Fixing a blurry old photo

Soft focus, camera shake or a tiny print scanned large: what kind of blur your photo has, what can honestly be recovered, and what AI does with soft faces.

Blur is the damage people forgive at the time and regret decades later — the only photo of a great-grandfather turns out to be the one where the camera moved. Old cameras made blur easily: slow lenses, slow film, guessed focus, and prints the size of a matchbox. The modern remedy is real but has honest limits, and knowing which kind of blur you have tells you what to expect before you start.

Bring the face into focus

Choose a photo

Know your blur

Focus blur means the lens was focused elsewhere — the face is soft but evenly so, and often something else in the frame is sharp. Motion blur means the camera or the subject moved; edges smear in one direction, sometimes doubling. Resolution blur is different again: the print is tiny or the scan is coarse, and the face is only a handful of pixels doing its best. Many old photos carry two of the three at once.

Modern enhancement reconstructs plausible fine detail on top of what the blur left behind. It works best when the underlying structure survives: a softly focused portrait can come back convincingly sharp, and a small print can come back far cleaner. The honest limit is heavy blur on a small face — reconstruction there is conservative by design, because inventing a crisp stranger would be worse than leaving softness. Epoha tells the model, face by face, what may not change, and shows you the result next to the original so you can catch sharpening that has turned into replacement.

One practical note before blaming the print: a surprising share of “blurry old photos” are actually blurry scans — the phone focused on the album page, or the scanner was set to a coarse resolution. Rescan carefully before concluding the photo itself is soft; our digitizing guide covers how.

The lens, the era, and the shaky phone

A fourth kind of softness has nothing to do with what happened when the shutter opened. Cheaper lenses and coatings from earlier decades produced photographs that read soft overall, evenly, even in a shot that was perfectly focused and still — a character of the equipment, not a mistake. It shows up as an image that never reaches a hard edge anywhere in the frame, unlike focus blur, which usually leaves something else sharp. Reconstruction still helps, but a portrait shot on simple period glass will not come back looking like it was taken on a modern lens — and should not have to. The same softness turns up often in low-light wedding portraits shot without flash, on lenses already soft to begin with; the wedding photo guide covers what to expect there.

A fifth cause belongs to the digitizing step, not the original photograph: a shaky handheld phone photo of a print. That motion blur gets baked into the file before restoration ever sees it, and it is the easiest kind to fix — not with software, but by redoing the shot with a steadier hold or a flatbed scan, as the digitizing guide covers.

What the scan fixes, what the tool fixes, and where sharpening turns into invention

These are different problems. A poor scan — low resolution, a careless phone shot, a setting left for the web instead of an archive — throws away real information restoration can never get back, because the pixels were never captured. Rescanning the original print at a proper resolution, commonly 600 dots per inch, often does more for a soft image than sharpening a thin file afterward. If the print still exists, check the scan before blaming the tool.

Restoration proper recovers and sharpens detail that is genuinely present but obscured — soft focus, mild motion smear, film grain. On a well-scanned file that recovery is often substantial, and the two remedies stack: a print rescanned properly, then enhanced, goes further than either step alone. But there is a real point past which “sharper” stops meaning “more accurate” and starts meaning “invented.” A face that is a genuine blur of twenty or thirty pixels contains roughly that much information, and nothing past it is recoverable, only guessable — the honest response is conservative sharpening, not a manufactured stranger who merely resembles the blur.

Fixing a blurry old photo, in order

  1. Diagnose the blur before doing anything

    Focus blur is even and often leaves something else sharp. Motion blur smears in one direction, sometimes doubling an edge. A tiny, closely cropped face is a resolution problem, not a focus problem.

  2. Rescan before you reach for a tool

    Check whether the scan itself is the weak link — low resolution, a rushed phone shot, a web-quality setting. A proper rescan, commonly 600 dots per inch, often recovers more than sharpening a thin file.

  3. Run enhancement, then compare it to the original

    Upload the best scan you have. On a small or heavily blurred face, reconstruction stays conservative on purpose rather than inventing features.

  4. Check the face against the original, feature by feature

    Eyes, nose, mouth, the shape of the jaw — the sharpened face should still be recognizably the same person.

  5. Decide what the photo needs to do next

    A small frame forgives a soft face more than a large print does — the print-size math is worth checking before enlarging further.

Blurry photo questions

Can you sharpen a face that is just a smudge?

Sometimes, partially. If eyes, nose and mouth are distinguishable, reconstruction has anchors to work from. If the face is a featureless blob, honest enhancement will improve it modestly rather than invent a new person. Either way you see the result before paying.

Will sharpening add fake detail to skin and hair?

Reconstruction does generate fine detail — that is what sharpening a blurry photo means. The guardrails are conservatism on faces and explicit rules about what in a face may not change, so texture returns without the person changing — and you compare the result with your scan before paying.

Should I fix blur before or after other restoration?

After. Sharpening a scratched or stained image etches the damage deeper. Repair first, sharpen last — the pipeline here applies that order automatically.

Should I rescan an old photo before trying to fix the blur?

Often, yes — check this first. If the original print exists, a scan taken at low resolution or a careless phone shot throws away detail restoration cannot recover, while a proper rescan, commonly at 600 dots per inch, sometimes fixes more than any amount of sharpening applied to a thin file.

Can you sharpen a face from a tiny old snapshot enough to print it large?

Sharpening and cleaning happen together, but neither invents detail the original never captured. A small, soft face comes back cleaner, not high-resolution, and the file comes back at about 1,400 pixels on its long side — check the print-size math before ordering anything large, so the expectation matches what the file can actually support.

Bring the face into focus

Upload the blurry scan and judge the sharpened result with your own eyes.

Free during early access · 2 restorations a day

Sources