Skip to main content
Automatic image restoration & repair

Upload up to at the same time, Pinin will restore them automatically. Each photo restoration takes approximately 47 seconds and costs only one credit.

Try Pinin for free

Experience our AI-powered image restoration with your first photo completely free. No payment required.

Login required

Please log in to use our automatic image restoration service.

Add up to 10 images

Recent photo restorations

Pro tips

A few quick habits before you upload make a big difference in how sharp, clean and lifelike your restored photos turn out.

Scan it flat

Lay photos completely flat on the glass — no wrinkles, folds or curled corners.

Clean the surface

Wipe your scanner glass or camera lens so no dust or smudges get baked in.

Go high resolution

Scan at 600 DPI or higher so the AI has real detail to work with.

Even lighting

Use bright, consistent light and avoid harsh shadows, glare or reflections.

Remove dust first

Gently brush the print with a soft cloth to clear loose particles before scanning.

Keep it lossless

Save as PNG or TIFF when you can to preserve the maximum amount of detail.

Repair by damage type

Different damage, different physics

Every kind of damage leaves a different trace in the emulsion, and what can be recovered depends on which one you have. Torn prints need inpainting across a gap. Water damage leaves a tide line but usually keeps the image underneath. Mould etches. Fading is dye chemistry, not dirt. Pick the one that matches your photograph and the page will tell you what comes back.

Torn and ripped prints

Handling cracks a brittle emulsion along a fold; old tape stains the paper it was meant to hold.

The gap is filled by inpainting — plausible texture, consistent with the surroundings, not the original pixels.

Water and flood damage

Water swells the gelatin binder and carries dissolved dye outward, which is what draws the tide line.

Staining and the tide line clean up well. Emulsion that lifted off the paper has nothing left to recover.

Mould and mildew damage

Fungal hyphae digest the gelatin layer, leaving foxing spots and a haze that reads as softness.

Surface haze and spotting respond. Where the fungus has etched into the emulsion the texture is gone.

Faded vintage prints

Chromogenic dyes fade unevenly — cyan first, which is why old colour prints drift magenta.

Colour balance and contrast can be rebuilt. The densities that faded are not stored anywhere.

Ancestry and family archive photos

Decades in an album: adhesive yellowing, curled corners, and a dozen small damages on one sheet.

Usually all of it at once. The constraint is provenance, not pixels — keep the scan with its notes.

Pixelated and low-resolution photos

Resampling below the detail the file held, or a screen grab of a screen grab.

Edges are inferred, not returned. Fine at print size, unusable for identification.

Blurry and soft photos

Camera shake, a slow shutter, or a scan from a small print: the detail was captured but smeared across neighbouring pixels.

Sharpness is reconstructed from the image itself. Mild softness comes back convincingly; extreme blur is partly guessed.

Wedding photographs

A dark church, a long dress, and a photographer working at the edge of the available light — usually on the only surviving print.

The light problem is named rather than glossed over, and the standard for what counts as finished is higher.

Two honest limits apply to everything above. Large missing areas are filled with texture the model invents, so a face that was torn away is a reconstruction and should be treated as one. And where the print is the only surviving copy, the original should be scanned once at archival quality and left alone — the restored file is a derivative, not a replacement. See before-and-after restorations by damage type before uploading anything.

What restoration does, and what it refuses to do

Restoration here means five operations, not one: an all-in-one repair that regenerates the frame, a dust-and-scratch pass that spots and inpaints, a deblur-and-upscale pass for soft prints, a colourisation pass, and an upscale-only pass for files that are clean but small. You pick one per photograph. Each render costs one credit, the first is free, and the original file is never modified — you get a new file beside it.

The repair is inpainting, and inpainting is a statistical argument. Given the pixels around a tear, the model proposes what the emulsion probably held: continuous texture, plausible tone, an edge that carries the same direction it carried on both sides of the gap. That is genuinely useful across a scratch, a crease or a chipped corner. It is also a proposal, not a recovery, and the difference is visible the moment anyone asks what was under the tear.

Which is why the honest limit on this page is about the original: if the print, the negative or the slide still exists, scan it again rather than repairing a copy of a copy. A second capture produces information; no operation here does. Fading is the same story in chemical form — a print that has lost its cyan dye layer has lost it, and the yellow cast can be neutralised but the colour that was there cannot be computed back.

Before and after are worth looking at as a pair rather than as a gallery. The restored half tells you what the model inferred; the original tells you how much it had to infer, and a result that looks astonishing beside a badly damaged print is often a small change beside a good one. See before-and-after restorations from each damage type, including the ones that came back worse than they went in.

Some photographs deserve their own process rather than the general one. Wedding photographs are the clearest case: usually the only print, frequently handled, often stored in an album with adhesive that has since stained the reverse, and a subject nobody is willing to accept a plausible guess about. Those run on the same model but with a different standard for what counts as finished.

Which tool can actually do this has its own page per assistant, and each of them says what the tool returns before you spend anything on it: ChatGPT and Gemini hand you a re-drawn image, Claude cannot return an image at all, Meta AI works at feed size, and Remini sharpens a face by reconstructing it. If you would rather do the work by hand, Photoshop remains the most capable tool there is — the honest comparison is on AI restoration vs Photoshop. What free photo restoration means when the first photo genuinely costs nothing is the last of them.

Photo restoration — frequently asked questions

Everything you need to know about restoring your old and damaged photos with Pinin.

How do I restore an old photo with Pinin?

Upload a scan or photo of your old print, choose a restoration mode, and Pinin's AI repairs damage, sharpens detail, and can colorize it in seconds. Your first restoration is free — no credit card required to try it.

Is the first photo restoration really free?

Yes. Your first photo restoration is completely free, with no watermark and no credit card required. After that, one credit restores one photo.

What kinds of damage can AI photo restoration fix?

Pinin repairs scratches, tears, creases, fading, discoloration, water and mold stains, blur, and low resolution. It can also colorize black-and-white photos and upscale small or grainy scans to a sharp, print-ready result.

What image formats and sizes are supported?

You can upload JPEG, PNG, and most common image formats. For best results, scan prints at 600 DPI or higher and keep the file lossless so the AI has real detail to reconstruct.

Are my uploaded photos kept private and secure?

Yes. Every upload is encrypted in transit and during processing, your original is never altered, and images are not shared with third parties. See our Privacy Policy for full details.

What if my photo is too damaged for automatic restoration?

For severely damaged or antique photos where large areas are missing, we offer a manual restoration service where our expert artists rebuild the image by hand.