Photo restoration: How to repair an old photo with AI

A faded family portrait or a torn print does not have to stay hidden in an album. Start with Deep-Image Photo Restoration to repair an old photo online, then use selective retouching for any damage that needs more attention.

Updated 23 October 2026: a simpler restoration workflow

There is now a dedicated tool for old and damaged photographs. Try an automatic restoration first. The comparison below comes from our earlier Auto Enhance update; the manual case studies later in this guide show a separate workflow and the interface used at the time.

How to restore an old photo online

1. Digitize the print. Scan it or photograph it straight on, in even light. Avoid reflections and save a separate, unedited copy.
2. Open Photo Restoration and upload your image. Start with Natural Photo Correction, the preset available without an account.
3. Restore and compare. Inspect faces, clothing and repaired edges. Try another preset if you have access to it and compare the versions.
4. Download the right file. Guests can restore up to three photos a day and download a JPG Preview with a 500 px long edge. Full-resolution downloads require an account. Free-account results have a watermark; paid plans offer watermark-free downloads.

Which restoration preset should you choose?

For a family portrait where likeness matters, begin with Natural Photo Correction or, with an account, Auto Enhance Tone Correction. For extensive damage, compare a stronger preset such as Auto Enhance Pro or Cinematic feel, both available with paid access. Preset choice also affects color: check whether a monochrome photo stays close to the original or gains new tones.

AI reconstruction is an interpretation. A repaired eye, nose or missing part of a face may look plausible without matching the person exactly. Added colors are not evidence of the original colors. Keep the original scan alongside any restored version, especially for family history or archival use.

When does manual retouching help?

If an automatic restoration leaves one damaged patch, Inpainting lets you select that area for another edit. Use a small selection and compare the result after each pass. The two case studies below document the earlier manual approach, including screenshots of the interface used at the time. They illustrate selective repair rather than the steps required by today’s Photo Restoration tool.

Step by step guide through a challenging photo restoration

These two earlier examples show how selective edits can repair a badly damaged print. The aim is to reduce visible damage while keeping the photograph’s character. Start with the current automatic tool above; use these techniques when a particular area still needs work.

As we can see, the photo is very badly damaged. There is a huge chunk of photo completely missing. The face on the right has melted nose and burned marks on the chin. It all looks very bad.

Our first step will be with the Inpainting tool, and we will try to get rid of this big white chunk of missing photo.

I have used the brush, to select this teared off area. And in the text prompt section I simply wrote "clear background and a bit of a female head". But probably just "clear background" would be enough.

And this is my result. It doesn't look like much just yet, but this is a step-by-step process so we have still some work ahead of us. Now, let's try to fix that nose and chin.

I have reuploaded this photo to the Inpainting tool again. This time I have selected only the nose and in the text prompt wrote "female nose".

As you can see on the image above, I have already fixed the nose, now I am doing exactly the same process with the chin. I have selected it using an Inpainting brush. Then in the text prompt wrote "female chin"

Now we have fixed the most crucial damage on the photo. What we need now is a little bit of a general fix applied to the whole photo. Great for this purpose will be a generative Upscale. For this, we will use AI Image Enhancer.

Here we have used the Generative Upscaling. But it's a little trick, as we set the width for 1200 pixels, which is exactly the same value as an input photo, we will not really upscale the photo, but the generative upscaling feature will still apply to the image. For better effect, we also turned on "reduce noise" and "sharpen" options.

Of course, we could still continue to work on that photo and get it in a much better shape. This example is to show you how many options you have available, when you are fully aware of the possibilities that specific tools offer. Once you realise you can combine various tools and features, you will gain a lot more creative options, that could be used for any purpose.

Portrait photo restoration example

Second example is of a vertical portrait photo, that we will try to repair.

The main issues with this photo are two white lines due to folding the photo and also terrible noise that kills most of the details in the photo. Our first step will be getting rid of those fold lines. Just like previously, for this, we will use our Inpainting tool.

For the second portrait, I selected the fold lines and used the prompt "remove scratches". If one selection changes too much, work on the background and shirt separately. Review each result before moving on; the number of edits and their cost depend on the tools and settings you use.

With fold lines removed, this is not yet looking that great. But we will now apply some effects on a whole photo to fix it all up. For this we will switch to AI Enhancer PRO, which has all features in one place and together with AI Generator PRO are the most comfortable tools for advanced users.

In here we apply Generative Upscaling together with noise reduction and sharpening. While using generative upscaling we don't really have to actually upscale the image, we can keep the same size, but still apply the effect. This time we did upscale image a bit, as it was really small to begin with.

And here we have the final result. I personally would still do some work on it to add this final touch. But it's a good example on how you can work on your own old photos and achieve some satisfying results with relatively small time and money investment.

List of tools used for Photo Restoration

For a new restoration, start with Photo Restoration. The older case studies above used these additional tools:

• Inpainting Image AI - to redraw the most damaged and completely destroyed elements of the picture
• AI Photo Enhancer - to apply global effects on a whole photo
• Generative Upscale - to help restore completely lost details.