Prompts de Imagen para Diagram Chart

【入力データ】 ・左ページの料理名: {argument name="左ページの料理名" default="グリーンカレー"} ・左ページの材料リスト: {ingredientsLeft} ・左ページの調理手順: ①{stepLeft1} → ②{stepLeft2} → ③{stepLeft3} ・右ページの料理名: {argument name="右ページの料理名" default="トムヤムクン"} ・右ページの材料リスト: {ingredientsRight} ・右ページの調理手順: ①{stepRight1} → ②{stepRight2} → ③{stepRight3} ・全体のカラーテーマ: {argument name="カラーテーマ" default="オーガニック・ナチュラルカラー"} 【構造とレイアウトの指定】 ・スタイル: 薄いグリッド線（方眼）や温かみのある紙の質感が表現された、1枚の味のあるオフホワイトのメモ帳用紙（方眼紙）の上に描かれた、フラットな2Dの手書き風インフォグラフィックレシピ図解。 ・背景と構図の制約: 物理的な本、立体的な手帳の輪郭、ページの厚み、影、木のテーブルなどのリアルな写真背景要素は描かないでください（Absolutely NO wooden table, NO realistic book mockup, NO book thickness, NO notebook pages, NO realistic photo background. It must be a flat 2D vector layout directly on the textured grid notepad paper canvas）。 ・左側のレイアウト: - 上部に大きく「{dishLeft}」という手書きの日本語タイトル。 - その下に、入力された材料リスト「{ingredientsLeft}」を綺麗な縦並び（箇条書き）の手書き日本語とシンプルな食材イラストで描画。 - 下半分には、調理手順「①{stepLeft1} → ②{stepLeft2} → ③{stepLeft3}」を横方向または縦方向の矢印フローで配置。各ステップの内容（例：切る、炒める、煮込むなど）と完全に整合する可愛い手描きイラストを配置。 - 左下に完成した美味しそうな「{dishLeft}」の料理イラストを配置。 ・右側のレイアウト: - 上部に大きく「{dishRight}」という手書きの日本語タイトル。 - その下に、入力された材料リスト「{ingredientsRight}」を手書きの日本語とシンプルな食材イラストで描画。 - 下半分には、調理手順「①{stepRight1} → ②{stepRight2} → ③{stepRight3}」を横方向または縦方向の矢印フローで配置。各ステップの内容（例：切る、炒める、煮込むなど）と完全に整合する可愛い手描きイラストを配置。 - 右下に完成した美味しそうな「{dishRight}」の料理イラストを配置. ・色彩: 「{colorTheme}」を基調とした、優しく暖かみのあるオーガニックな色合い。 【品質および除外の制約（必須）】 ・画像の中に、日付（例: 「8/7」など）、曜日（例: 「SUN」など）、カレンダー枠、日付スタンプ等の日付関連要素は絶対に描かないでください（NO date stamp, NO calendar, NO day of the week, NO timeline stamp）。 ・フォントは手書き風の親しみやすい日本語フォント（ゴシックまたは丸文字）で、入力された日本語を崩れやスペルミスなく正確に描画してください。 ・アスペクト比: --ar 4:3

Goal: Create a polished Japanese typography comparison poster testing 30 different font styles for the same phrase, in a premium editorial infographic style for {argument name="model name" default="ChatGPT Images 2.5"}. Canvas: Landscape 3:2 poster, approximately 1200×800 px, ivory paper background with subtle texture, thin gray grid lines, high-resolution print layout. Header: A dark navy-to-slate gradient banner across the top. Large white serif title on the left: 「{argument name="headline text" default="ChatGPT Images 2.5　日本語フォント30種類 書き分けテスト"}」. Smaller white subtitle underneath: 「同じ文章でも、フォントが変わると、伝わる印象はここまで変わる。」 On the upper right, add a small refined slogan block: 「AIで、表現の幅が広がる。文字が変われば、世界が変わる。」 and below it in tiny spaced English capitals: “SAME WORDS, A THOUSAND IMPRESSIONS.” Main layout: Below the header, create an exact 5-row by 6-column grid containing exactly 30 numbered font sample cards. Each card has a small bold Japanese label at the top, one large sample phrase in the specified font style, and a tiny Japanese impression note below. The repeated sample phrase must be 「{argument name="sample phrase" default="生成AIで、仕事は変わる。"}」. Keep the phrase visually dominant in every cell while changing only typography, weight, color, texture, and decorative background according to the style. The 30 cards, in order from left to right, top to bottom: 1. 「明朝体」 elegant traditional Mincho serif, black, formal and trustworthy. 2. 「ゴシック体」 modern bold Gothic sans, black, simple and highly legible. 3. 「丸ゴシック体」 rounded Gothic, soft friendly black text. 4. 「極太ゴシック体」 very heavy extra-bold black Gothic, strong impact. 5. 「細身ゴシック体」 thin stylish Gothic, light gray-black, refined. 6. 「新聞明朝体」 newspaper-style Mincho, readable and intellectual. 7. 「教科書体」 textbook style, clean educational Japanese letterforms. 8. 「筆文字」 bold expressive brush calligraphy, energetic ink strokes. 9. 「毛筆楷書」 formal brush kaisho, traditional and dignified. 10. 「毛筆行書」 flowing semi-cursive brush script, elegant. 11. 「手書き風（ペン書き）」 casual pen handwriting, warm and natural. 12. 「サインペン風」 marker-pen handwriting, casual and cheerful. 13. 「子どもの手書き風」 childlike handwritten style, cute uneven strokes. 14. 「昭和レトロ」 warm retro Showa-era display type, deep red-brown, nostalgic. 15. 「大正ロマン」 Taisho romantic serif display, framed with delicate ornamental corner borders and beige paper. 16. 「江戸看板風」 Edo-period signboard style, thick black lively characters. 17. 「和風高級旅館」 refined Japanese luxury ryokan style, calm serif over pale patterned background. 18. 「女性誌風」 women’s magazine style, elegant thin serif over soft pink floral background. 19. 「ビジネス誌風」 business magazine style, serious black serif, trustworthy. 20. 「テック系」 tech style, blue-black sans over pale circuit-line background. 21. 「SF風」 science-fiction style, glowing cyan-white text on dark starry space background. 22. 「ミニマル」 minimalist thin sans/serif, lots of whitespace, quiet beauty. 23. 「高級ブランド風」 luxury brand style, white serif on black background with subtle gold accents. 24. 「ポップ」 playful pop style, bright blue and pink rounded letters on yellow with colorful dots. 25. 「かわいい」 cute pastel pink rounded letters with small hearts. 26. 「漫画タイトル風」 manga title style, bold black dynamic letters with speed lines and orange burst background. 27. 「ホラー風」 horror style, distressed pale text on dark red-black eerie background. 28. 「スポーツ新聞風」 sports newspaper style, bold red and blue italic headline with dynamic diagonal shapes. 29. 「選挙ポスター風」 election poster style, bold white text on blue and red campaign-poster blocks. 30. 「公共サイン風」 public sign style, yellow signboard with black arrows and heavy black text, practical and highly readable. Footer: A clean footer strip along the bottom. Left: “AICLASS” in bold dark type, followed by tiny Japanese tagline 「AIと言葉の力で、可能性を広げる。」 Center: small Japanese sentence 「フォントは、ただの文字ではない。それは、伝えたい想いのかたち。」 Right: handwritten signature-style text “Genie G2”. Visual style: Crisp Japanese typography specimen sheet, professional magazine design, accurate grid alignment, varied font rendering, subtle paper grain, balanced margins, no random extra panels. Use exactly 30 cards and keep every card numbered.

