Nano Banana is conversational: instead of crafting a single mega-prompt, you describe what you want in plain language and iterate by asking the AI to 'make it warmer', 'add a character on the left' or 'remove the cup'. Each turn keeps the previous image as context, so refinement feels more like talking to an art director than tweaking parameters.

Be specific about subject, environment, lighting and style — for example 'a 35 mm film photo of an elderly Japanese fisherman at golden hour, shallow depth of field' beats 'old man fishing'. After the first generation, refine in small steps: change one thing per message. You can also paste reference images and ask for a similar look.

Yes. Send the previous image back into the chat and ask for the change you want — 'make the sky stormy', 'replace the dog with a cat', 'extend the canvas to 16:9'. Nano Banana keeps composition and identity stable across edits, which is much harder with prompt-only generators.

You own the images you generate and can use them commercially — social media, ads, blog headers, book covers, merchandise. We block obvious trademark and named-celebrity prompts to keep your output safe to publish; for regulated industries (medical, financial) please review the result with your legal team before shipping.

Square (1:1), portrait (3:4, 9:16) and landscape (4:3, 16:9) are all available, with output up to roughly 2K on the long edge. For larger prints, generate at 1:1 then run the result through our AI Image Enhancer for a 4× upscale.

Free credits on signup, then a flat per-image cost on paid plans — typically much cheaper than premium ChatGPT or Gemini tiers because we don't bundle a chat assistant. If you only need image generation and editing, Nano Banana is the most cost-efficient option in our toolkit.