GPT Image 2 Prompts

A dreamy vertical spring portrait photo of a young woman standing in profile in a vast field of tiny blue nemophila flowers under a bright blue sky with soft white clouds. She wears a fitted ribbed knit dress in {argument name="dress color" default="pale sky blue"} with elbow-length sleeves, and her long straight brown hair is blowing dramatically to the right in the wind. Frame her from about mid-thigh upward, slightly off-center to the right, with the flower field filling the background and a small rope fence with wooden posts visible along the lower right edge beside a gravel path. Use soft natural morning sunlight, airy high-key exposure, gentle pastel tones, shallow depth of field, subtle lens bloom, and a romantic social-media snapshot aesthetic. Add a few large out-of-focus white flower petals drifting across the foreground. Overlay handwritten white Japanese text directly on the image: at the upper left, "{argument name="greeting text" default="おはよう"}" with a small doodled cloud beside it; below that, the date "{argument name="date text" default="2026.04.22"}"; across the lower middle-right, "{argument name="main note text" default="嫁と ネモフィラデート✌︎"}"; and beneath it, "{argument name="sub note text" default="に使っていいよっ♡"}". Keep the handwriting casual, playful, and slightly uneven like a marker pen on a photo.

{"type":"hand-drawn Chinese infographic poster","style":"retro risograph zine aesthetic, textured off-white paper background, rough brush lettering, sketchy pen lines, halftone shading, limited ink palette of deep purple and teal with slight misregistration, energetic doodles and arrows","topic":"AI cognition externalization trilogy","headline":{"text":"AI的认知外化三部曲","position":"top center","style":"large expressive Chinese brush lettering, with the letters AI oversized in teal and purple"},"layout":{"flow":"left to right, three-stage evolution diagram connected by arrows, with a large sweeping curved arrow across the bottom pointing right","sections":[{"title":"2022-2023 推参数","position":"left","count":4,"labels":["大脑","GPT-4","Gemini","DeepSeek"],"scene":"a large illustrated brain containing a tall stack of rectangular layers like model parameters or books, with three small pinned labels on the left side reading GPT-4, Gemini, and DeepSeek; the large word 大脑 appears prominently inside the brain"},{"title":"2023-2024 递纸条","position":"center","count":5,"labels":["提示词","检索信息","AI","提示词 + RAG","用完即弃"],"scene":"a human hand extends from the left holding a note toward a side-profile human head silhouette labeled AI; two paper notes are visible, one labeled 提示词 with smaller handwritten Chinese lines beneath, and one labeled 检索信息 with smaller bullet-like Chinese lines; below the head is the phrase 提示词 + RAG, and near the bottom right is a small trash can with the label 用完即弃; a dashed line curves toward the trash can"},{"title":"2024-至今 搭基础设施","position":"right","count":5,"labels":["AI","记忆库","技能库","MCP","沙箱"],"scene":"a head silhouette labeled AI embedded inside a scaffold-like technical framework with pipes, connectors, and platforms; four surrounding modules are attached: a cylindrical database tower labeled 记忆库, a toolbox-like container labeled 技能库, a rectangular node card labeled MCP, and a cube-like container labeled 沙箱"}],"centerpiece":"the three phases of AI evolution from internal model scaling to prompt-based interaction to infrastructure-based cognition externalization","bottom_banner":"从内卷大脑 → 到借纸条 → 再到搭基础设施 认知外化，AI 才真正拥有了翅膀！"},"objects":{"left_section_details":"brain drawn as a cloud-like organic form, stacked slabs in the center, small lightning doodles around it","middle_section_details":"the prompt note includes the heading 提示词, the retrieval note includes the heading 检索信息, the human head is a simple side silhouette, and the interaction is shown as feeding notes into the AI head","right_section_details":{"memory_module":{"title":"记忆库","visible_text_items_count":4,"items":["长期记忆","用户偏好","历史对话","知识沉淀"]},"skills_module":{"title":"技能库","visible_text_items_count":4,"items":["工具集合","API调用","工作流","插件"]},"mcp_module":{"title":"MCP","visible_text_items_count":4,"items":["通信协议","能力发现","连接服务","安全控制"]},"sandbox_module":{"title":"沙箱","visible_text_items_count":4,"items":["隔离执行","代码运行","环境隔离","安全防护"]}}},"decorative_elements":{"count":8,"items":["sparkles","star doodles","radiating lines","lightning bolts","ink splatters","scribble underlines","small motion marks","curved directional arrows"]},"composition":"wide landscape poster, dense but readable, playful and conceptual, combines editorial illustration and systems diagram, all text in Chinese, visually communicates the evolution of AI systems across three time periods"}