Using the provided reference image as the geometric base, turn the rough hand-drawn nested cube/tunnel form into a polished three-column style exploration poster. Keep the same core structure: an outer cube frame with inward sloping faces leading to a smaller centered cube opening. Create exactly 3 vertical panels separated by white gutters, each showing the same object interpreted in a different visual style, with a caption block underneath each image. Canvas and layout: Wide horizontal poster, 3 equal columns. Each column has a large square image area on top and a clean white text area below. Use elegant editorial spacing and centered typography. Panel 1: Render the reference form as a realistic sculptural object made of rough stone or concrete, sitting on a dark reflective surface. Add warm internal lighting along the inner edges and openings so the structure feels physically deep and architectural. Panel 2: Render the same form as a modern minimalist geometric illustration: clean black outlines, transparent glass-like cube faces, subtle blue and gray planes, white studio background, crisp shadows, and a restrained architectural drawing feel. Panel 3: Render the same form as a dreamy fantasy art object: a glowing transparent crystalline cube floating in the sky, reflecting sunset clouds, mountains, stars, and cosmic light, with luminous pink-orange-blue highlights. Text: Add exactly 3 numbered Chinese caption blocks beneath the panels: 1. Title: {argument name="first panel title" default="1. 写实风"}; subtitle: “光影与材质，呈现空间的真实感”; small line: “— 看见结构中的深度 —” 2. Title: {argument name="second panel title" default="2. 现代简约风"}; subtitle: “简洁的线条与色彩，突出几何美感”; small line: “— 简单的形状，丰富的可能 —” 3. Title: {argument name="third panel title" default="3. 梦幻艺术风"}; subtitle: “用光影与想象，赋予空间诗意”; small line: “— 在有限的空间，看到无限的世界 —” Style constraints: Preserve the reference object’s recognizable nested-cube geometry in all 3 panels. Make the poster clean, premium, and presentation-ready. Do not add extra panels, extra objects, watermarks, logos, or unrelated text.

Goal: Create a hyper-realistic futuristic medical engineering visualization of an exploded transparent artificial human heart, presented like a premium bioengineering product render with technical callouts. Canvas: 4:3 horizontal image, dark laboratory background, glossy black reflective floor, cinematic rim lighting, shallow depth of field, cool blue highlights and warm amber internal glow. The heart floats centered above the floor with a soft reflection beneath it. Main subject: A transparent biomechanical anatomical heart labeled as {argument name="device concept" default="next-generation bioengineered artificial heart"}. The central heart body is made of clear glass-like biocompatible polymer, with visible internal tubes, valves, microfluidic channels, circuit traces, miniature mechanical pumps, red and blue glowing vascular pathways, and amber illuminated electronic modules. Use realistic refraction, caustics, chrome screws, polished metal rings, and black carbon-fiber structural components. Exploded layout: Show exactly 13 major separated component groups around the central heart: 1 central transparent heart chamber; 2 left outer clear curved shell; 3 left black carbon-fiber rib panel; 4 left inner transparent curved support layer; 5 upper-left aorta connector ring assembly; 6 upper central transparent aorta and vessel towers; 7 upper-right pulmonary artery connector ring assembly; 8 right-side cylindrical port and interface rings; 9 right black carbon-fiber rib panel; 10 right transparent circuit-interface module with amber glow; 11 right outer clear curved shell; 12 far-right small transparent cap/lens; 13 multiple small floating screws and bolts aligned around the exploded parts. Keep all parts suspended in precise horizontal layers, as if disassembled for inspection. Technical annotations: Add thin white/gray leader lines and small uppercase sci-fi typography. Include exactly 7 visible callout labels: 1 “AORTA” with subtext “HIGH-FLOW / BIOCOMPATIBLE / POLYMER”; 2 “RIGHT ATRIUM” with subtext “MICRO-FLUIDIC / CHANNELS”; 3 “CARBON-FIBER RIB STRUCTURE” with subtext “ULTRA-LIGHT / HIGH-STRENGTH”; 4 “VENTRICLE” with subtext “BIO-MIMETIC / PUMP CHAMBER”; 5 “PULMONARY ARTERY” with subtext “PRESSURE REGULATING / VALVE”; 6 “BIO-CIRCUIT INTERFACE” with subtext “NEURAL SYNC / WIRELESS POWER / REAL-TIME MONITORING”; 7 “MICRO-FLUID CHANNELS” with subtext “NANO-SCALE / SELF-CLEANING / FLOW OPTIMIZATION”. Place these labels around the heart without covering key details. Footer text: In the bottom left, small spaced lettering reads {argument name="left footer text" default="HUMANITY\nA STRONGER TOMORROW"}. In the bottom right, add a tiny schematic line icon and text reading {argument name="right footer text" default="BIOENGINEERING\nFOR A HEALTHIER WORLD"}. Visual style: Ultra-detailed photorealistic 3D render, luxury industrial design, transparent medical device, carbon fiber texture, chrome hardware, glass refraction, glowing red and blue tubes, amber circuit light, high contrast, clean futuristic interface typography, technical diagram aesthetic, 8k quality. Constraints: No people, no hands, no blood or gore, no cartoon styling, no messy background, no extra large titles. Keep the anatomy recognizable as a human heart while emphasizing transparent futuristic engineering.

Create a clean anime production setting illustration of a {argument name="fish species" default="Japanese sea bass"} in full left-facing side profile, centered horizontally on a deep navy blue background. The fish should be long and realistic but stylized like a hand-painted animation reference sheet, with a dark bluish-gray back, silver-white belly, visible overlapping scales, a pale lateral line, large golden eye, defined gill plates, and crisp ink outlines. Show exactly 7 visible fins: one spiny dorsal fin near the front of the back, one soft dorsal fin behind it, one large forked tail fin, one pectoral fin on the visible side, one pelvic fin underneath near the head, one anal fin underneath near the rear, and a small lower fin edge partially visible behind the body. Add a thin fishing line entering from the left into the fish’s mouth and a small metallic lure or hook rig positioned inside the open mouth so that the hook appears caught in the mouth, not floating outside; keep the lure subtle and partially hidden by the lips. Use detailed fin rays, delicate scale texture, muted natural colors, soft cel-shading, and a polished concept-art look. No labels, no extra objects, no watermark, no border.

