Restore and Colorize an Old Photo

Photo sharpening and restorationPhoto 3:2 · Website images, blog coversHotels and spasRestaurants and cafésService businesses

Before · input

Input: a sepia-toned old photo, scratched, creased and with a torn corner; a stone building entrance and an olive tree in a pot — An in-house photo, deliberately aged: sepia, scratches, dust, creases, a torn corner.AI-generated · made by us

Output: the same building entrance, restored and in color; cream stone facade, wooden door, olive tree in a terracotta pot, blue sky — AI-generated · made by us

Prompt for the model

Restore this old, damaged photograph. Remove the scratches, dust spots, the crease line and the torn corner; fix the fading and the contrast. Then colorize it with natural, period-appropriate colors: [warm limestone facade, dark brown wooden doors, green olive tree in a clay pot, pale blue sky]. Keep the composition, the architecture, the proportions and every object exactly as in the original; do not modernize it and do not add people, signs or details that are not there. Keep a light natural film grain. Photorealistic result, not a painting.

Replace the gold [bracketed] parts to match your own product, space or brand.

Keep unchanged

The prompt tells the model not to change these. Check the result anyway: these are the parts that break most often.

Architecture and proportions
Composition
Existing objects
No new people or signs

When not to use it

Colorization is a guess: the real color of the facade, the door or the clothing is unknown. If you'll publish it with a claim like "this is what our business looked like when it was founded", state that it was "restored and colorized with AI" and keep the original.

Don't use family or archive photos with people in them in ads without the permission of those people or their relatives.

What it does and how to adapt it

Repairs and colorizes an old photo that is scratched, creased, faded or has a torn corner. For an established business's founding story, a hotel's history, a shop's early years.

If you know the main colors in the photo (e.g. the color of the facade, the door, the sign), write them in the square brackets; if not, leave "natural, period-appropriate colors". Scan the photo or shoot it with your phone, flat and without shadows.

The example was generated with Nano Banana 2 via Higgsfield.

Generation log

Model: Nano Banana 2 (Google)
Date generated: October 2, 2026
Attempts: Got it on the first try
Aspect ratio: 3:2
Generated via: Higgsfield API (selected: Nano Banana Pro; job record shows nano_banana_2)
[Fields] in the example: colors: warm limestone facade, dark brown wooden doors, green olive tree in a clay pot, pale blue sky
Resolution: 2k
Job ID: 38aee8db-edb5-4c65-ba37-a272eaac9eee

What happened across attempts: On the first attempt, the scratches, dust, crease and torn corner were cleaned up and the photo was colorized. The architecture, the steps and the silhouette in the background were preserved. The colors are the ones written into the prompt plus the model's guesses.

Even with the same prompt and the same settings, you won't get an identical result: models generate something different every time. The example shows what kind of result this prompt produces; it is not a guarantee.

If you'll use this image in an adIn Turkey, AI-generated ad content falls under a labeling rule: when a label is required and how to word it.