Bring damaged family photographs back on your iPhone, iPad or Mac - scratches, tears, faded faces and missing colour - with everything processed on the device and the original never touched.
Five kinds of damage, each handled by its own model: scratches and cracks, stains and tears and worn corners, faded or blurry faces, missing colour in black and white photographs, and low resolution. Auto Restore runs the ones that apply; you can also apply them individually.
No. Everything runs on your iPhone, iPad or Mac. There is no account, no sign-in and no upload, and the app works with no network connection at all. Nothing about what you restored is sent anywhere.
Never. Every restoration is saved as a new file and the original is left exactly as it was. This is deliberate - restoration involves judgement, and you should always be able to go back to what the photograph actually looked like.
A flatbed scanner at 600 dpi gives the best input by a wide margin. If you are photographing instead, use diffuse light from the side rather than a flash, keep the print flat, and fill the frame with the photograph. The models amplify whatever is in the source, so a clean scan matters more than any setting in the app.
It can rebuild small missing areas from the surrounding texture - a crease through a cheek, a worn corner, a chunk of background. Where a face is largely gone, there is nothing to recover from, and the app will not invent a new one. That limitation is intentional.
Up to 16. Each face is detected and restored individually rather than the whole image being sharpened, which is why a group photograph comes out with everyone in focus instead of a uniformly sharpened blur.
That is the point of restoring faces separately. The model recovers the features present in the photograph rather than generating a plausible face. If a face is too degraded for that, the result stays soft - which is the honest outcome, and better than a confident stranger.
It understands skin, sky, fabric and foliage, so results are plausible rather than tinted. It cannot know that a dress was green - no colouriser can, because the information is not in a black and white photograph. Treat colour as an interpretation and keep the monochrome original.
Yes. Colour photographs from the 1960s to 1980s commonly shift towards red or yellow as the dyes age, and the colour pass corrects that rather than inventing colour from nothing.
After. Upscaling first enlarges the scratches into features, and the repair models then work around damage that has been made structural. Restore, check the result, then enlarge if you need the resolution for printing.
Reduce the strength and run it again. Over-restoration shows as waxy skin and lost grain, and it is the most common complaint about any restoration tool. Comparing against the original at full size is the way to catch it - a result that looks impressive on its own often looks wrong beside the photograph.
Yes, batch processing handles a folder of scans. It is worth checking the first few individually to settle the strength before running the rest, because a setting that suits one photograph usually suits the whole scanning session.
JPEG, PNG and HEIC. For an archive, keep the original scan and the restoration side by side - the restoration is a copy, not a replacement, and in twenty years you may want to redo it with better models.
iOS 17 or iPadOS 17 on iPhone and iPad, macOS 14 or later on Mac, and visionOS 1.0 or later on Apple Vision Pro. It runs natively on Apple Silicon, and one purchase covers every platform.
Repair, sharpen and colourise old photographs on device - nothing is uploaded and originals are never modified.
From a print in a shoebox to a finished restoration, in about five minutes.
Do the comparison before the colour step. Judging restoration and colourisation at the same time makes it hard to tell which one went too far.
Try this in Old Photo Restorer: AI Repair →The scanning is the slow part, so the aim is to scan once and decide later.
Photographs from the same era and the same camera usually respond to the same settings, so grouping the scans by decade before you start saves a pass.
Try this in Old Photo Restorer: AI Repair →Our support team usually replies within one business day.
Download Old Photo Restorer: AI Repair Contact support