Goal: Create an elegant Chinese literary presentation cover slide about {argument name="topic" default="李清照对当代女性的启示"}, blending Song dynasty poetic atmosphere with modern editorial design. Canvas: 16:9 horizontal slide, clean off-white background with a pale sage-green vertical panel occupying the right third. Use lots of negative space, refined margins, soft paper texture, faint ink-wash stains, and subtle gray botanical shadows across the middle. Layout: Left two-thirds is the text area; right third is a still-life photograph area. Place the large headline in the upper-left, arranged as two lines of dark forest-green Chinese serif type: line 1 “李清照对当代”, line 2 “女性的启示”. Under the headline add a thin muted red horizontal accent line. Below it place the subtitle in smaller black Chinese serif text: {argument name="subtitle text" default="从文学人生，看见女性如何表达自我、选择人生、抵抗困境。"} Middle-left feature row: include exactly 3 small icon cards in a horizontal row, separated by two thin vertical divider lines. Card 1 has a muted rose circular icon with a white leaf symbol and the label “以文学\n确立自我”. Card 2 has a warm beige circular icon with a white balance-scale symbol and the label “在时代与命运之间\n保持判断”. Card 3 has a desaturated teal circular icon with a white user silhouette and the label “不被单一\n身份定义”. Bottom-left keyword row: include exactly 3 keywords separated by vertical bars, with a small dark green leading mark: “表达自我 | 保持独立 | 承受变化”. Right still-life details: show an open antique Chinese letter or manuscript paper lying diagonally on a desk, with visible vertical handwritten Chinese calligraphy in gray ink. A black-and-brown calligraphy brush rests diagonally over the paper, pointing toward the lower-right. Add exactly 2 blooming plum blossom branches: one branch crosses the upper-right corner with several pale white-pink blossoms and buds, and one smaller branch extends from the center-left toward the right, casting soft shadows. The right panel background should be pale sage green; the paper is warm ivory with slightly curled edges and soft folds. Visual style: tasteful, literary, museum-catalog minimalism; calm Song dynasty aesthetics; natural daylight from upper left; soft shadows; muted palette of ivory, forest green, sage, rose, beige, and ink gray; high-resolution editorial composition, realistic still-life photo blended with clean graphic typography. Constraints: Keep all visible text in Chinese exactly as specified, with no extra captions or watermark. Maintain the 3 icon cards and 3 bottom keywords exactly. Avoid clutter, bright colors, cartoon styling, or modern tech UI effects.

【入力データ】 ・全体タイトル: {argument name="メインタイトル" default="夜勤導入3ステップのフロー図"} ・ステップ1: タイトル「{argument name="ステップ1タイトル" default="日勤習熟"}」/ 期間「{step1Period}」/ 説明「{step1Desc}」 ・ステップ2: タイトル「{step2Title}」/ 期間「{step2Period}」/ 説明「{step2Desc}」 ・ステップ3: タイトル「{step3Title}」/ 期間「{step3Period}」/ 説明「{step3Desc}」 ・カラーテーマ: {argument name="カラーテーマ" default="グリーン"} 【構造とレイアウトの指定】 ・スタイル: 日本のビジネススライドや教育用マニュアル、社内資料でそのまま使える、クリーンで親しみやすい3ステップのフローチャート図解。16:9の横長スライド構成。 ・背景: すっきりとしたオフホワイトまたは白の無地背景。 ・上部タイトル: - 左上に「図3」またはステップバッジを配置し、その右隣に太く見やすいゴシック体で「{mainTitle}」という日本語のメインタイトルを大きく描画してください。右端にはクリーンな抽象ロゴマークを小さく配置してください。 ・ステップカードの配置: - 横並びになった3つの美しい長方形カード。左から右へ順に配置され、ステップ間は太い接続矢印で結ばれています。各カードの枠線はカラーテーマ（{colorTheme}）に沿って綺麗に色分けされています。 ・カード1（左）の内部: - 丸バッジ「1」と、日本語タイトル「{step1Title}」を大きく描画。その下に「{step1Period}」と書かれたカプセル型ラベル。 - 中央には、メモをとる優しい手書きタッチの日本人女性と、太陽が描かれた窓のイラストを配置。 - 下部には、日本語の説明文「{step1Desc}」を配置。 ・カード2（中央）の内部: - 丸バッジ「2」と、日本語タイトル「{step2Title}」を大きく描画。その下に「{step2Period}」と書かれたカプセル型ラベル。 - 中央には、懐中電灯を持つ二人の日本人女性と、夜空とビルのイラストを配置。 - 下部には、日本語の説明文「{step2Desc}」を配置。 ・カード3（右）の内部: - 丸バッジ「3」と、日本語タイトル「{step3Title}」を大きく描画。その下に「{step3Period}」と書かれたカプセル型ラベル。 - 中央には、夜空と三日月のイラストと、ガッツポーズをして微笑む日本人女性のイラストを配置。 - 下部には、日本語の説明文「{step3Desc}」を配置。 【品質・フォントの指定】 ・シンプルでクリーンな2Dイラスト、フラットデザイン、すっきりとした明確な線画。 ・文字はすべてスペルミスなく、読みやすい正確な日本語フォント（ゴシック体）で美しく描画してください。 ・一切の3D効果や影、ノートパソコン等のモックアップ背景、カメラの傾きや歪みは排除し、フラットな図解画像そのものを画面全体いっぱいに表示してください。 ·アスペクト比: --ar 16:9

