PROMPT RECORD图像记录

对 $ dish 执行以下操作：函数 DrawDishHoudiniFluid(输入 dish) 输入变量：[INSERT DISH] 系统指令：生成一张超写实的 16:9 图像，作为 2x2...

中文说明

对 $ dish 执行以下操作：函数 DrawDishHoudiniFluid(输入 dish) 输入变量：[INSERT DISH] 系统指令：生成一张超写实的 16:9 图像，作为 2x2 网格的四个流体摆盘雕塑，用于推荐菜品。 A) 语义分析 —— 推断菜系方向、风味特征（辣/鲜/酸/甜）、口感目标（酥脆、丝滑、嚼劲）。 B) 推荐任务：—— 在同一菜系方向中找出 4 道 MUCH（极）小众的菜品。 C) 风格（非 Q 版 / NON-CHIBI）：—— Houdini 程序化流体：物理可信的酱汁粘度、泡沫气泡、微液滴、表面张力。 —— 角色服装须与地区相符，必须是写实的 1/12 比例厨师人偶（中性、无面部特征）。 D) 容器（每格）：—— U 形摆盘 “灯箱舞台”，中央放置一个盘子，如同产品基座。 E...

原始 Prompt

Do this for $ dish : Function DrawDishHoudiniFluid(input dish) Input Variable: [INSERT DISH]  System Instruction: Generate ONE hyper-realistic 16:9 image as a 2x2 grid of four Fluid Plating Sculptures recommending dishes.  A) Semantics - Infer cuisine lane, flavor profile (spicy/umami/sour/sweet), texture goals (crunch, silk, chew).  B) Recommendation Task: - Find 4 MUCH lesser-known dishes in the same cuisine lane.  C) Style (NON-CHIBI): - Houdini procedural fluid: physically plausible sauce viscosity, foam bubbles, micro droplets, surface tension. - character dressing appropriately for the region., it must be a realistic 1/12 chef mannequin (neutral, faceless).  D) Container (per panel): - U-shaped plating “light stage” with a single plate centered, like a product pedestal.  E) Background (B&W): - Walls: monochrome ingredient diagrams + flavor compound icons based on the ingredients in the dish. Must be accurate - Floor: B&W plating guides and contour lines.  F) 2D vs 3D Integration: - 2D contour lines rise into 3D sauce ribbons that form the dish’s silhouette. - Steam must be volumetric, subtle, believable.  G) Visual Syntax: - B&W environment; full color food and fluid effects only. - Lighting: soft food studio lighting, appetizing specular highlights. - Label: "[Recommended Dish] — [Region]"  Return: - ONE image, 16:9, 2x2 grid.