A vintage hand-drawn Chinese infographic on textured off-white paper with torn deckled edges, printed in a limited two-ink palette of deep purple and bright cyan accents, with halftone shading and poster-like comic line art. The composition compares intuitive manual cooking versus tool-assisted cooking and externalized cognition. On the left, a large adult cook in an apron is dramatically stir-frying in a wok with flames, flying ingredients, smoke swirls, and energetic motion lines; the cook’s face is obscured by a rectangular blur block. Above the head is a glowing brain-like thought bubble containing 3 small icons: a seasoning bottle, a flame, and a timer/clock, with rays radiating outward. At the lower left, a broken chain hangs beside large bold Chinese text reading “全凭手感”. Below the cook, a smaller confused person mixing in a bowl looks worried, surrounded by exactly 4 question marks. In the center, multiple thick arrows point from the cook’s intuition toward a structured cooking workflow. At upper center is large Chinese text “照着做”. Beneath it, a hand holds a smartphone recipe screen showing visible Chinese text including “5克盐” and “中火3分钟”, with a finger tapping the screen. Around the phone are measured cooking tools and ingredients: 2 measuring spoons, 1 measuring cup, 1 digital scale, and 6 small bowls arranged neatly, plus a pot on a stove with steam rising and a finished plated dish nearby. Near the plated dish is Chinese text “七八分” and a 5-star rating row with exactly 4 filled stars and 1 empty star. On the right, show a concept map of externalized cognition around a simple smiling stick-figure cook standing behind a table of food. The node network is connected with cyan lines and includes exactly 6 tool/info nodes: a smart oven at the top, 2 seasoning packets labeled “调料包” and “榨酱包”, 1 clipboard labeled “精准配方”, 1 blender, 1 rice cooker, and 1 recipe note sheet. Centered in this network is large Chinese text “认知外化”. The table in front of the figure holds exactly 5 dishes/bowls/plates of prepared food. Along the bottom is a simplified cognition equation: a large brain icon, then an arrow to a smaller brain icon, then a cluster of external tools/documents including a phone screen, a microwave-like appliance, and 3 paper/manual items, then an equals sign, then a muscular human silhouette with gears inside the torso and an upward arrow near the shoulder. Beneath or beside this equation is bold Chinese text “脑 + 外部工具 = 更强能力”, with the first “脑” also shown separately under the left brain icon. Use expressive sketched arrows, doodle lightning bolts, comic emphasis marks, and a balanced left-to-right flow that visually communicates the transition from pure instinct to repeatable systems and augmented ability.