Goal: Create a futuristic Chinese tech infographic cover slide explaining {argument name="headline text" default="GPT Image 2.5 Sunburst 模型能力解说"}, based on OpenAI image prompting documentation, in a glowing cyber-blue presentation style. Canvas: 16:9 horizontal slide, deep navy-to-cyan gradient background with soft radial light beams, neon arcs, thin orbital lines, lens glow, glassmorphism panels, and subtle high-tech UI grid details. Use crisp vector-like typography and luminous cyan borders. Add small decorative English microcopy in the corners. Main header: At the top center, large bold white title: “GPT Image 2.5 Sunburst 模型能力解说”. Directly below it, smaller bold subtitle: “基于 OpenAI《Image Prompting》文档的图示化学习版”. Under the subtitle, place a rounded URL pill with a chain-link icon and the text “developers.openai.com/api/docs/guides/image-prompting”. Central diagram: In the center, create a circular sunburst capability map. Put a bright glowing white-cyan core circle labeled “GPT Image 2.5 Sunburst” with a smaller Chinese line underneath: “一张图，开启更多可能”. Around it, place exactly 6 numbered colored capability nodes connected by orbit rings and radial lines: 1. Cyan node with cube icon: “1. 模型选择”, smaller text “选对模型 / 发挥最佳效果”. 2. Green-teal node with sliders icon: “2. 参数设置”, smaller text “掌握关键参数 / 控制生成效果”. 3. Purple node with document-and-pencil icon: “3. Prompt 编写原则”, smaller text “用好的提示词 / 表达你的意图”. 4. Pink-magenta node with image icon: “4. 生成能力”, smaller text “从想法到画面 / 的多样化生成”. 5. Orange node with magic-wand icon: “5. 编辑能力”, smaller text “基于现有图像 / 进行精准编辑”. 6. Violet-blue node with refresh arrows icon: “6. 跨轮迭代”, smaller text “多轮优化 / 逐步达成理想结果”. Left side feature cards: Arrange exactly 3 rounded translucent cards stacked vertically, each with a square icon tile on the left and Chinese text on the right: 1. Lightbulb icon, title “把文档转成易懂图示”, body “将官方文档的关键要点 / 转化为直观、清晰的视觉内容”. 2. Target icon, title “聚焦可实操方法”, body “提供真实可用的技巧与示例 / 帮助你快速上手”. 3. Image icon, title “覆盖生成与编辑”, body “从文本生成到图像编辑 / 全面梳理核心能力”. Right side feature cards: Arrange exactly 3 matching rounded translucent cards stacked vertically: 1. Shield-check icon, title “强调可读性与准确性”, body “在简洁的表达中 / 保持内容的专业性与准确性”. 2. Group/users icon, title “适合传播与内部培训”, body “便于分享、学习与团队使用 / 提升 AI 图像创作能力”. 3. Palette icon, title “统一系列风格”, body “10 页完整体系 / 一致的设计语言与视觉风格”. Bottom content-structure band: Add a wide rounded glass panel across the bottom. On the left of this panel, show a large open-book icon and the bold heading “本套内容结构”, with smaller body text “10 页带你系统掌握 / GPT Image 2.5 的图像提示词方法”. To the right, show exactly 10 chapter rows in two columns of five, each row in a small rounded blue pill with a two-digit number: 01 “封面导读” 02 “模型选择” 03 “参数与尺寸规则” 04 “Prompt 编写原则” 05 “控制风格与光线” 06 “过程/知识可视化” 07 “文本广告与 Logo” 08 “叙事/历史/UI/商业页” 09 “参考图驱动编辑” 10 “透明背景与跨轮迭代”. Decorative side text: Upper left tiny vertical microcopy reads “MORE / CREATIVE / POSSIBILITIES / WITH AI”; below it add Chinese copy “用更好的提示词 / 创造更美好的视觉世界” with a small underline. Upper right tiny Chinese copy reads “更清晰的思路 / 更强大的表达 / 让想象力 / 变成可实现的图像”. Add handwritten script near the upper-right diagram: “From Docs to Mastery”. Add handwritten script near bottom right: “Good Prompts / Better Images / A More Creative You”. Footer: Bottom left add tiny separated words “IDEAS / PROMPTS / IMAGES / A BRIGHTER TOMORROW”. Bottom right add page marker “Page 1 / 10”. Visual style: Premium SaaS keynote cover, luminous blue glass cards, neon cyan outlines, high contrast white Chinese typography, rounded UI panels, subtle blur, volumetric light, polished infographic composition. Keep all major elements symmetrical and readable, with the central sunburst as the focal point. Constraints: Use exactly 6 central capability nodes, exactly 6 side feature cards, and exactly 10 bottom chapter rows. Do not add extra chapters or cards. Preserve the Chinese headings and labels exactly as specified. No logos other than text references to OpenAI and GPT Image 2.5. No watermark.

Goal: Create a wide black-background comparison graphic showing two generated YouTube home page screenshots for a tech-bro recommendation feed, comparing {argument name="left model label" default="GPT-Image 2.5"} versus {argument name="right model label" default="GPT Image 2"}. Canvas: Very wide 3:1 landscape image, approximately 1200×640 pixels. Use a pure black outer background with generous margins. Center the composition horizontally. Layout: Place exactly 2 large YouTube dark-mode browser screenshots side by side. The left screenshot sits in the left half and the right screenshot sits in the right half, with a clear gap between them. Above each screenshot, center a white model label in a clean sans-serif font. Under both screenshots, centered near the bottom, place one caption: {argument name="bottom caption" default="A screenshot of Youtube home page for a tech bro."} Left screenshot details: A dark-mode YouTube homepage mockup labeled GPT-Image 2.5. Include the YouTube logo at top left, a search bar along the top, sidebar navigation on the far left, and category chips across the top. Show exactly 12 visible video thumbnail cards arranged in a 4-column by 3-row grid. The thumbnails should look like startup, productivity, investing, AI, and self-improvement content. Include these 12 visible thumbnail themes: 1) “The next 10 years” with a young male tech founder portrait, 2) “How to Build a Startup in 2025,” 3) “A Calmer More Focused You” landscape/meditation thumbnail, 4) “ELON MUSK Full Interview,” 5) “M4 Worth it?” laptop thumbnail, 6) “Build a Second Brain” Notion-style thumbnail, 7) “S&P 500” green stock chart, 8) “Stop Scrolling. Build Something.” creator portrait, 9) “The Future of Work” scenic thumbnail, 10) “a day in my life” desk/window lifestyle thumbnail, 11) “Tech Products That Changed My Life” gadgets on table, 12) one additional tech/productivity card consistent with the grid. Use small channel avatars, view counts, durations, and metadata text, but keep most small text only partially legible like a real screenshot. Right screenshot details: A dark-mode YouTube homepage mockup labeled GPT Image 2. Include the YouTube logo, search bar, sidebar navigation, top category chips, and profile/create icons. Show exactly 8 visible video thumbnail cards arranged in a 4-column by 2-row grid. The thumbnails should be sharper and more spacious than the left panel. Include these 8 visible thumbnail themes: 1) “how I built this.” with a Shopify bag icon and founder portrait, 2) “The Era of AI Employees” with two men and AI icons, 3) “$0 to $1M ARR” black chart thumbnail, 4) “The Truth About Silicon Valley” with a man’s portrait, 5) “Hard Truths” with Elon Musk portrait, 6) “minimal. but effective.” desk setup thumbnail, 7) “$10K/MO SIDE PROJECT” with creator portrait and app icons, 8) “How Bitcoin Works” with Bitcoin icon network graphic. Add small video durations and metadata beneath each card. Visual style: Photorealistic screenshot-comparison aesthetic, crisp UI details, dark YouTube theme with charcoal panels, white and gray text, red YouTube accents, subtle thumbnail colors, realistic browser screenshot proportions. The overall graphic should look like a social media post or review image comparing two AI image models. Constraints: Do not add an AI Search video. Keep exactly 2 screenshots, exactly 2 top labels, exactly 1 bottom caption, exactly 12 video cards in the left screenshot, and exactly 8 video cards in the right screenshot. Avoid watermarks and avoid any extra explanatory text outside the specified labels and caption.

