Images · guidance updated 22 Nov 2026
Google Gemini (Nano Banana 2 / Nano Banana Pro) prompt guide
Nano Banana 2.1 (paid tier) has enhanced quality, reasoning, and world knowledge. Nano Banana 2 (free) covers basics. Both use Gemini's native multimodal architecture. Gemini 2.5 Flash Image powers generation.
How to prompt Google Gemini (Nano Banana 2 / Nano Banana Pro)
- Describe scenes as narrative paragraphs with five core elements: specific subject, composition, action, location/time, aesthetic style.
- Use photography language for realistic work: shot type, angle, lens, aperture, lighting direction and quality.
- State canvas size and aspect ratio in words (e.g., '9:16 vertical poster', '21:9 cinematic').
- Put exact text in quotes; specify font, weight, color, position, and language.
- For edits, specify what changes and what stays unchanged; refine conversationally one adjustment at a time.
- Assign each reference image a clear role (pose from Image A, style from Image B, background from Image C).
- Include factual constraints for diagrams; leverage Gemini 3's real-world knowledge and reasoning.
- Frame exclusions positively ('empty, deserted street') rather than negatively ('no cars').
- Provide purpose and context ('logo for minimalist skincare brand') so the model infers correct style and quality.
- Use camera and lighting details to direct mood: golden-hour backlighting, shallow depth of field, cinematic grading.
- Build prompts using formula: [Action] + [Subject] + [Setting] + [Atmosphere]; add composition, style, quality and ratio.
- Compose with deep language understanding; narrative descriptions outperform disconnected words.
The shape of a good prompt
Create a [image type/style] of [specific subject details] [action], in [location/time]. [Composition: shot type, angle, framing]. [Lighting and atmosphere]. [Camera: lens, depth of field, colour grading]. [Text: "exact words" in position, font style, colour]. [Aspect ratio in words]. [Purpose/context for style inference].
Avoid
- Comma-separated keyword lists with no relationships.
- Negative phrasing ('no people', 'without text').
- Vague one-line prompts without purpose or context.
- Ambiguous pronouns in edits ('make it bigger' without specifying what).
- Midjourney-style parameter syntax (--parameter).
- Skipping reference-image role assignment when using multiple images.
Example
Request a cozy coffee shop poster for autumn
Create a vertical 2:3 poster in a vintage screen-printed illustration style for a neighbourhood coffee shop's autumn menu. A steaming ceramic mug of spiced latte sits on a wooden window ledge in the foreground; behind it, a rainy street at dusk with maple leaves scattered on the wet pavement and warm amber light glowing from the café's tall windows. Soft, cozy atmosphere with a burnt orange, mustard and deep teal palette and visible paper grain. The headline "AUTUMN BREWS" runs across the top in bold cream serif capitals, and the line "Pumpkin spice • Maple latte • Chai" sits along the bottom in a clean sans-serif. The street is quiet and empty.