{"type":"anime food comic page","style":"clean manga-like digital illustration with warm evening lighting, glossy food rendering, detailed ramen shop background, soft pastel character coloring, expressive comedic Japanese gourmet comic tone","format":"vertical four-panel comic with black panel borders","subject":{"main character":{"gender_presentation":"young woman","age_appearance":"late teens to early 20s","hair":{"color":"{argument name=\"hair color\" default=\"golden blonde\"}","style":"twin ponytails with curled ends, straight bangs"},"outfit":"light gray floral button-up cardigan or knit top, black shoulder bag","face_visibility":"face obscured by a centered rectangular blur censor in panels 1 and 2, fully visible in panels 3 chibi inset and 4"},"theme":"craving and enjoying rich Iekei ramen"},"layout":{"panel_count":4,"panels":[{"index":1,"scene":"outside a ramen shop at dusk on a city street","background_elements":{"count":6,"items":["large red lantern labeled 家系","red shop noren labeled ラーメン","standing signboard labeled 濃厚豚骨醤油ラーメン with ramen photo","glass storefront entrance","urban evening street with pedestrians","speech balloons with vertical Japanese text"]},"character_pose":"standing in front of the shop, one finger near lips, looking tempted","text":{"count":4,"items":["家系","ラーメン","濃厚豚骨醤油ラーメン","speech balloons reading 家系ラーメン抗えなくなる響き・・・ and これは食べてくっきゃないでしょ！"]}},{"index":2,"scene":"inside the ramen shop at the counter","background_elements":{"count":10,"items":["water pitcher","water cup","chopsticks container","red condiment jar","small soy sauce or seasoning bottle","black menu board with ramen prices","vertical sign ライス無料","wooden counter","chair","blurred wall poster in the back"]},"character_pose":"seated at the counter drinking water from a glass with one hand while holding a menu with the other","text":{"count":3,"items":["ライス無料","menu heading ラーメン","speech text ライスいっちゃうか・・・"]}},{"index":3,"scene":"close-up hero shot of the ramen bowl","food":{"dish":"Iekei ramen","components":{"count":5,"items":["thick glossy broth","medium-thick yellow noodles","4 overlapping slices of braised chashu pork","dark green spinach","3 large sheets of nori"]}},"background":"radiant orange burst with sparkles","extra_character":"small chibi head of the girl at bottom right, smiling excitedly","text":{"count":1,"items":["speech balloon これこれ〜待ってました！"]}},{"index":4,"scene":"the girl happily eating ramen at the table","food_and_drink":{"count":5,"items":["main ramen bowl","separate bowl of white rice","glass of beer","floating garnish illustration of nori and chashu near the text","chopsticks lifting noodles"]},"character_pose":"eyes closed with bliss, smiling, cheeks slightly flushed, lifting noodles with chopsticks and holding bowl with the other hand","background":"soft pink-orange gradient with sparkles and small flower motifs","text":{"count":2,"items":["large vertical slogan 今日はライスもつけてチートデイ！","speech balloon DNAに素早く届く♥"]}}]},"camera":"panel 1 medium shot, panel 2 medium interior shot, panel 3 extreme food close-up, panel 4 medium close-up eating scene","rendering_notes":"make the ramen especially appetizing with glossy fat sheen on the pork, rich broth highlights, and realistic noodle texture; keep the comic readable with authentic Japanese signage and speech balloons; emphasize cozy evening ramen shop atmosphere and playful indulgent humor"}

create the image of lymphocytes, neutrophils and basophils as they would be seen under the microscope after giemsa staining

Using REFERENCE_0 as the visual foundation, expand the single sky-palace scene into a complete cinematic world-building presentation for a fantasy civilization. Preserve the same East Asian celestial imperial design language, floating architecture above clouds, gilded ornamentation, warm sunset lighting, and ultra-detailed realistic concept-art look, then turn it into a polished multi-panel lore board on a dark background with elegant editorial typography. Include 8 labeled sections: 1 large hero overview panel for the civilization, 1 architecture section with 9 images, 1 characters section with 7 full-body character designs, 1 clothing and armor section with 6 outfit variations, 1 vehicles and transport section with 4 vehicle designs, 1 world map section with 1 illustrated map, 1 environments section with 6 environment images, and 1 culture and details section with 6 close-up detail images. Add a refined fantasy title and short lore paragraph for the civilization, using the name {argument name="civilization name" default="Celestial Dynasty"}. The architecture section should show 9 distinct items: palace, temple, gate, pillar, bridge, throne hall, garden, library, and observatory. The characters section should show 7 archetypes: emperor, empress, high priest, general, royal guard, scholar, and artisan. The clothing and armor section should show 6 designs: ceremonial robe, battle armor, noble attire, priest robes, artisan garments, and dancer costume. The vehicles and transport section should show 4 designs: an airship, a flying beast mount, a sky chariot, and a ceremonial palanquin. The environments section should show 6 scenes: one wide banner panorama plus 5 smaller locations, including forest, mountain peaks, lotus valley, golden fields, and a celestial port. The culture and details section should show 6 images: dragon motif, lantern or artifact, palace interior, scholarly scene, mechanical instruments, and decorative symbols. Keep all panels visually unified as if from one official franchise world bible, with extremely rich detail, believable material design, and cinematic realism.

