirt with a dark vest.
- Replace the “悪魔ファブち” placeholder with a handsome silver-haired demon man, black horns, dark bat wings, a stern expression, and dark clothing.
- Replace the “天使ファブち” placeholder with a smiling silver-haired angel version of the same man, white wings, a glowing halo, clasped hands, and a warm heavenly atmosphere.

Panel-specific rendering:
1. Top panel: Show the young woman in a kitchen/dining setting holding a bowl of tofu topped with green garnish, speaking the same line from the reference in a vertical speech balloon.
2. Middle-left panel: Show the demon man arms crossed against a dark purple background, with two vertical speech balloons matching the reference dialogue.
3. Middle-right panel: Show the angel man smiling with hands together, halo and wings visible, against a bright golden sparkling background with small flowers; keep two speech balloons matching the reference dialogue.
4. Bottom panel: Show the young woman holding a small bottle of shichimi spice on the left, surprised; include one spiky emphasis balloon for “七味も…かける…!”, one scalloped balloon for the angel-side line, and one oval balloon for the final line.

Style: Polished Japanese shoujo manga/manhwa illustration, soft gradients, delicate line art, attractive character rendering, subtle blush, warm indoor lighting, expressive faces, vertical Japanese speech balloons. Keep the dialogue legible and in Japanese, using the original text from the storyboard as much as possible.

Constraints: Do not leave any placeholder boxes or construction labels. Do not add extra panels. Keep the manga page clean, finished, and print-ready.

Using REFERENCE_0 as the manga name/storyboard, convert it into a polished finished manga page while preserving the same overall page structure, panel order, dialogue placement, and page number. Replace the rough placeholder boxes with fully illustrated characters and props, keeping the scene flow from the storyboard.

Goal: Finish the page as a clean, full-color shoujo / josei manga page about {argument name="scene title" default="夜食"}, with professional line art, soft gradients, detailed character rendering, and crisp black panel borders.

Canvas and layout: Keep exactly 4 panels: 1 wide top panel, 2 diagonal middle panels, and 1 wide bottom panel. Add the title 「夜食」 at the top left, page number 「2/4」 at the top right, and the artist handle 「@_sagyo」 centered at the bottom.

Panel transformations:
1. Top wide panel: Illustrate the girl from the storyboard as a beautiful young woman in a warm kitchen, holding a bowl of tofu topped with green garnish. Keep one vertical speech bubble with 「じゃ…お豆腐にめんつゆかけて…！」.
2. Middle-left diagonal panel: Replace the “evil” placeholder with a silver-haired demon version of the male character, with black horns, dark bat wings, a stern crossed-arm pose, and a purple ominous background. Include exactly 2 speech bubbles: 「ちっ…冷奴は正直おいしいので何も言えません…」 and 「せめてネギとかかは載せましょう それくらいの悪は許されます」.
3. Middle-right diagonal panel: Replace the “angel” placeholder with the same silver-haired male as an angel, smiling with hands clasped, white wings, glowing halo, sparkles, flowers, and a warm golden heavenly background. Include exactly 2 speech bubbles: 「まぁ…！夜食界の優等生ですよ…」 and 「えらいです…」.
4. Bottom wide panel: Show the girl again at left holding a small red-capped bottle labeled 「七味」, looking determined. Add exactly 3 dialogue effects/bubbles across the panel: a spiky burst reading 「七味も…かける…！」, a wavy bubble reading 「くっ…今夜は天使側の勝ちですか…」, and an oval bubble reading 「それは悪ではなく彩りです」. Use a simple warm tabletop/background.

Style constraints: Keep the Japanese dialogue legible and vertically typeset where appropriate. Make the finished art much more detailed than the storyboard, with attractive anime faces, soft lighting, clean coloring, and no extra panels or extra speech bubbles. Preserve the comedic angel-versus-devil interpretation of the original placeholders.

