เอกสารคำถามภาษาจีนสำหรับ GPT ภาพที่ 2

A clean screenshot-like image showing a dark charcoal rounded-rectangle panel filled with dense simplified Chinese prompt text in white sans-serif font, presented as a text document or note on a flat dark background. The composition is a single full-frame panel with generous padding and subtle shadow, like a modern UI card. At the top is a sentence describing the creation of a high-end technology magazine double-page spread poster, with the English title in quotation marks: “GPT IMAGE 2: From Ideas to Images”. Below it are multiple left-aligned paragraphs of Chinese text with consistent line spacing, describing layout ratio, modern editorial design style, and a detailed two-page composition. In the middle section, there is a numbered list with exactly 6 items labeled 1 through 6, each item naming a module in English inside quotation marks followed by Chinese explanations. The 6 module names are exactly: “Photorealism”, “Readable Text”, “Product Mockup”, “Character Consistency”, “Storyboard”, and “World Knowledge”. Under item 2, include 5 separate quoted multilingual text lines as subitems, exactly these: “IDEAS → IMAGES”, “想法 → 图像”, “アイデア → 画像”, “아이디어 → 이미지”, and “فكرة → صورة”. Later in the text, add a section specifying the main title and subtitle as clearly readable quoted English lines: “GPT IMAGE 2” and “From Ideas to Images”. Near the lower portion, add another Chinese line introducing a small caption at the bottom right, followed by the exact quoted English caption: “Typography • Photorealism • Layout • Consistency • Multilingual Text”. At the bottom, include a final Chinese paragraph about visual style requirements such as premium ad feel, Apple-style product photography, Swiss graphic design, soft natural light, refined materials, clear layout, ample whitespace, and avoiding low-quality textures, blurry text, gibberish, or deformed fingers. The text should appear crisp and fully legible, with realistic screen-capture fidelity, high contrast, subtle anti-aliased edges, no illustrations, no icons, no photos, no extra interface chrome beyond the single dark card, and an overall polished editorial-tech presentation.