AI for Old Photos UK: Restore & Colourise

Last summer, I found a cardboard box at the back of a spare room. It held about sixty photographs from the 1950s and 1960s: black-and-white portraits, a family seaside trip, a christening gathering at a house that no longer exists. Several were creased or faded. One had a long scratch across the centre.

Within an afternoon, using free tools available to anyone with a phone or computer, most of those photographs were repaired or colourised. One now sits in a frame on the wall. This guide explains how to do the same with your own old prints.

What can AI do with an old photograph?

AI photo tools can handle several tasks that would previously have required professional darkroom work or hours of painstaking manual editing.

• Remove scratches, creases, dust marks, and small tears
• Repair fading and restore the original tonal range of the image
• Sharpen blurry areas, particularly around faces
• Add realistic colour to black-and-white photographs
• Increase the resolution of a small or low-quality scan so it is suitable for printing

Most of these tools work directly in your browser. You upload a photo, the AI processes it in seconds, and you download the result. The results are not perfect, and it helps to know the limits before you start. But for most family prints, the result is noticeably better than what you started with.

Step one: get your photos into digital form

Before any AI tool can help, the photo needs to be on your phone or computer. There are two ways to do this.

If you have a scanner: scan at 600 dpi or higher. Most flatbed scanners have a photo setting that selects the right resolution automatically. For very old or damaged prints, 1200 dpi gives the AI more detail to work with.

If you don't have a scanner: a phone camera works well enough for most purposes. Here is what to do:

1. Place the photograph flat on a table, making sure there are no ripples or shadows across it.
2. Find good even lighting, away from direct sunlight. Sunlight causes bright reflections on glossy prints that are difficult to remove later.
3. Hold your phone directly above the photo, parallel to it rather than at an angle. Tap the photo on screen to focus, then take the shot.
4. Check the image before moving on: the print should fill most of the frame and be in clear focus.

For better results, two free apps are worth trying. Microsoft Lens (available for both Android and iPhone) straightens the image and reduces lens distortion. Google PhotoScan is designed specifically for photographing prints with glossy surfaces: it takes multiple overlapping shots and stitches them together to remove reflections. Note that availability varies by device and country; if it is not in your app store, Microsoft Lens handles glossy prints well as an alternative.

The best free tools to restore old photos

These three options are the most accessible for someone starting out. All work in a web browser and require no software to install.

Canva

Canva's photo restoration tool is one of the simplest available. Upload your scanned image, click Restore, and the AI removes scratches, corrects fading, and improves general clarity. A free Canva account is sufficient; you do not need to pay.

We tested Canva on a faded colour print from the early 1970s: a family garden party with washed-out reds and a horizontal fold mark across the lower third. The fold disappeared entirely. The colours came back in a way that looked natural rather than oversaturated. The whole process took about forty seconds.

MyHeritage Photo Repair

MyHeritage is a genealogy platform that has developed strong AI photo tools in recent years. Their Photo Repair function handles damage typical of old family prints: yellowing, creases, small tears, and general loss of contrast. MyHeritage offers a limited number of free restorations before asking you to register.

We tested MyHeritage Photo Repair on a scan of a 1960s black-and-white family portrait: a formal group photograph with significant yellowing and a long horizontal crease across the lower half. The crease was almost entirely removed. Contrast across the faces improved noticeably, and the overall image looked closer to how it would have appeared when first developed. The process took under a minute.

The tool also connects directly with MyHeritage's family tree feature, which makes it a natural choice if you are researching your family history alongside preserving photos. More on that below.

ChatGPT

If you have a ChatGPT Plus account, you can upload a photograph and ask for suggestions about what to repair and which tools to try first. ChatGPT does not restore photos directly, but it is genuinely useful for planning an approach to a particular type of damage. See our ChatGPT for seniors guide if you have not used it before, or try a free AI conversation here to get comfortable with it first.

How to colourise a black-and-white photo with AI

Colourising a black-and-white photograph means the AI analyses shapes, textures, and context in the image and adds what it calculates to be historically plausible colours. It does not know what colour your grandmother's dress actually was. It makes an educated estimate based on patterns learned from millions of photographs.