Create a photorealistic two-photo vertical split collage of the same adult East Asian woman in soft romantic fashion, designed as eye-soothing social media AI art. Canvas: square 1:1 image, divided into exactly 2 equal side-by-side panels with no visible border, warm indoor lighting, high-detail smartphone-photo realism, natural skin texture, shallow depth of field, cozy feminine mood. Left panel: the woman is seated at a small cafe table at night, looking down and to her left with a calm expression, dark hair in a loose messy updo with wispy bangs, wearing a pale blue-gray lace and knit cutout coquette outfit with crisscross halter straps, long distressed lace sleeves, small bow ties, a ruffled cream mini skirt, and delicate necklace; include exactly 2 visible cafe items on the table: 1 iced latte in a clear glass with black straw and 1 slice of layered berry cake on a blue-patterned plate with a spoon. Background shows a dark window, a white wall, shelves or counter decor, and dried flowers in a vase. Right panel: the woman stands in front of a mirror taking a phone selfie, looking at the phone screen, long dark wavy hair with bangs, wearing a lavender purple floral sheer ruffled coquette set with off-shoulder translucent sleeves, lace-up bodice, ribbon details, and a short layered ruffle skirt; she holds a light-colored smartphone with a cute cat sticker on the case. Background shows a dark restroom or fitting-room door, tiled wall, hexagon floor tiles at the bottom, and a Japanese-style wall sign. Overall styling: soft glossy beauty photography, delicate pastel colors, romantic lace textures, subtle film grain, realistic reflections, flattering composition, no watermark, no extra people, no distorted hands or faces. Use {argument name="subject" default="an adult East Asian woman"}, {argument name="left outfit color" default="pale blue-gray"}, {argument name="right outfit color" default="lavender purple"}, {argument name="setting mood" default="warm cozy nighttime cafe and mirror selfie"}, and {argument name="image style" default="photorealistic smartphone fashion collage"}.

Goal: Create a humorous, ultra-realistic, cinematic vertical split-screen vacation fantasy image featuring the same cute young woman in both panels, with playful oversized beach props and vibrant tropical colors.

Canvas: Vertical 2:3 image, divided into exactly 2 horizontal panels of equal height with a thin clean white divider line between them. Use bright midday sunlight, saturated turquoise water, warm golden sand, shallow depth of field, high-detail textures, and glossy cinematic realism.

Top panel: On a tropical beach, show {argument name="character name" default="a young woman with long dark brown hair"} as a tiny, doll-like person clinging excitedly to an enormous fresh green coconut as if it is a giant beach house. She has a surprised open-mouth expression, sunglasses resting on her head, a pink floral crop top, loose white pants, and white sneakers with pink accents. The coconut is cut open at the top with white coconut flesh and splashing coconut water visible. Add exactly 4 coconut accessories: 1 pink paper cocktail umbrella stuck in the coconut, 1 blue bendy straw, 1 fresh mint sprig, and 1 small rope-and-wood ladder leaning against the coconut. The beach scene includes exactly 3 foreground decorative items: 1 orange starfish, 1 small pale seashell, and 1 pink striped seashell. In the background, include 1 elaborate sandcastle on the left, leaning palm trees, blue ocean waves, distant island mountains, and a bright blue sky with soft clouds.

Bottom panel: Show the same woman lounging luxuriously on a huge yellow inflatable duck float in clear turquoise ocean water. She wears a straw sunhat, round dark sunglasses, the same pink floral top, high-waisted white shorts, and white sneakers with pink accents. Her pose is relaxed with both hands behind her head and legs stretched out. The inflatable duck has an orange beak and wears oversized black sunglasses. Include exactly 1 small fluffy kitten sitting on the float beside her, also wearing tiny dark sunglasses. Add exactly 1 small round wooden tray table on the float holding exactly 2 tropical drinks: 1 orange drink with an orange slice garnish and 1 pink-orange drink with a tiny pink umbrella and straw. Background shows sparkling sea, a palm-lined tropical resort shoreline with small huts, distant hazy land, and bright blue sky.

Visual style: Ultra-realistic but whimsical forced-perspective comedy, crisp focus on the subjects, vibrant vacation postcard palette, detailed water reflections, realistic sand grains, playful luxury travel mood, no text, no logos, no watermark.

Constraints: Keep exactly 2 panels, exactly 1 woman per panel, exactly 1 giant coconut in the top panel, exactly 1 yellow duck float in the bottom panel, exactly 1 kitten, exactly 2 drinks, and avoid adding extra characters or captions.