Using REFERENCE_0 as the architectural source, reorganize it into a cleaner, more focused Japanese technical drawing sheet for the corridor window assembly. Keep the same subject and hand-drafted architectural illustration style, but simplify the page and redraw the construction details as an inferred measured detail rather than a broad explanatory board. Goal: Create a detailed drawing titled {argument name="main title" default="廊下 窓まわり詳細図"} with subtitle {argument name="subtitle" default="(断面図・展開図・部分詳細図)"}. Treat all dimensions as estimates from the reference/photo and note that the unit is {argument name="unit note" default="mm"}. Layout: Use exactly 4 drawing areas: 1 large A-A section detail on the left, 1 interior elevation view at the upper right, and 2 small partial detail diagrams at the lower right. Make the left section the dominant focus and include a small key plan/section marker at the top of it. Left main section: Redraw the corridor window section at S=1/10 with clearer construction layers and orange dimension lines. Show exterior planting and exterior ground on the left, the tall wood sash/glass opening in the center, and the interior corridor side on the right. Include labels for exterior side, interior side, ceiling, steel column, wood sash, double glass, brick sill/window base, brick exterior wall, steel angle/bracket, floor build-up, existing RC slab/foundation, GL ±0 and FL ±0. Emphasize estimated dimensions including opening height about 2100, window base about 300, brick rise 180, brick ledge width 350, exterior passage about 600, wall 100, sill/bench 350, and corridor width about 1200. Upper-right drawing: Create an interior elevation view labeled {argument name="elevation label" default="内観立面図（廊下側） S=1/50"}. Show the corridor-side window wall with repeated vertical wooden frames, glass panes, brick/bench base, steel columns, exterior planting visible through the glass, and orange A-A section markers. Lower-right details: Add exactly 2 partial detail diagrams labeled a and b: a. upper/head detail showing ceiling, PB substrate/paint finish, wood sash, and double glass; b. lower/sill detail showing brick finish, mortar substrate, brick support, and steel angle plus bracket, with orange dimensions 350, about 300, and 180. Style and constraints: Maintain the reference’s clean Japanese architectural presentation: thin black linework, light gray hatching, muted wood and brick tones, orange dimensions, white background, precise leader lines, and a professional measured-drawing look. Use Japanese technical labels where shown, but avoid overcrowding. Add a bottom caution note {argument name="caution note" default="※寸法は写真からの推定です。実施設計時は現地調査のうえ決定してください。"}. Do not add unrelated photos, people in the interior, decorative icons, or extra drawing panels.

Goal: Create a warm vertical Xiaohongshu-style illustrated poster explaining that letting AI think longer can reduce the bill, using a cozy home DIY scene and clear handwritten Chinese typography. Canvas: Vertical 3:4 social media poster, soft golden afternoon light, high-resolution anime/editorial illustration, inviting and optimistic. Main headline: At the very top, place a large cream brush-stroke banner with bold hand-painted Chinese text: {argument name="headline text" default="让 AI 多想一会儿，账单竟然少了近一半"}. Emphasize the final phrase in orange-red while the rest is dark brown-black. Add small yellow-orange emphasis rays on both sides. Subtitle: Directly below the headline, add a smaller rounded white label with Chinese text: {argument name="subtitle text" default="Astra 的能力曲线为什么开始拐弯"}. Central scene: A cute young man with {argument name="hair color" default="messy dark brown hair"}, wearing a forest-green sweater and dark pants, sits cross-legged on a rug in a sunlit room. He looks upward thoughtfully with a small smile, holding a pencil near his chin and an open instruction booklet in his lap. The booklet shows a simple line drawing of a shelf on the left page and a checklist on the right page. Counted text elements and labels: Include exactly 1 thought bubble above the character with Chinese text: {argument name="thought bubble text" default="先想清楚 再动手..."}. Include exactly 4 checklist rows in the booklet: 3 checked items labeled 「思考」, 「规划」, 「执行」, and 1 unchecked grey item labeled 「返工」. Include exactly 1 whiteboard chart in the background labeled 「效果 vs 成本」 with a vertical axis 「效果 (%)」 and horizontal axis 「成本 ($)」. On the chart, show exactly 2 curves: an orange solid curve with round markers bending upward then flattening, labeled 「Astra 开始拐弯」 with an orange arrow, and a grey dashed curve labeled 「不够思考 成本更高」. Room and objects: The setting is a cozy living room or study with blurred plants, framed pictures, warm window light, and a wooden shelving unit being assembled. Show exactly 1 sleeping orange-and-white cat curled on top of the shelf. Show scattered DIY parts in the foreground: wooden boards, screws, and 1 orange-black screwdriver. Add exactly 1 white mug near the character with English text stacked as “Good Plan Lower Bill” and a small smiley face. Sticky notes and cardboard signs: Include exactly 3 yellow note/sign elements: 1 sticky note on the upper right wall saying 「多一步思考 少很多返工 更聪明 也更省钱」 with a smiley face; 1 sticky note on the wooden shelf saying 「一次做好 省时省钱」 with a bold check mark and small yellow emphasis rays; 1 cardboard box sign in the lower right foreground saying 「更聪明的 AI 更轻松的生活」 with a simple house icon. Visual style: Soft painterly anime illustration mixed with infographic design, rounded shapes, thick friendly hand-lettering, warm beige/orange/green palette, gentle depth of field, clean composition, lively but not cluttered. Constraints: Preserve all Chinese text exactly as specified, keep the poster readable, use no logos or watermarks, and do not add extra charts, extra sticky notes, extra characters, or extra checklist items.

Create a wide cropped comparison graphic showing two side-by-side image-generation quality results on a plain white background. The canvas is an ultra-wide horizontal strip, about 5:1 aspect ratio, cropped tightly so only the top portion of each result is visible. At the top left above the first crop place the bold black uppercase label “MEDIUM”; near the upper center above the second crop place the bold black uppercase label “HIGH”. Each of the 2 comparison panels shows the same fitness-class style tiled layout: a photographic band at the top with gym equipment and dark exercise machines, underneath a deep teal rectangular title bar with large bold white Japanese text {argument name="main tile text" default="有酸素"}. Use thick white borders and sharp diagonal white dividers between tiles, giving the layout a dynamic slanted collage look. Count exactly 2 main visible teal title tiles with the full text “有酸素”, one in each panel, plus exactly 2 partially visible neighboring teal tiles at the right edge showing only the first character or fragment {argument name="partial tile text" default="ス"}. A thin lower white divider and the top edge of another teal row are barely visible along the bottom, cropped off. Emphasize the artifact being compared: around the white Japanese letters on the solid teal backgrounds, include subtle uneven brightness and saturation patches, faint haloing, and mottled texture, with the “HIGH” panel slightly cleaner but still showing visible nonuniformity. Keep the composition as a technical screenshot/comparison crop, not a polished poster. Use crisp sans-serif typography, high contrast, no extra captions, no watermark.

