Here is a new way to visualize food. This one covers some of the best ones from my home country. 

Prompt: 
Function DrawDish(input dish name)
Input Variable: [insert dish name]
(e.g., ramen, biryani, tacos al pastor)

System Instruction:
Generate a hyper-realistic, 1:1 product photo of a "Bento Cross-Section Diorama" (miniature food exhibit).

1. Semantic Extraction:
> Identify cuisine, flavor profile (spicy, umami, sweet), and the dish’s signature components.
> Figure: Chibi “chef mascot” OR a cute ingredient character (e.g., a chibi chili pepper).
> Pose: Serving action or sprinkling garnish.

2.  Container (Bento Box):
> Structure: U-shaped bento display (two walls + floor) with compartment lines.
> Desktop size, premium lacquer look.

3. Background 
> Walls: B&W recipe-card typography, ingredient diagrams, plating sketches (original, inspired).
> Floor: Pop-up “steam” ribbons, spice flecks, herb confetti cut-outs rising up.

4. Integration (2D vs 3D):
> The 3D figure is “pulling” a 2D sauce swirl that becomes a 3D glossy ribbon.

5. Visual
> B&W design background; full-color food and figure.
> Soft studio lighting with appetizing highlights.
> Label: "[Dish] — [Region/Cuisine]"

Return: ONE image, 1:1, ultra-real food miniatures, product-photo lighting.

<instructions>
Input: dish name
Analyze cuisine, flavor profile, typical ingredients, cultural context. Find 4 MUCH lesser-known dishes in the same cuisine/style.
Output: a 2x2 grid, each panel is DrawDish for a lesser-known recommendation.
</instructions>