Create a vertical 3:4 photorealistic AI-art portrait of a young Japanese woman in a warm steampunk clock-themed setting. She stands on the right side of the frame, looking slightly to the viewer's right with a gentle smile, long wavy dark brown hair half-up with straight bangs, and small antique clock-and-gear hair accessories. She wears a cream-white corset dress printed with black Roman numerals and clock markings, gold trim, delicate ruffles, a black choker with a small clock pendant, and exactly 2 large round antique clock ornaments hanging on the front of the bodice with gold chains. The background is softly blurred with golden bokeh lights, brass tones, and one enormous vintage clock face behind her on the upper right, using shallow depth of field and cinematic warm lighting.

On the left side, she holds a transparent acrylic signboard with rounded corners, visible screw caps at the corners, and handwritten Japanese text in black, purple, blue, and a small yellow sparkle doodle. The signboard should be clear enough to read, with a realistic hand holding its left edge. Include exactly 5 visible screw caps on the acrylic board: one near each corner and one along the upper edge. The board text should read:
「AI が勝手に考えた
今日のテーマ♡
“ジャネーの法則”

子どもは初めての経験が多くて
記憶に残りやすいのに対し、
大人は慣れた日常が多くて
感動が減るため、時間が
早く過ぎると感じます。
体感上の人生の半分は
『20歳』で終わると
言われています。

おすすめ
・新しいことにチャレンジする
・いつもと違う道を通ってみる

衣装のポイント3
時計の歯車と数字で
時間の流れを表現！

※間違えてたらごめんなさい」

At the lower right of the signboard, draw exactly 1 tiny simple doodle of a smiling girl face. The image should feel like a polished Japanese social-media AI art post: elegant, cute, educational, warm, detailed, high-resolution, with realistic skin texture and glossy acrylic reflections. Use the customizable subject as {argument name="character description" default="a young Japanese woman with long wavy dark brown hair and soft makeup"}, the outfit theme as {argument name="outfit theme" default="cream-white steampunk clock dress with Roman numerals and antique clock ornaments"}, the board title as {argument name="board title" default="ジャネーの法則"}, the background as {argument name="background setting" default="warm antique clock workshop with golden bokeh and a giant clock face"}, and the mood as {argument name="mood" default="gentle morning greeting, elegant and hopeful"}.

Create a cinematic widescreen sci-fi illustration of a humanoid artificial intelligence figure in right-side profile, reaching out with one delicate robotic finger to touch a glowing microchip on a vast dark circuit board. The AI figure has a smooth pale {argument name="robot material" default="porcelain-white metallic"} face and hand, elegant biomechanical panel lines, etched circuit-like grooves, exposed layered contours around the head, cheek, neck, ear, and fingers, and a calm focused expression. The fingertip makes contact with the central processor, producing a bright warm {argument name="contact glow color" default="golden-orange"} flare and small lens bloom. The background is a dense motherboard landscape filling the left and center of the frame, with one square microchip near the center-left, countless copper traces radiating outward, tiny components, and many red-orange LED points scattered across the board like a neural network. Use a dark teal, black, and gunmetal palette with warm sparks and glow accents, high-detail 3D render quality, shallow depth of field, dramatic side lighting from the chip, subtle volumetric haze, futuristic atmosphere, and a composition where the robot occupies the right third while the circuitry expands across the left two thirds. Format as {argument name="aspect ratio" default="wide 16:9 cinematic banner"}, ultra-detailed, sharp focus on the fingertip and chip, no text, no logos, no watermark.

Create a realistic vertical portrait photo of a beautiful young South Asian woman standing beside a bed in a warmly decorated birthday-themed bedroom. She has fair skin, long straight light brown hair parted slightly to one side, soft natural makeup, red lipstick, and a calm direct gaze at the camera. She wears an elegant traditional Pakistani salwar kameez in {argument name="outfit colors" default="royal purple and mustard"}, with intricate golden embroidery on the neckline, sleeve cuffs, hem, and dupatta border; the kameez has small white dotted motifs, matching purple trousers, and a sheer purple chiffon dupatta draped over one shoulder and held at the front. The room is cozy and festive: a bed with a beige tufted headboard and pale floral bedspread sits on the right, with wrapped gifts or colorful pillows on it; behind her are curtains with floral and geometric patterns. Include exactly 2 visible balloon clusters: one cluster of red balloons near the upper center-right and one cluster of red and white balloons in the upper right. Add red metallic letter balloons spelling a birthday message partially visible behind her, with the readable letters resembling “H B D”. Use warm indoor lighting, shallow depth of field, soft smartphone portrait realism, slightly creamy skin tones, and a celebratory home-bedroom atmosphere. Add a subtle TikTok-style watermark near the lower right with {argument name="watermark text" default="@userot8jfzpcc4"}.

