Nano Banana Prompt Generator

What is Nano Banana?

“Nano Banana” is the popular name for Google’s Gemini image generation and editing model. It became famous for conversational photo editing: you upload a picture, describe a change in plain language, and the model edits the image while trying to keep the person, object or scene recognizable. It can also create images from text, combine several input images, and render short text. Google has released newer and “Pro” versions over time. You can use it in the Gemini app, in Google AI Studio and through the Gemini API; limits and features depend on your account, so check Google’s current details. Images made with Google’s models carry an invisible SynthID watermark.

Because the model understands natural language well, the best prompts read like clear instructions to a skilled photographer or retoucher, not like a pile of keywords.

Two kinds of Nano Banana prompts

Generation prompts

Describe the subject, setting, composition, lighting, style, colors and mood in a few full sentences. Mention the aspect ratio you want.

Edit prompts

Refer to the uploaded image, state the single change you want, list what must stay identical, and ask the edit to match the existing light and perspective.

A structure for edit prompts

1. Reference: “Using the uploaded photo…”
2. Change: one clear edit (“replace the background with…”, “change the jacket to red leather”).
3. Preserve: what must not change (“keep the face, hairstyle, pose and framing exactly the same”).
4. Blend: how it should look (“match the warm afternoon light and add natural shadows”).
5. Output: optional notes such as aspect ratio or “photorealistic”.

Nano Banana prompt examples

A photorealistic close-up of a cozy coffee shop window on a rainy evening. A white ceramic cup sits on the wooden sill with steam curling upward, raindrops streak the glass, and warm city lights glow as soft bokeh outside. Shot with a shallow depth of field, warm tungsten interior light against cool blue street tones. Calm, nostalgic mood. Square 1:1 format.

Using the uploaded portrait, replace the background with a sunny Mediterranean street with whitewashed walls and bougainvillea. Keep the person’s face, expression, hairstyle, clothing and pose exactly the same, and keep the original framing. Match the lighting on the person to bright late-afternoon sun coming from the left and add a soft natural shadow.

A minimalist poster for a jazz night. A silhouette of a saxophone player against a large deep-blue circle on a cream background, with subtle paper texture. At the top, the words "Blue Hour Sessions" in bold retro sans-serif lettering; at the bottom, "Friday 9 PM" in smaller type. Flat vector style, limited palette of navy, cream and mustard. Portrait 3:4.

Tips for better Nano Banana images

• check_circleEdit step by step. Ask for one change, review it, then ask for the next change in a follow-up.
• check_circleBe explicit about identity: “keep the face exactly the same” helps preserve likeness in edits.
• check_circleUse photography language for realism: lens, depth of field, light direction, time of day.
• check_circleQuote any text exactly and keep it short; shorter text renders more reliably.
• check_circleWhen combining images, say which image provides which element (“the dress from image 2 on the person in image 1”).
• check_circleState the aspect ratio in words if your interface does not offer a setting.

Mistakes to avoid

• check_circleAsking for many edits at once, which increases the chance that faces or details drift.
• check_circleLeaving out the “preserve” part of an edit prompt.
• check_circleUsing keyword soup (“8k, masterpiece, ultra detailed”) instead of describing the scene.
• check_circleEditing photos of other people without their permission, or creating misleading images of real people.
• check_circleExpecting pixel-perfect logos or long paragraphs of text.

From image to video

A Nano Banana image makes an excellent first frame for image-to-video models. Create the frame here, then animate it with a motion prompt from our Runway Gen-4 or Kling generators. To reverse-engineer a picture you already have, use Image to Prompt. For video ideas, see the Video Prompt Generator and Video to Prompt, and for structured prompt libraries the JSON Prompt Generator.

Frequently asked questions