What it does This is an “awesome list” of copy-pasteable prompts for Nano Banana Pro (also called Nano Banana 2), Google’s AI image model. The repo collects prompts from Twitter, WeChat, and Replicate, organized into categories like photorealism, e-commerce, interior design, and photo restoration. Each entry includes the full prompt text, a sample image, and attribution.

The interesting bit The prompts read like cinematography call sheets rather than vague “make it pretty” instructions. One business-headshot prompt specifies a Sony A7III with 85mm f/1.4 lens, three-point lighting, and “visible pores, not an airbrushed look.” Another uses structured JSON to define a 2000s mirror selfie down to the CD player and beaded door curtain. The implicit thesis: treat the model like a camera crew, not a genie.

Key highlights

• Heavy focus on face consistency — multiple prompts explicitly demand “keep the face 100% accurate” or “preserve original facial features”
• JSON-structured prompts for complex scenes with nested subject/accessories/photography/background objects
• Categories span practical use cases: professional headshots, product photography, interior design, social media marketing
• Sources cited per prompt (Twitter handles, WeChat articles, Replicate links)
• CC BY 4.0 license; actively maintained with sponsor slots claimed at 20,000 daily visitors

Caveats

• The README is truncated mid-prompt in the available source, so later sections are incomplete
• Heavy ad load: two sponsor banners plus a direct pitch for paid slots before the table of contents
• No explanation of why these prompts work — just the prompts themselves; you’ll need to experiment

Verdict Useful if you’re already using Nano Banana Pro and want proven starting points for specific aesthetics. Skip it if you’re looking for general prompt engineering theory or don’t use this particular model.