Goal: Create a 16:9 cinematic historical PowerPoint cover slide about the {argument name="historical event" default="Wenxi Fire in Changsha, 1938"}, using a heavy archival collage aesthetic that feels solemn, painful, and reflective.

Canvas: Wide presentation slide, 1200×675 or 16:9. Use an aged beige paper background with faint grid lines, scratches, stains, fold marks, dust, and worn print texture. The overall palette is sepia, charcoal black, brick red, smoky gray, and deep teal.

Layout: Design one dense poster-like slide with exactly 5 main visual zones: 1) a narrow vertical brick-red strip on the far left with faint city imagery and vertical Chinese text, 2) a huge central title occupying the left and middle of the slide, 3) a burning ruined city panorama across the lower half and right center, 4) a torn-paper map collage in the upper-right corner, and 5) a dark teal torn-paper information block in the lower-right corner.

Text content: At the top-left, set small bilingual/location text: “长沙  CHANGSHA”, then “1938”, then “WENXI FIRE”, separated by thin horizontal rules. The main title must be enormous Chinese typography reading “文夕大火”, with “文夕” in distressed black and “大火” in distressed dark red. Under the title, place a beige label strip reading “1938年长沙的伤害、余痛与重生”. Below it, add a smaller subtitle: “一场由战时决策失控引发的城市灾难”. Under that, add a small black rectangular tag reading “课件视觉图组”. In the upper-right, add a bold “1938” with a small red label “WENXI FIRE”, plus a vertical framed Chinese stamp-like label reading “长沙文夕大火”. In the lower-right, add a large page indicator “01 / 10”. Include one short vertical handwritten-style Chinese note over the teal panel, subtle and atmospheric.

Subject details: The city scene should show old Changsha streets in ruins, dark silhouettes of people walking, damaged traditional and early modern buildings, telephone poles, rubble piles, and a large building burning with orange flames and thick black smoke rising into the sky. Add torn collage edges where the city photo meets the paper background. Include exactly 1 prominent fire plume, exactly 1 large smoke column, exactly 1 rubble pile in the lower-left foreground, exactly 1 torn map fragment in the upper-right, and exactly 1 teal torn-paper block in the lower-right.

Visual style: Heavy historical aesthetics, archival documentary poster, Chinese modernist editorial design, distressed letterpress typography, old newspaper collage, torn paper, grainy halftone, scratches, muted contrast, somber memorial mood. Use bold geometric Chinese title characters with rough ink erosion and printed wear. Keep the composition dramatic but readable.

Constraints: Do not make it look modern, glossy, or clean. Avoid extra decorative icons, people portraits, watermarks, or random English filler text. Preserve the exact slide/page feeling and the solemn historical tone.

create a premium studio portrait using my uploaded face as the exact facial reference. Preserve my identity with maximum accuracy, including my face shape, hairstyle, beard, eyebrows, skin tone, lips, jawline, and all unique facial features. Do not change my age or recognizable appearance. I am wearing a {argument name="clothing" default="fitted black ribbed knit sweater and matte black rectangular sunglasses"}. Medium-length textured hair with natural volume. Pose with my body turned slightly to the side while looking confidently away from the camera. {argument name="background" default="Deep crimson red gradient background"} with {argument name="lighting" default="dramatic cinematic red rim lighting"} around my hair and shoulders, soft key light on the face, subtle shadows, ultra-clean luxury editorial aesthetic, high contrast, shallow depth of field, razor-sharp facial details, realistic skin texture, premium fashion campaign look, centered composition, 85mm lens, f/1.8, photorealistic, 8K, IMAX quality, no text, no watermark, no logo.