社论概念插图设计提示词系统
来源: @Gdgtify模型: Nano Banana 2
📸 示例图片

提示词
English
2x2 grid, AI infers topics, 16:9 CLASSIFY(input) → {
domain: editorial_illustration × conceptual_art × political_commentary
intent: persuade + provoke + distill (the "unforgettable glance")
audience: op-ed readers, magazine subscribers, visual intellectuals
platform: print (magazine cover, editorial header), scalable to 4cm thumbnail
complexity: extreme (requires strict metaphor singularity, physical print simulation,
and absolute rejection of literalism and digital rendering)
}
CORE DESIGN LAWS (IMMUTABLE):
⚠ ONE METAPHOR. ONE SUBSTITUTION. ZERO TEXT.
• Substitution: Change Scale XOR Material XOR Function. Never two.
• Restraint: No text, no arrows, no forehead symbols. Ambiguity ≠ Confusion.
• Treatment: 3-4 flat colors, 1 texture, shape-first, generic faces.
• Medium: Physical print character (Riso, screenprint, litho).
• Survival: Must read perfectly at 4cm wide.
SEMANTIC_VARIABLES (AI infers entirely from {{THESIS}}):
{{THESIS}} → [USER PLUGS IN: e.g., "Tech monopolies are extracting
behavioral data like 19th-century colonial rubber barons."]
{{BENEFICIARY}} → AI infers the winner (e.g., The Algorithm / The Corporation)
{{VICTIM}} → AI infers the loser (e.g., The User / The Public)
{{TENSION}} → AI infers the core conflict (e.g., Extraction vs. Autonomy)
{{FAMILIAR_SCENE}} → AI infers the baseline visual (e.g., A Victorian rubber plantation)
{{SUBSTITUTION}} → AI selects exactly ONE wrong element (Scale, Material, OR Function)
{{PALETTE}} → 3-4 flat colors inferred from the tension
{{TEXTURE}} → The single print artifact (Riso grain, screenprint halftone)中文
2x2 网格,AI 推断主题,16:9 分类(输入) → {
领域: 社论插图 × 概念艺术 × 政治评论
意图: 说服 + 挑衅 + 提炼(“难忘的一瞥”)
受众: 专栏读者、杂志订阅者、视觉知识分子
平台: 印刷品(杂志封面、社论页眉),可缩放到 4cm 缩略图
复杂度: 极高(需要严格的隐喻单一性、物理印刷模拟,并绝对拒绝写实主义和数字渲染)
}
核心设计法则(不可更改):
⚠ 一个隐喻。一个替换。零文本。
• 替换:改变 规模 XOR 材质 XOR 功能。绝不改变两者。
• 克制:无文字、无箭头、无额头符号。模糊 ≠ 困惑。
• 处理:3-4 种平涂颜色,1 种纹理,形状优先,通用面部。
• 媒介:物理印刷特性(Riso、丝网印刷、石版印刷)。
• 生存:必须在 4cm 宽时完美阅读。
语义变量(AI 完全从 {{THESIS}} 推断):
{{THESIS}} → [用户插入:例如,“科技垄断企业正在提取行为数据,就像 19 世纪的殖民橡胶大亨一样。”]
{{BENEFICIARY}} → AI 推断获胜者(例如:算法 / 公司)
{{VICTIM}} → AI 推断失败者(例如:用户 / 公众)
{{TENSION}} → AI 推断核心冲突(例如:提取 vs. 自主性)
{{FAMILIAR_SCENE}} → AI 推断基准视觉(例如:维多利亚时代的橡胶种植园)
{{SUBSTITUTION}} → AI 恰好选择一个错误元素(规模、材质 或 功能)
{{PALETTE}} → 从冲突张力中推断出的 3-4 种平铺颜色
{{TEXTURE}} → 单一印刷痕迹(Riso 颗粒、丝网印刷半色调)