Nano Banana 2 retro photo prompts: 6 controlled edits

Runbo Li
Runbo Li
·
· 5 min read
ai

Quick answer

To make a convincing retro photo with Nano Banana 2, describe the capture medium, lighting, color response, texture and era-specific production details while naming what must remain unchanged. Upload a source you can use, apply one style direction, compare the whole result with the original, and continue only from an approved version.

Nano Banana 2 is Gemini 3.1 Flash Image, the current general-purpose model in Google’s image family. The original Nano Banana name refers to the legacy Gemini 2.5 Flash Image model. This prompt guide uses the current family and was checked September 13, 2026.

Retro edit workflow

  1. Choose one era and medium. A 1970s studio portrait, 1990s point-and-shoot and Y2K catalog image need different light, color and texture.

  2. Prepare the source. Use the highest-quality original you are allowed to process; crop only after deciding the delivery ratio.

  3. Paste one prompt. Keep the subject and scene constraints explicit instead of stacking several aesthetics.

  4. Inspect the full image. Check identity, hands, hair, clothing seams, text, logos, reflections, objects and background.

  5. Correct one problem at a time. Continue from the last approved version and name the exact region to repair.

  6. Export and document. Save the original, prompt, model, selected output, edits, rights and date.

Create one retro photo edit

Upload a photo you can use, choose one era prompt, and compare the result with the original for identity, hands, text, clothing and background before iterating.

Edit a Photo

A reusable preservation suffix

Append this to any prompt when those properties should remain stable:

“Preserve the subject’s identity, facial features, expression, pose, hands, body proportions and clothing shape. Preserve the camera angle, crop, background layout and all existing readable text. Change only the photographic treatment described above. Add no logos, jewelry or objects.”

A preservation instruction improves reviewability but does not guarantee exact fidelity. Compare every output with the original at full resolution.

1. 1970s warm studio portrait

Restyle this as a professionally scanned 1970s color studio portrait. Use a warm tungsten key light, soft fill, muted earth tones, gentle highlight bloom, fine film grain and slightly softened lens contrast. Keep skin texture natural. Preserve the subject and composition exactly. Do not add period props or change the clothing design.

Check: skin tone, highlight clipping, garment color, invented accessories and whether the backdrop changed shape.

2. Direct-flash 1990s snapshot

Prompt: “Make this look like a late-1990s compact-camera print scanned from a family album. Use direct on-camera flash, a slightly dark background, modest red-eye risk without distorting the eyes, consumer-film color, fine print grain and a narrow white photo border. Preserve every person, face, hand, pose and object. Add no timestamp unless requested.”

Check: eye detail, flash shadows, background people, fingers and any invented date or camera mark.

3. 1980s editorial portrait

Prompt: “Transform only the photographic treatment into an early-1980s editorial portrait. Use a controlled hard key light, subtle rim light, deep but readable shadows, restrained color saturation, crisp medium-format detail and light print texture. Preserve the subject’s identity, hairstyle shape, pose, wardrobe construction and background geometry.”

Check: face shape, hairline, fabric edges, shadow detail and whether the model introduced era stereotypes you did not request.

4. 1960s black-and-white contact print

Prompt: “Render this as a carefully printed 1960s black-and-white documentary photograph. Use panchromatic monochrome response, moderate contrast, visible but fine silver-halide grain, natural edge softness and a thin contact-print border. Preserve the exact people, scene, signage and crop. Do not simulate damage, scratches or dust.”

Check: readable signage, facial separation from the background, lost shadow detail and fake damage.

5. Y2K catalog product photo

Prompt: “Restyle this product image as a clean early-2000s mail-order catalog photograph. Use an off-white seamless background, broad softbox light, small natural contact shadow, slightly cool color balance and restrained digital sharpening. Preserve the product shape, label, logo, color, material and proportions exactly. Add no copy, badges or packaging.”

Check: label spelling, logo geometry, product dimensions, material texture and new claims or badges.

6. Faded instant-film portrait

Prompt: “Create the look of an aged instant-film print without changing the scene. Use a soft central exposure, mild cyan shadows, gently faded warm highlights, low micro-contrast and a wide blank instant-film border. Preserve the subject’s identity, hands, clothing, pose and background. Do not add handwriting, dates, light leaks or heavy damage.”

Check: face fidelity, hand anatomy, clothing color, border shape and invented writing.

How to repair a near-miss

  • Identity drift: restart from the source and reduce the requested change to light, color and texture.

  • Unwanted props: list the additions to remove and repeat that no objects may be introduced.

  • Damaged text or logo: restore it from the original in a deterministic editor or mask it from the generative edit.

  • Too much grain or damage: ask for fine grain and a clean professional scan, then prohibit scratches, dust and light leaks.

  • Wrong decade: describe the camera, light, palette and print process; the year alone is too ambiguous.

  • Repeated-edit degradation: return to the last approved file instead of correcting a correction.

Choosing the right Nano Banana model

Google’s current image-generation guide recommends Nano Banana 2 for general work, Nano Banana 2 Lite when speed and cost dominate, and Nano Banana Pro for complex professional assets. Google labels the original Gemini 2.5 Flash Image model as legacy. The Gemini Apps help page explains current app access, age, plan and regional limits.

For the exact model IDs, app steps and multi-reference workflow, use the complete Nano Banana 2 tutorial.

When another workflow fits better

Use Magic Hour’s AI Image Editor for a guided source-image edit, or the AI Image Generator when you want a new retro scene without a source photo. Browse AI art styles for medium vocabulary. After the still passes review, use Image-to-Video if the final asset needs motion.

Frequently asked questions

Name the era, capture medium, lighting, color response, contrast, texture and print treatment. Then list the subject, composition and product details that must remain unchanged.

Start from the original, request one style-only change, explicitly preserve identity and compare the entire face with the source. If identity drifts, restart from the original with a narrower prompt.

Only when they support the specific reference. Heavy damage is a generic retro cliché and can hide important detail. A clean period print is often more credible.

Commercial use depends on your rights to the source image, people, products, marks and output, plus the current tool and platform terms. Review those inputs and the final asset for the intended campaign.

No. Nano Banana 2 is Gemini 3.1 Flash Image. The original Nano Banana is Gemini 2.5 Flash Image, which Google now describes as the legacy model in the family.

Runbo Li
Runbo Li
CEO of Magic Hour
Runbo Li is the Co-founder and CEO of Magic Hour, where he builds AI video and image tools for content creation. He is a Y Combinator W24 founder and former Data Scientist at Meta, where he worked on 0-1 consumer social products in New Product Experimentation. He writes about AI video generation, AI image creation, creative workflows, and creator tools.
View author →

Continue Reading

Nano Banana AI Editing Guide
Nano Banana 2 image editing tutorial (2026)
The Best AI Image Editors
6 best AI image editors in 2026: choose the right one
Collage of the best AI image generator logos
10 best AI image generators: features, costs and examples
best ai image and video apis
9 best AI image and video APIs: costs and integration
Collage of logos from the best text-to-image APIs.
Text-to-image API integration guide: request to accepted image
Imagen
Google Imagen 4 guide: capabilities, current access, and alternatives