Goal: Create a square technical comparison graphic showing how focal length changes perspective in a simple 3D room scene, comparing Blender renders against GPT Images 2.5 outputs. Canvas: Square image, approximately 1080 x 1080 px, with thin black outer border and two horizontal black title bars. The graphic contains exactly 8 rendered panels arranged as two stacked 2 x 2 grids. Layout: The upper half is labeled {argument name="top section label" default="Blender"} in bold yellow text centered on a black bar at the very top. The lower half is labeled {argument name="bottom section label" default="GPT Images 2.5"} in bold yellow text centered on a black divider bar between the two grids. Each section contains exactly 4 panels: top-left focal length 16, top-right focal length 24, bottom-left focal length 36, bottom-right focal length 50. In every panel, place the focal-length label in bold white text at the lower-right corner: “Focal length 16”, “Focal length 24”, “Focal length 36”, and “Focal length 50”. Scene repeated in all panels: A clean minimalist 3D test room with white walls, white ceiling, glossy pale gray tiled floor with visible grid lines, and a rear wall with a centered dark gray vertical door or rectangular panel. Place exactly 3 colored cubes in the scene: 1 large glossy {argument name="center cube color" default="red"} cube in the foreground center, 1 small glossy {argument name="left cube color" default="blue"} cube to the mid-left, and 1 small glossy {argument name="right cube color" default="yellow"} cube to the mid-right. The cubes cast subtle soft reflections on the polished floor. Use symmetrical perspective lines converging toward the back wall. Panel-specific camera behavior: In the Blender section, make the focal length progression visually accurate: 16 mm appears very wide-angle with the red cube smaller and farther away, strong room expansion and visible side walls; 24 mm slightly less wide with the red cube larger; 36 mm more zoomed with the red cube prominent; 50 mm most zoomed with the red cube very large and the side cubes closer to the edges. In the GPT Images 2.5 section, show the same scene and same four labels, but make the 16 mm and 24 mm panels less wide than expected, with object scale changing only modestly across focal lengths; the 16 mm panel should not shrink the objects as much as the Blender 16 mm panel. Visual style: Clean synthetic 3D render, soft global illumination, high-key white room, subtle shadows and reflections, crisp tiled floor, no people, no extra objects, no watermark. Maintain exact count: 2 labeled sections, 8 panels, 4 focal-length labels per section, and 3 cubes per panel.

将此说明图表转换为图标。完全删除其中的文字和背景。*background="{argument name="background" default="transparent"}"

Research the poster regulations for presentation at the {argument name="conference name" default="2026 Neuroscience Conference"} and transform the attached paper into a poster accordingly.

Using REFERENCE_0 as the sole source, faithfully recreate the Japanese Kasumigaseki-style policy infographic as a clean, high-resolution 16:9 slide. Preserve the original overall layout, dense information structure, blue title bar, yellow/red emphasis bands, arrows, callout boxes, illustrated landscape/business icons, and all visible Japanese labels as closely as possible. Improve the output by making it sharper, brighter, less compressed, and more legible, with cleaner vector-like text boxes and clearer small illustrations, while keeping the same crowded official-government presentation aesthetic. Keep the top title text as {argument name="headline text" default="地域循環共生圏（日本発の脱炭素化・SDGs構想）"}, keep the version label as {argument name="version label" default="Ver. 26"}, and retain the main central label {argument name="central concept label" default="地域循環共生圏"}. Do not simplify the diagram, do not modernize it, do not remove small labels, and do not add new branding, watermarks, people, or unrelated decorative elements.

请将我上传的每一张照片分别制作成一张独立的“旅行观察档案”设计海报，不进行多图拼贴，每张照片单独输出。 整体采用2:3竖版构图，由上下两个相互呼应的区域组成。上方实景摄影约占画面43%，下方旅行观察档案约占57%。分界清晰、平直，不使用撕纸、渐变融合或装饰性边框。 ## 上方：实景摄影 忠实保留原始照片，准确维持主体身份、数量、外观、结构、姿态、方向、地形、建筑、道路、水域、空间关系、真实材质、自然光影与原有色彩氛围。 只进行克制的展览级摄影调色，适度改善曝光、层次、清晰度和色彩统一性，使其具有旅行杂志、独立出版物与摄影展览的质感。 为适配竖版画幅，可以自然调整裁切，或合理扩展天空、地面、水面及周边环境，但不得拉伸、扭曲、移动、替换或重新设计主体。 如果原图包含雾、云、雪或弱光，应保留真实天气特征，但不要让浓雾、乳白色空气感或低对比度遮蔽主体。主体边缘、关键材质和空间层次必须清楚可辨。 ## 下方：旅行观察档案 在温暖的米白色天然档案纸上，重新绘制照片中的同一场景。 首先识别照片中最具有旅行叙事性的视觉锚点，例如： 船、灯塔、孤屋、房车、桥梁、车站、亭阁、门楼、塔楼、道路、码头或其他能够代表地点与旅程的主体。 将这一主体作为下方观察绘图的核心，并保持其外形、比例、方向、结构特征及其与环境之间的关系，使人能够一眼认出原照片。 使用以下绘画语言： - 铅笔结构素描 - 干墨线条 - 细密但克制的排线 - 少量从照片中提取的低饱和设色 - 局部纸张透底 - 轻微旧档案颗粒 - 精细引线与观察标注 - 尺寸参考、方向线和环境记录 - 松散但准确的旅行手稿质感 不要把下方做成精确的建筑CAD、工程制图或密集科普信息图。它应该像旅行者、建筑观察者或自然研究者留下的一页私人观察档案，具有知识感，但仍然安静、诗意并保留大量留白。 下方主体绘图应占据主要视觉区域。周围可以出现少量细线标注，指出照片中真实存在的重要结构、自然特征和空间关系。 底部设置5个尺寸统一的小型观察图板，根据照片内容自动选择： - 地理位置示意 - 地形或天际线轮廓 - 主体正视图或侧视图 - 局部结构剖面 - 天气、水面、云层、植被或道路观察 所有图板均应来自原照片内容，不得凭空添加无关结构。 ## 中文排版 根据照片中的地点、主体和旅行意象，自动生成： - 一个2至6字的中文主标题 - 编号，例如“编号 01” - 四位数年份 - 一句简短、克制且具有画面感的中文说明 - 3至4项场景信息 - 4至6个结构观察标注 - 5个底部图板标题…