Using the provided reference image as the visual civilization seed, expand it into a new cinematic world scene that keeps the same grand East Asian fantasy design language, monumental scale, intricate architecture-first worldbuilding, and ultra-detailed realism, but shift the setting from a luminous cloud palace into an ancient mechanical ruin on the ground. Replace the gilded celestial temple environment with colossal circular ring structures, giant embedded gears, heavy industrial stone-and-metal architecture, broken conduits, and overgrown ruins with pale flowering trees. Change the mood to cold, misty, desaturated blue-gray lighting with soft atmospheric fog, while preserving the epic, awe-inspiring worldbuilding feel from the reference. Add 1 lone wandering swordsman seen from behind in the foreground, wearing flowing white robes and carrying a sheathed sword, centered beneath the largest ring to emphasize scale. Make it feel like a frame from a high-budget fantasy film, with extremely rich environmental detail and a cohesive visual language derived from the reference image.

创建一张 {argument name="课程主题" default="Crypto"} 的选修课考试卷，真实照片感

William and Piggy shaped ice cream, pink square ice cream with pixel eyes and pig noses, strawberry flavor, served on a cone, cute food design. Egg Cone: A standard waffle cone, golden and crispy in color, with a “Minecraft” co-branded paper sleeve. The background is an outdoor theme park scene with a blue sky, white clouds, and brightly colored rides and buildings blurring into the background. The hand of the person holding the ice cream is clear, enhancing the realism of the “hands-on experience”. The overall style is bright, fun and playful, making it an eye-catching IP co-branded summer ice cream.

根据佛经对{argument name="角色" default="观音菩萨"}的形象描述，原原本本的还原一张真实的{argument name="角色" default="观音菩萨"}形象照片，皮肤与衣服接近真实，画质{argument name="画质" default="iPhone 15 pro"}

A clean minimalist tech logo composition on a light gray background, wide horizontal layout. On the left, show a stacked cluster of 4 overlapping Microsoft Office-style app icons with soft gradients, rounded corners, subtle depth, and glossy modern vector shading. The front icon is a large blue square tile with a smaller raised blue panel containing a bold white capital P. Behind and slightly above it, place an orange presentation icon with a white pie-chart symbol. Behind the blue icon to the right, place a white document page icon with a folded blue corner and 4 horizontal blue text lines. At the back on the right, place a green spreadsheet icon with a white grid. Arrange the 4 icons so they overlap tightly and read as a unified hybrid Office app mark. On the right side, place the product name in a clean sans-serif font: small black text reading Microsoft above a large multicolor wordmark reading {argument name="headline text" default="PowerWordEx"}. Color the word in segmented Microsoft-style hues: warm orange-red for Power, deep blue for Word, and teal-green for Ex. Center the whole composition vertically, keep generous whitespace, and render it as polished corporate branding mockup, crisp vector illustration, soft shadowing, no extra objects.

