Five difficult points to check with this prompt:
If this one image is generated, please zoom in and check the following points.

- Text Consistency (AI EVOLUTION 2026)
Is the text written on the cardboard sign rendered correctly without spelling mistakes, while maintaining a "handwritten" texture?

- Anatomical Accuracy and Interaction
Are the old man's "10 fingers" naturally holding the fragile object, a glass butterfly? Are the fingers dissolving into the background or is the number incorrect?

- Material Differentiation (Material Mix)
Are materials with completely different reflectivity and transparency—"old man's skin (wrinkles)," "chrome plating (mirror surface)," "glass (transparency)," "velvet (napped)," and "liquid (steam)"—depicted without contradiction under the same light source?

- Calculation of Space and Light (Ray Tracing Understanding)
Is the blue luminescence of the android reflecting correctly on the old man's glasses and surrounding metal? Are there no contradictions in the shadows caused by multiple light sources?

- Prompt Adherence and Attribute Separation
Is the specified arrangement of "old man on the left" and "android on the right" maintained? Is "Color Bleeding" occurring, where the color of the "red velvet" transfers to the android's skin or the old man's clothes turn blue?
Infographic / Edu Visual - Benchmark Prompt for Nano Banana/Gemini Image Generation
0
0
0

A detailed, multi-faceted benchmark prompt designed to test the limits of image generation models like Nano Banana/Gemini by checking for five specific 'difficult points' related to text consistency, anatomical accuracy, material rendering, spatial lighting, and prompt adherence/attribute separation.

Language
EnglishDeutschEspañolFrançais日本語한국어Português中文
Created3 months ago
Last updated3 months ago
Creator

Services with a clipboard icon will copy the prompt to your clipboard first.

Version History
Prompt documentation
Comments (0)
Please log in to leave a comment.

Be the first to comment

to start the conversation.

Related Prompts
English中文Español日本語PortuguêsDeutsch한국어Français

Stop listing specs. Start selling value. This prompt converts your raw technical features (e.g., "WebSockets support") into high-converting marketing copy (e.g., "Real-time collaboration"). Ideal for Indie Hackers and Engineers.

1
0
0

EnglishDeutschEspañolFrançais日本語한국어Português中文

A structured prompt template for generating hyperrealistic food photography, specifically an exploded view of a dish with each ingredient layer floating in mid-air. It includes placeholders for the dish name, color scheme, and up to seven ingredient layers, designed for premium editorial food advertising.

1
0
0

EnglishDeutschEspañolFrançais日本語한국어Português中文

This prompt set is for generating ultra-realistic food photography, including a motion prompt for video generation (Kling AI) and two image prompts: one for a standard, high-detail pizza slice, and another for an exploded infographic view of the pizza layers.

1
1
0

EnglishDeutschEspañolFrançais日本語한국어Português中文

A detailed prompt for generating a technical infographic poster in the style of a vintage blueprint, illustrating the timeline and evolution of a product (placeholder), emphasizing sharp ink lines, watercolor accents, and high information density.

1
0
0

EnglishDeutschEspañolFrançais日本語한국어Português中文

A detailed prompt for Nano Banana 2 to create a watercolor-style map of a specified country, focusing on clean design, high legibility, and specific artistic constraints (no cities, roads, flags, logos, or watermarks).

1
0
0

EnglishDeutschEspañolFrançais日本語한국어Português中文

A prompt used to generate missing numbered assets (11 and 12) for a Canva design, ensuring the AI matches the style, texture, and color of the existing assets (1-10) provided as a reference image.

1
0
0