Use the uploaded photo as the primary reference. Create a sophisticated 4:3 landscape {argument name="style" default="architectural specimen card inspired by museum archives, architectural surveys, and vintage scientific documentation"}. Keep the {argument name="location" default="original location"} recognizable, but transform the presentation into a curated architectural study rather than a normal travel poster. * Place the main landmark as a large monochrome architectural cutout or finely detailed ink drawing. * Add a smaller original-color photographic fragment showing one distinctive detail of the location. * Include a simplified topographic contour map behind the architecture. * Add tiny numbered reference points, measurement marks, arrows, elevation lines, and architectural annotation symbols. * Use an aged cream paper background with subtle grain and imperfections. * Palette: {argument name="palette" default="black, warm beige, muted olive, and one restrained accent color extracted from the landmark"}. * Typography should be small, elegant, and archival: location name, country, year, No. 019, and 3 descriptive keywords. * Create subtle imperfections resembling screen printing, drafting ink, pencil marks, and old paper. * Keep generous negative space and a sophisticated museum-catalog aesthetic. Important: Avoid postcard layouts, circular stamps, tourist graphics, excessive text, cartoon styling, glossy effects, 3D rendering, and overly detailed illustrations. Overall mood: architectural archive × museum specimen × old survey document × contemporary editorial design.

Goal: Create a richly detailed ancient Chinese ink-and-wash strategic visualization map of {argument name="battle name" default="赤壁之战"}, showing the Battle of Red Cliffs as a historical military map with painterly mountains, rivers, fleets, arrows, burning ships, labeled cities, and parchment texture. Canvas: Horizontal 4:3 map poster, aged tan rice-paper background with darkened worn edges, decorative corner border ornaments, ink stains, faded seals, and a hand-painted Song/Ming dynasty atlas aesthetic. Use a slightly elevated bird’s-eye view with the Yangtze River as the dominant blue waterway cutting across the center. Main title area: Place a huge black-and-dark-red Chinese brush-calligraphy title in the upper left reading {argument name="main title" default="赤壁之战"}. Beneath it, add a dark red banner subtitle reading {argument name="subtitle" default="孙刘联军 vs 曹操"}. Add a small supporting line below in fine Chinese type: “天下大势 · 江河为战 · 火映赤壁 · 三国鼎立”. Map geography and labels: Show exactly 11 major geographic/place labels in Chinese across the map: 1) 武陵 on the left-central mountains, 2) 巴丘 lower left of the river, 3) 长江 written large on the central river, 4) 南郡（今江陵）upper center-right, 5) 荆州 upper right, 6) 江夏（今武昌）right of river, 7) 夏口（今汉口）far right river bend, 8) 东吴 far right, 9) 赤壁（今赤壁市）near the central fire battle, 10) 乌林（曹军大败之地）below Red Cliffs, 11) 柴桑 lower right. Include misty mountain ranges, watchtowers, walled towns, small villages, docks, and forested islands. Battle layout: Depict exactly 2 opposing forces. The northern/upper force is Cao Cao’s army, colored blue, with a blue vertical banner “曹” and a label “曹操军 约二十余万”; show blue infantry ranks and fleets moving south along the river. The southern/lower-right force is the Sun–Liu allied army, colored red, with a large red banner “孙” and label “孙刘联军 约五六万”; show red infantry ranks and warships launching the attack. Use dramatic red fire arrows and burning ships at the central river near Red Cliffs. Add the key action label “黄盖火攻” in red near the flames, and label the blue fleet formation “曹军水军布阵（以铁链连船）”. Movement arrows: Include exactly 4 arrow systems: 1) blue dashed arrows descending from the north labeled “曹操南下 顺江而下”, 2) red attack arrows curving from the southeast toward the burning fleet labeled “黄盖火攻”, 3) green dashed retreat arrows running west/southwest labeled “曹军大败 向华容道撤退” and “至华容道（曹军败退）”, 4) red reinforcement arrows from the far right labeled “来自江上游（孙刘联军会师）” and “至柴桑（朱吴后方）”. Inset information panel: In the upper right, add a parchment-framed rectangular inset with a small monochrome river battle illustration and exactly 4 Chinese text lines: 1) “时间：建安十三年（208年）”, 2) “关键看点：火攻、联军、长江水战”, 3) “曹操南下，意图一统江南；”, 4) “孙刘联手于赤壁以火攻破敌，曹军大败，三国鼎立之势由此形成。” Add a small red seal marked “三国”. Legend and scale: In the lower right, create a framed legend titled “图例” with exactly 6 legend items: 1) blue dashed arrow “进军路线（曹操军）”, 2) red arrow “火攻方向（孙刘联军）”, 3) green dashed arrow “败退路线（曹操军）”, 4) blue boat icon “曹操军战船”, 5) red boat icon “孙刘联军战船”, 6) flame icon “火攻地点”. Include a simple scale bar marked 0, 25, 50, 100 里 and a compass rose labeled 北. Foreground atmosphere: Along the bottom edge, add a dark silhouetted battlefield foreground with wooden watchtowers, banners, spears, soldiers, smoky mist, and a broken black war flag bearing the phrase “一分天下 始于此战”. Visual style: Classical Chinese shanshui map painting mixed with modern infographic clarity; sepia parchment, muted teal-blue river, black ink mountains, red and blue army color coding, delicate Chinese typography, dramatic but readable. Use dense historical detail but keep all labels legible and map symbols consistent. No photorealism, no modern objects, no watermark.

Goal: Create a polished launch-feature hero graphic for {argument name="product name" default="GPT Image 2.5"}, showing a futuristic image-generation interface with multiple capability panels floating in a soft studio environment. Canvas: Wide 16:9 landscape composition, rounded-corner poster, premium tech-ad style. Use a warm off-white background with subtle glassmorphism, soft reflections, luminous circular rings on the floor, and shallow-depth-of-field greenery and stone foreground blur. Layout: Place the large headline “GPT Image 2.5” centered near the top, with “GPT” in black and “Image 2.5” in a blue-to-purple gradient. In the center, show one large rounded rectangular showcase image with a glowing border: a photorealistic Mediterranean coastal terrace at sunset, overlooking cliffs, villas, mountains, a calm sea, and a small sailboat; include a round wooden table with flowers, a cup, a book, and a cozy chair in the foreground. Feature panels: Include exactly 4 surrounding feature areas, each with a visible label. Top left panel labeled “Precise Edits”: two side-by-side dog photos of a golden retriever indoors, separated by a white arrow; the left image has a dashed selection box around a yellow tennis ball, the right image shows the edited object as a soccer ball with a dashed box. Bottom left panel labeled “GIF Creation”: a sequence of exactly 5 overlapping rounded cards showing a golden retriever running and splashing through water, progressing from faint/transparent frames to a sharp final action frame. Top right panel labeled “Sketch to Image”: a simple pencil sketch of an armchair on the left, a white arrow, and a photorealistic green cushioned wooden armchair with a draped blanket on the right. Bottom right panel labeled “Natural Detail”: a 2 by 3 grid of exactly 6 close-up texture tiles: human eye, glossy dark hair, water droplets on glass, reflective curved metal, wood grain corner, and green woven fabric. Visual style: High-end glossy product marketing render, photorealistic image thumbnails, crisp typography, rounded translucent cards, soft drop shadows, white glow trails connecting the panels, balanced symmetry, warm sunlight mixed with cool blue-purple UI accents. Constraints: Keep all text legible and limited to the headline plus the four labels. Do not add extra logos, watermarks, menus, buttons, or unrelated icons.