The results are usually convincing in areas with predictable colours: skin tones, grass, sky, and stone walls. Clothing is less reliable because fashion varied so much across decades and regions. Interior furnishings are the least predictable of all.

MyHeritage In Color adds colour in a single click, within the same platform as Photo Repair. Skin tones are handled well, and the output for most portraits from the 1940s and 1950s looks natural rather than artificial.

Colorize.cc focuses specifically on colourisation and has a free tier. For most family portraits, the output is striking. The site also offers different AI models if the first result does not look right.

One practical note: save the original file before colourising. The AI's colour choices, while often convincing, are not a verified historical record. Keeping the original black-and-white alongside the colourised version means you have both: the vivid version to share and the authentic original to preserve.

Pairing photo restoration with family history research

For many people, old photographs and family history go hand in hand. A restored image can be attached to a family tree record, shared with relatives working on the same branch, or used in a printed family history book.

MyHeritage handles both functions in the same account, which makes it a natural starting point. A photo you repair in their tool is immediately available within your family tree. You can tag the people in it, link them to records, and share the restored version with relatives who have their own MyHeritage accounts.

If you are just starting to look into your family history using AI, our guide to using AI for family history research covers how AI tools can help interpret old documents, suggest record matches, and map out family connections in a way that would have taken weeks by hand.

Is it safe to upload old family photos to an AI tool?

This is a fair question. When you upload a photograph to a web-based tool, that image is sent to servers, often in the United States, for processing.

Under the UK GDPR, photographs of identifiable people count as personal data. Uploading them to a third-party service means that service is processing personal data on your behalf. Reputable tools such as Canva and MyHeritage have privacy policies stating they do not sell or share uploaded images. Each service has its own retention terms, so it is worth checking the current policy of any tool you use before uploading.

The Information Commissioner's Office (ICO) provides guidance on photographs as personal data. For personal use with a reputable named service, the practical risk is low. A few sensible precautions are worth taking:

• Avoid uploading photos of children to tools you are not familiar with.
• Check the privacy policy before using a new tool. Look for a clear statement that uploaded images are not used for AI training or shared with third parties.
• If a photo includes visible personal details in the background, such as a house number or a car registration plate, consider cropping those before uploading.

For a broader look at privacy with AI tools, see our piece on AI and your personal data. Our guide to staying safe with AI also covers what information is worth keeping to yourself online.

What AI restoration cannot fix

It helps to have realistic expectations before you start.

Faces under heavy damage. If a face has been torn away or covered by extensive damage, the AI will add detail that looks plausible but is not an accurate reconstruction. Several tools note openly that their face-recovery features can subtly alter a person's likeness. Treat a restored portrait as a close approximation of the original, not a definitive image of how that person looked.

Historical colour accuracy. The colours added during colourisation are educated guesses. A 1940s military uniform will be coloured based on what uniforms of that type typically looked like, not on specific knowledge of what your relative was actually wearing. For heritage purposes, a note alongside the colourised image explaining that colours are AI-generated estimates is good practice.

Severe deterioration. AI tools perform well on light to moderate damage: scratches, fading, small tears, and general yellowing. If a print has been stored in damp conditions for many years, the damage may be too extensive for AI to address reliably. UK-based professional services such as Image Restore offer archival-quality hand restoration for irreplaceable prints where an AI result is not good enough.

Film negatives. Browser-based tools are designed for prints rather than negatives. ON1 Photo RAW added a dedicated Restore AI module in April 2026 that handles older negatives more reliably, though it is a paid desktop application with a free trial.

How to save, print, and share your restored photos

Once you have a result you are happy with, download it at the highest resolution the tool offers. For most free plans this is adequate for sharing on screen or by email. For printing at A4 or larger, a higher-resolution export matters: check whether the tool offers one before committing to a free account.

UK printing services that handle photo-quality prints include Photobox, Snapfish, and Boots Photo. For anything larger than A5, a glossy finish shows the restored detail better than matte paper.

For sharing with family, a shared Google Photos or iCloud album is simple and does not require family members to create accounts. You can place the original and the restored version side by side so they can see the difference. If you are adding restored images to a genealogy project, most platforms accept standard JPEG files up to around 20 megabytes.

Frequently asked questions