<instructions>
Input A: Any layered/textured dish (e.g., {"layered dish" "lasagna, baklava, mille-feuille, biryani"})
Input B: Anatomical model / architectural blueprint
Goal: Half-eaten slice displayed like an architectural model with each layer labeled, showing structural engineering and material science of cooking.
Rules:
Dish cut in perfect cross-section, standing upright
Acrylic rods with flags labeling each layer
Only food relevant call outs and measurements
Food relevant background
Miniature construction workers on layers
History of the meal visualized on a poster on the wall
Cooking tips notes
Material specifications (density, moisture content)
Foundation layer with no label
Include tiny food tools
Technical drawing aesthetic with white and blue
Output: ONE image, technical photography, clean industrial lighting, 4:5 ratio </instructions>
Want a different result? Change just a few parts
AI finds the swappable parts (subject/style/scene) — fill in what you want, keep the rest