You are an information editor, diagram designer, and 3D art director. - Compare 3 visual proposals matching the content, and adopt the most intuitive option. - Explain using objects or materials specific to the subject. Use books, bamboo slips, or paper for classical literature; parts or assembly for manufacturing. - Elegant matte materials, soft lighting, and natural depth. Use 3D representations to convey sequence, relationships, and changes.

Goal: Create a realistic vertical smartphone screenshot of a ChatGPT conversation showing an AI-generated Chinese history infographic about {argument name="topic" default="安史之乱"}. Canvas: 576 × 1024 portrait phone screen, clean white iOS-style interface, high-resolution screenshot look with soft shadows and rounded UI elements. Top status and navigation: Show the iPhone status bar with exactly 5 visible status items: time “8:47”, a small user/person icon, black Dynamic Island with one tiny orange dot, cellular/Wi‑Fi indicators, and a battery badge reading “16”. Below it, show exactly 3 floating navigation controls: a left circular hamburger/menu button with a small blue notification dot, a right rounded pill containing an edit/compose icon, and a three-dot menu icon. Chat layout: On the right side near the top, place one light-blue user message bubble containing exactly this Chinese text: “生成图片：全面讲一讲\n安史之乱”. Beneath it, show one generated image card with rounded corners. Add exactly 2 small action icons under the card on the left: a share/upload icon and an ellipsis icon. Add one white circular share button overlay at the lower-right corner of the generated image card. Generated image card content: The card is a dense, horizontal Chinese educational infographic poster about the An Lushan Rebellion. Use an antique parchment and ink-wash historical style with dark red section headers, beige paper background, weathered borders, and a dramatic top banner painting of a burning Tang dynasty city, smoke, soldiers, banners, and a dark horse silhouette. Large brush-calligraphy title at upper left: “安史之乱”. Include a smaller subtitle about the major turning point of the Tang dynasty from prosperity to decline. Keep all infographic text in Chinese and make it appear mostly legible. Infographic sections: Use exactly 9 visible content sections in the poster: 1) “什么是安史之乱?” explanatory text block, 2) “主要人物” row with exactly 5 illustrated portrait figures labeled An Lushan, Shi Siming, Tang Xuanzong, Tang Suzong, and Yang Guifei, 3) “叛乱的原因” list with exactly 4 numbered causes, 4) “叛乱的过程” vertical timeline with exactly 6 dated milestones, 5) a central “叛乱形势图” map of Tang China with red and black arrows showing rebel and imperial movements, 6) “重要战役与事件” column with exactly 4 thumbnail battle/event images, 7) “叛乱的影响” block with exactly 4 icon-and-text impact items, 8) “历史意义” block with exactly 3 icon-and-text meaning items, and 9) “后世评价” block with exactly 3 quote cards. Add a bottom slogan line in Chinese resembling “以史为鉴，方知兴替。” with a small grayscale battlefield silhouette along the bottom edge. Bottom input bar: At the bottom of the phone screenshot, show the ChatGPT message composer as one rounded white pill with exactly 4 elements: a plus icon, gray placeholder text “询问 ChatGPT”, a microphone icon, and a blue circular voice-wave button. Visual style: Modern ChatGPT mobile UI combined with a richly detailed Chinese history infographic. The screenshot should feel like an authentic mobile app capture, with white margins, subtle drop shadows, rounded corners, and no external watermark. Constraints: Do not add extra chat messages. Keep the infographic fully contained inside the rounded image card. Preserve the visible Chinese text where specified. Use accurate Tang dynasty historical visual motifs and avoid fantasy elements.

Create a clean infographic summarizing {argument name="subject" default="Teva’s revenue report"}, with clear headings and a {argument name="chart type" default="chart showing annual growth"}.

【主题 / 作品名】：{argument name="主题" default="____"} 【左侧状态】：{argument name="左侧状态" default="____"} 【右侧状态】：{argument name="右侧状态" default="____"} 【关系类型】：正面与背面 / 概念与成品 / 来源与传播 / 实体与转译 / 其他 【中轴变化】：____ → ____ → ____ 【中文标题】：____ 【英文标题】：____ 【主色调】：____ 【辅助色】：____ 【画面比例】：9:16 生成一张具有成熟 Editorial Design 与展览视觉质感的左右构图海报。 不要简单把画面切成左右两栏，而是让左右两个视觉区域表达同一个主题的不同状态，并让“中间边界”成为变化真正发生的位置。 左右区域采用不等比例布局，根据主体重量使用 40:60、45:55、55:45 等关系，不要机械 50:50。左侧负责建立起点或原始状态，右侧负责展示变化后的结果，例如： 正面作品 → 背后的结构与制作路径； 设计草图 → 建成后的真实空间； 声源主体 → 向外扩散的振动与声场； 真实实体 → 纸面上的拓印、痕迹或图像。 最重要的是中轴转译。不要用明显竖线分隔，而要选择一个核心元素穿过左右区域，并在跨越中轴时连续发生变化. 例如：花瓣逐渐化成针脚和丝线；建筑线稿逐渐变成真实墙面；器物震动逐渐扩散成声纹；石刻浮雕逐渐压平成为黑墨拓印。 这种变化必须连续、自然，让观众一眼理解左右两边存在因果、过程或转译关系，而不是两张无关图片拼在一起。 Typography 也参与左右关系，但所有文字必须属于作品本身。使用真正的作品标题、副标题、短句 and 展览信息，不出现“左右构图”“设计案例”“视觉研究”等模板占位文字。 图像较重的一侧减少信息，较空的一侧可以承担更多标题与文字，通过字体大小、密度和位置完成视觉配重。 色彩控制在 2–4 个主要色系。左右可以有明显明暗、冷暖或材质差异，但至少保留一种颜色、纹理或视觉元素跨越中轴，保证整张作品统一。 最终阅读顺序：第一眼看到左右两种明显不同的视觉状态；第二眼发现中轴处正在发生连接 and 变化；第三眼理解“左边为什么会变成右边”；最后再阅读标题与作品信息。 整体要求：两个视觉域各有作用，中间有明确转译过程，左右不平均但视觉平衡，既有内容逻辑，也有成熟作品感。