A clean corporate parody graphic on a plain light gray background, centered horizontally, styled like an official Microsoft product banner. On the left, show a layered cluster of 4 overlapping Microsoft Office-like app icons with glossy 3D depth and soft shadows: 1 large blue square tile in front with a bold white capital letter P, 1 orange rounded rectangle behind it at the upper left with a white pie-chart symbol, 1 white document sheet in the middle with a folded top-right corner, three blue horizontal text lines and a blue bar chart at the bottom, and 1 green spreadsheet tile behind on the right with a white grid pattern. On the right side, place the word Microsoft in small dark gray sans-serif type above a large product-name line that reads {argument name="headline text" default="Power harassment Ex"}. Render the large title in a Microsoft-style multicolor gradient spanning warm orange-red into purple-blue and cyan, with a crisp modern sans-serif look. Keep the composition minimal, wide, and balanced, with lots of empty space, resembling a polished tech company branding image or fake software logo announcement.

A clean corporate editorial infographic illustration for a Japanese business article about manufacturing industry challenges, set inside a realistic modern factory with metalworking machines and a cool gray-blue color palette. The composition is split into two thematic halves over a blurred industrial workshop background. At the top center, place a large bold Japanese headline in dark navy text: {argument name="headline text" default="製造業が直面する根深い課題: 「人手不足」と「技術伝承」"}. In the left foreground, show 1 factory worker in a light gray uniform and cap standing with arms crossed, looking concerned; nearby add a circular white icon showing exactly 6 human silhouettes, with 3 filled dark figures and 3 dotted outline figures to symbolize labor shortage. In the right foreground, show exactly 2 factory workers beside a large lathe machine: 1 veteran worker in a navy work jacket and navy cap pointing and demonstrating machine operation, and 1 younger worker in a light gray uniform and cap observing closely. Near them, add a white thought bubble icon containing a navy head silhouette with 2 gear symbols and exactly 3 small Japanese word bubbles reading "経験", "勘", and "コツ". In the center background, include exactly 2 additional workers in light gray uniforms operating machinery, slightly out of focus, for a total of 5 workers visible in the factory scene. Overlay two rounded white information panels with dark navy title tabs. The left panel title is "人手不足" and contains exactly 3 bullet points with blue circular check icons: "若手人材の確保が困難", "現場の負担増大", and "将来の生産性低下への懸念". The right panel title is "技術伝承" and contains exactly 3 bullet points with blue circular check icons: "熟練技術が属人化", "暗黙知が言語化・標準化されていない", and "ベテランの退職による技術の断絶リスク". At the bottom center, place a dark navy rounded rectangular callout box with white Japanese text: {argument name="callout text" default="製造業の持続的な成長のために、これらの課題解決が急務"}. Add exactly 2 large white arrows pointing inward toward this bottom callout, one from the left and one from the right. Style the whole image like a polished consulting-firm presentation visual or business magazine insert: photorealistic workers and machinery, subtle depth of field, high clarity, balanced layout, soft lighting, crisp typography, and a professional Japanese manufacturing-industry infographic aesthetic.

A candid documentary-style indoor family snapshot in a modest early-2000s living room in China, photographed like a real consumer camera photo with soft ambient daylight, slight grain, natural color, and a yellow date stamp in the bottom right reading {argument name="date stamp" default="2003.02.02"}. The scene shows exactly 3 people: a little girl in the foreground sitting cross-legged on a patterned carpet and playing with a colorful toy car, one adult kneeling on the floor near the TV stand in a light gray sweater and blue pants, and one adult leaning over electronics equipment in a dark blue long-sleeve shirt near a sofa. Make the image feel unposed, nostalgic, and convincingly photographic, as if captured during an ordinary afternoon at home. On the left side, include 1 bulky CRT television displaying black-and-white static snow, sitting on a dark media stand with stacked electronics and shelves below. In the center background, place a wall calendar showing {argument name="calendar year" default="2003"}, and a wooden shelving unit holding framed family photos, books, tapes, and small household items. On the right side, include 1 sofa draped in a light floral lace cover with a blue folder or notebook resting on the arm. Add domestic clutter and lived-in details: exactly 2 potted plants on the floor by the TV, a patterned rug covering most of the floor, and a slightly dim, warm, everyday apartment atmosphere. The child should wear a pink sweater and dark pants, with small tied pigtails, focused downward on the toy. The overall style should be highly realistic, like a genuine archival family photograph rather than an illustration, with believable perspective, soft shadows, and subtle lens imperfections.

