Nano Banana Pro / Image Model: Prompts & Workflows

Cluster collects Nano Banana Pro image prompts, workflow notes, and prompt-engineering principles for photoreal outputs. - Multiple sample prompts for city scenes and seasons - Prompts specify time, lighting, camera style, and foreground subject - Author states outputs are 100% AI, not real photos (claim) - Two prompt principles: brevity and broad adaptability - 7000-word workflow notes summarize a 2-hour Nano talk - Collections: site lists 500+ Nano Banana Pro prompt recipes - Prompt example converts anime characters to hyperreal humans - Prompts include technical camera specs and negative prompts - Some users format prompts as JSON-like structures - Emphasis on giving model authority to adapt scenarios
Key insights
Nano Banana Pro achieves photoreal scene generation
Multiple example prompts produce high-fidelity photoreal images of real-world scenes (Paris café, Tokyo rain, Santa Monica pier, NYC Central Park, etc.), each specifying date/time, lighting, foreground subject, background blur, and an iPhone-like photographic style. One post explicitly states these images are entirely AI-generated rather than photographs.
Concise and adaptable prompts are recommended
A posted prompt principle advises keeping prompts as short as possible to save tokens and making prompts broadly applicable so the model can decide scenario-specific details. The guidance prioritizes granting the model authority to adapt and be creative within concise instructions.
Resources and workflows available for creators
A 7000-word note compresses a two-hour Nano Banana Pro lecture, detailing a creation workflow. A separate site curates 500+ Nano Banana Pro prompt recipes. These resources aim to teach end-to-end prompt and production workflows for creators.
Technical prompt templates include camera and negative specs
Example prompts provide explicit camera settings (e.g., mobile flagship, lens, aperture f/8–f/11 for deep depth of field), image resolution targets (4K+), lighting styles, and reverse negative prompts to avoid cartoonish or low-quality artifacts. Some users structure prompt inputs in JSON-like formats.
Prompt examples include character-to-photoreal conversions
A prompt example instructs converting an anime character into a hyperreal human while strictly preserving hairstyle, clothing, facial features, and overall temperament, with deep depth of field and cinematic mobile photography styling.
Source posts
- Original post by @FinanceYF5 on X
- Original post by @op7418 on X
- Original post by @howie_serious on X
- Original post by @Jackywine on X
- Original post by @dotey on X
- Original post by @FinanceYF5 on X
- Original post by @FinanceYF5 on X
- Original post by @FinanceYF5 on X
- Original post by @FinanceYF5 on X
- Original post by @FinanceYF5 on X
- Original post by @FinanceYF5 on X
- Original post by @FinanceYF5 on X
- Original post by @FinanceYF5 on X
- Original post by @FinanceYF5 on X
- Original post by @Jackywine on X