
原始文档
GPT Image 2 · en
Scene:
A square commercial food-culture poster designed as a clean subway map on warm white paper. The style is bright, friendly, and precise, with flat vector lines, circular station dots, and a compact icon legend. Use appetite-friendly but restrained colors: chili red, noodle yellow, herb green, ocean blue, spice orange, rice white, and charcoal. The poster should feel like something sold in a culinary museum shop or used in a food festival exhibit.
Subject:
A thematic subway map of world cuisine relationships. Route lines represent staple grains, spice routes, fermentation, street food, noodle families, seafood traditions, baking, and tea-and-dessert culture. Transfer hubs show shared techniques such as steaming, grilling, pickling, broth-making, flatbread, dumplings, and chili oil. Station clusters imply regional families without drawing a real geography map. Small icons show bowls, chopsticks, bread, fish, jars, steam, leaves, and clay pots.
Important details:
The map must show relationship logic without readable cuisine names. Use line colors and transfer hubs to suggest shared ingredients and cooking methods. Keep the station intervals regular, the legend orderly, and the route intersections easy to follow. Avoid food photography; this is a technical-diagram poster with playful culinary icons. All label positions should use short non-readable glyph marks only. The composition should be dense enough to reward inspection but not so crowded that it loses transit-map clarity.
Use case:
Thematic Subway Map single study 7/8, testing culinary knowledge abstraction, commercial poster appeal, and structured route readability.
Constraints:
No readable text, no watermark, no logos, preserve exact 1:1 aspect ratio, no real restaurant brands, no national flags, no food photos, no tourist map base, no messy ingredient pile, no fake menu text, no random icon clutter.使用方法
- 复制完整原始提示词,并保留其中的参数。
- 按需求替换主体、文字和构图要求。
- 在兼容的外部工具中使用,核对当前使用条件并检查结果。
开始前需了解
- 先确认是否需要参考图。标注兼容模型,不代表该示例由这个模型生成。
- 核对名称、数字与画面文字;不同模型版本的结果可能不同。