{argument name="邮票主题" default="中国航天纪念邮票"}小全张，{argument name="场景" default="火箭发射场景"}，烫金边框工艺

A cute hand-drawn scrapbook-style horizontal educational infographic in Chinese on a warm beige paper background with rounded borders, pastel doodles, tape stickers, tiny stars, and a classroom-notebook aesthetic. The top center has a large headline in playful handwritten Chinese: "什么是分词？" with the second half in coral pink and the rest in dark brown. On the upper left, a chibi Shiba Inu teacher character stands facing forward, wearing a pale yellow shirt and blue necktie, holding a pointer stick upward, with a speech bubble that says "今天来学习分词(Tokenization)!". On the upper right, a cute blue square token mascot labeled "TOKEN" sits smiling, with blush cheeks, tiny legs and arms, and a magnifying glass; next to it is a speech bubble saying "我是 Token！被切得越合理，模型越聪明！". Near the right edge, add a small hanging notepad card reading "知识点 收藏夹" with a heart icon. Beneath the headline, place a rounded dashed text box containing the definition: "分词是将人类可读的文本切分成 Token 的过程。" with the word "Token" highlighted in blue. Below that, add a line of explanatory text: "下面以 “程序员Left” 和 “人工智能” 为例，看看不同的分词方式：" with "程序员Left" highlighted in light blue and "人工智能" highlighted in pink. The lower half contains exactly 4 rounded pastel panels arranged left to right, each with a numbered badge and title. Panel 1 is mint green and titled "1 词级分词". It includes the subtitle "按完整语义划分文本" and exactly 3 token bubbles: "程序员", "Left", "人工智能". At the bottom of this panel, show the Shiba Inu teacher giving a thumbs up with a speech bubble saying "语义完整，容易理解！". Panel 2 is pale yellow and titled "2 字符级分词". It includes the subtitle "逐字切分文本" and exactly 9 circular token bubbles arranged in rows: "程", "序", "员", "L", "e", "f", "t", "人", "工", "智", "能"—visually presented as individual characters from the example text, in a compact grid. At the bottom, show the blue token mascot holding a pencil, with a speech bubble saying "粒度最细，但序列会很长~". Panel 3 is pale pink and titled "3 子词分词". It includes the subtitle "按词根以及前后缀划分" and exactly 5 token bubbles: "程序", "员", "Left", "人工", "智能". At the bottom, show the Shiba Inu teacher thinking with one paw at its mouth and a question mark above the head, with a speech bubble saying "折中方案，效果 often 很好！". Panel 4 is pale blue and titled "4 字节级分词". It includes the subtitle "将字词转为字节编码，根据转换后的内容进行组合和划分" and displays exactly 14 small circular code bubbles: "E7", "A8", "8B", "...", "66", "74", "E4", "BA", "BA", "E8", "83", "BD", "..." arranged in multiple rows like encoded bytes. At the bottom, show the blue token mascot again, smiling, with a speech bubble saying "适合处理任意文本！" and a tiny light bulb doodle nearby. Across the very bottom, place a rounded banner note that reads: "不同分词方式各有优势，实际应用中会根据任务和数据选择合适的方案哦！". Use soft pastel green, yellow, pink, and blue for the four panels, dark brown handwritten text, subtle shadows, rounded shapes, and adorable kawaii expressions throughout. Keep the composition balanced, clean, and friendly, like a polished social-media science explainer poster.

{"type":"hand-drawn educational infographic illustration","style":"cute notebook doodle style, hand-painted handbook aesthetic, warm beige paper background with faint texture, playful sketchy outlines, soft pastel colors, small flower and star doodles, taped paper-card layout","aspect_ratio":"wide horizontal","subject":{"characters":[{"name":"{argument name=\"dog character name\" default=\"柴小七\"}","species":"Shiba Inu mascot","appearance":"cute orange-brown Shiba Inu with cream muzzle and belly, round face, upright ears, friendly expression, simplified cartoon anatomy"},{"name":"{argument name=\"blue character name\" default=\"Token\"}","species":"blue token mascot","appearance":"rounded triangular blue droplet-shaped character with glossy gradient body, tiny arms and legs, rosy cheeks, smiling face, the word \"{argument name=\"token label\" default=\"TOKEN\"}\" written across the body in white"}],"theme":"compare four levels of text segmentation in a friendly science-explainer format"},"layout":{"sections":[{"title":"1.词级分词","position":"far left panel","count":1,"subtitle":"按完整语义划分","scene":"the Shiba Inu and blue token character stand beside a basket labeled \"Tokens\" filled with paper strips. Visible strips and tags read \"程序员\", \"Left\", \"人工智能\", and a small sign reads \"程序员Left\". The token character points toward the basket while the dog presents the categorized pieces."},{"title":"2.字符级分词","position":"left-center panel","count":1,"subtitle":"逐字切分文本","scene":"the blue token character holds a magnifying glass and examines scattered small character tiles on the floor while the Shiba Inu sorts them. A label strip at the bottom reads \"程序员Left人工智能\" as one continuous text sample being split into individual characters."},{"title":"3.子词分词","position":"right-center panel","count":1,"subtitle":"按词根及前后缀划分","scene":"the Shiba Inu gestures at a flowchart board showing broken word parts and arrows, while the blue token mascot stands nearby. Small paper labels include mixed chunks such as \"程序\", \"员\", \"Left\", \"人工\", and \"智能\", emphasizing morpheme or subword grouping."},{"title":"4.字节级分词","position":"far right panel","count":1,"subtitle":"转换为字节编码组合划分","scene":"the Shiba Inu uses a retro computer with a long printed strip emerging from it, and the blue token character stands to the side. Across the lower half are ribbon-like strips filled with repeated byte numbers and symbols, especially \"1\", \"2\", \"3\", \"4\", and \"-3\", illustrating byte encoding combinations."}],"decorations":{"count":10,"items":["4 pieces of taped corners in pale yellow and pink","small five-point stars scattered around panels","tiny flower doodles between panels","curved motion lines near character gestures","notebook-style vertical panel dividers","rounded rectangular panel frames","light shadows behind paper panels","faint horizontal paper grain","small spark icons","hand-drawn accent marks"]}},"composition":"four equal vertical panels arranged left to right across one canvas, each panel with a bold numbered Chinese heading and a smaller subtitle beneath it, each panel showing the two mascots acting out a different tokenization concept","text":{"headings":["1.词级分词","2.字符级分词","3.子词分词","4.字节级分词"],"subtitles":["按完整语义划分","逐字切分文本","按词根及前后缀划分","转换为字节编码组合划分"]},"quality":"clean, polished, adorable, presentation-ready educational visual","use_case":"social media science explainer comparing tokenization methods for AI and NLP"}

A cute horizontal educational infographic in a hand-drawn scrapbook style on a pastel desk-and-paper collage background, explaining Chinese word segmentation with byte-level BPE. The scene is divided into 3 clearly separated instructional sections arranged from left to right across a wide banner. On the far left stands a chibi Shiba Inu mascot, {argument name="character name" default="柴小七"}, with warm tan and cream fur, round face, small triangular ears, rosy cheeks, and a curious expression, holding a cup and standing beside a small desk with drawers, pencils, and a chair. Above the dog is a bold rounded white title box with black Chinese text: {argument name="headline text" default="中文分词：Byte-level BPE（BBPE）流程科普"}. In the first teaching section near the upper middle, show 4 small translucent blue token-shaped tiles lined up on a wooden shelf, each labeled “Token”, with a curved arrow and small handwritten Chinese note “词频语料统计” pointing toward the next step. In the second section, place a large magnifying glass highlighting 3 small frequency tiles labeled exactly “E7”, “94”, and “B5”, with the section label in a yellow note reading “2. 频率统计与合并”; beneath and inside the magnified area include a large black Chinese character “电”, and nearby a handwritten note “频繁字节对”. In the third section at lower middle-left, add a wooden sign and a larger merged translucent blue token tile labeled “Token”, with a small yellow note reading “3. 跨字合并”, large black Chinese text “我们→”, and a caption strip below that says “高频词组合并为 Token”. On the far right, show the final explanatory result with 3 small byte boxes labeled “E7”, “94”, and “B5” above a large black Chinese character “电”, then a note card reading “1. 字节级编码（UTF-8）”, and below that the large black Chinese word “我们”. Connect the sections with curved arrows in pink, blue, and green to show process flow. Include 1 animated blue token mascot with tiny arms and legs near the center-bottom, smiling and waving. Use soft cream, pink, beige, and light blue colors, thick clean outlines, sticker-like cutout shapes, taped paper corners, notebook textures, pencils around the edges, and a friendly hand-account illustration style suitable for a science explainer graphic.

A hyper-realistic cinematic action photo of a young man mid-jump above a rooftop at golden-hour sunset, captured from a dramatic low angle against a glowing orange sky filled with scattered clouds. He is centered in the frame, airborne with both knees bent, wearing a woven straw hat, an open {argument name="jacket color" default="bright red"} overshirt flaring in the wind, a white tank top, dark loose cargo-style pants, rugged dark boots, layered necklaces, fingerless gloves, and a belt with a launcher holster clipped at his waist. His right arm is thrust toward the camera, launching a spinning top toy on a visible ripcord, with the toy in the extreme foreground on the right side rendered large and motion-blurred for depth and speed. Place a transparent plastic Beyblade battle stadium at the bottom foreground, catching warm reflections from the sunset, with the clearly visible printed logo text {argument name="brand text" default="BEYBLADE X"}. The city skyline stretches across the lower background with several high-rise buildings silhouetted by the setting sun near the horizon. Lighting is intensely warm and backlit, with strong rim light, cinematic contrast, subtle lens flare, and dynamic fabric movement. The composition should feel like a live-action promotional poster for {argument name="toy franchise" default="Beyblade X"}, emphasizing energy, youth, speed, and urban rooftop competition, in vertical poster format.

A cinematic live-action action scene in a narrow old European stone street at golden-hour sunset, shot from an extremely low, close-up perspective with a dramatic wide-angle lens. In the center-left foreground, a dynamic male fighter inspired by {argument name="character name" default="Luffy"} lunges forward in mid-motion, wearing a straw hat, an open {argument name="jacket color" default="red"} shirt, a light tank top, and dark pants. His body is twisted aggressively as he launches a spinning-top battle attack with a handheld launcher. In the right foreground, a sharply dressed gangster-like man in a dark suit and white shirt leans in, also firing a launcher, with two more suited men behind him and a crowd of onlookers filling the street. In the immediate foreground, exactly 2 spinning tops collide on a glossy dark surface, producing showers of bright orange sparks and multiple circular light trails that arc around them, emphasizing speed and impact. The environment is dense and atmospheric: weathered stone buildings rise on both sides, the street is packed with spectators, and a grand baroque cathedral dome and towers stand in the sunlit background. Use intense warm amber and copper lighting, deep shadows, high contrast, dust and spark particles, gritty texture, dramatic motion blur, shallow depth of field, and an epic blockbuster tone. The composition should feel like a hybrid of underground battle, street duel, and anime-inspired live-action adaptation, with kinetic energy, tension, and hyper-detailed realism.