How to fix blurry old photos
Updated 13 October 2026 · UnBlur AI guides · 8 sources
AI results are estimates. AI restoration predicts what a sharper or repaired photo might look like. It can invent or change details, especially faces, text and fine patterns. Always keep your original, and do not use an AI-restored image as evidence or to identify someone.
- Many old photos were never sharp: slow film, simple lenses and movement all softened them at the time they were taken.
- A good scan of the best original fixes more blur than any software. Every generation of copying loses some quality. 1,2
- Gentle sharpening increases edge contrast; it cannot add detail that is not in the photo. 3
- AI deblurring predicts new detail, which can change faces. Compare results carefully with the original. 4,5
To fix a blurry old photo, first make sure the blur is really in the photo and not in your copy of it. Then sharpen gently, and only after that consider an AI tool, comparing its result closely with the original. The order matters because a sharper scan of the best original you can find often helps more than any software.
Where the blur comes from
| Cause | What it looks like | Can it be improved? |
|---|---|---|
| Missed focus when the photo was taken | Everything soft, no direction | Somewhat, by sharpening or AI prediction |
| Camera or subject movement | Smears in one direction | Hard; AI may help with mild cases |
| A copy of a copy | Soft, low contrast, sometimes a print pattern | Yes: find the original print or negative |
| Low-resolution or shaky scan | Blocky or soft at full size | Yes: rescan |
| Dirty scanner glass or phone lens | Hazy, low contrast | Yes: clean and rescan |
The guide to motion blur vs focus blur shows how to tell the first two apart.
Step 1: Find the best original
The federal digitization guidelines point out that every generation of photographic copying loses some quality, so scanning a duplicate or copy print means starting with less 1. Ask relatives whether anyone has the original print, a larger print, or the negatives. A negative usually holds more detail than a small print made from it.
Step 2: Scan it properly
- Clean the scanner glass with a lint-free cloth dampened with water, spraying the cloth rather than the scanner 2.
- Scan snapshots at 300 to 400 dpi at least, and very small prints at much higher settings 2. Archival guidelines use 600 ppi for top-quality scans of prints 1.
- Turn off automatic sharpening and dust removal in the scanner software for the master file 1.
- Save the master as a TIFF 2.
Scanning far beyond what the print holds will not make it sharper; the Library of Congress notes there is a limit past which more dpi adds nothing 2. See how to scan old photos properly for the full method, or scanning with your iPhone if you have no scanner.
Step 3: Sharpen gently
Classic sharpening, often called unsharp mask, boosts the contrast along edges. Free desktop editors such as GIMP offer it with three controls: radius (how many pixels around each edge are affected), amount (strength) and threshold (the minimum difference that counts as an edge, which protects smooth areas such as skin and sky) 3. The same documentation notes that strong settings create dark or light halos along edges 3.
On iPhone, open the photo in Photos, tap Edit, then Adjust, and swipe through the light and color settings 6; on current versions the list also includes sharpening and definition controls like the ones Apple documents for Photos on Mac 7,8. Small amounts look natural. If the result shows halos or crunchy grain, back off. The difference between these tools and AI is explained in sharpening vs AI deblurring.
Step 4: Consider an AI tool, carefully
AI deblurring does not undo the blur mathematically. It predicts a plausible sharp image, and research shows that pushing restored images to look more realistic moves them further from a strict reconstruction of the input 4. Face restoration models rely on what faces generally look like, which helps realism but means fine features may be generated rather than recovered 5. In family photos that matters: a great-grandmother's eyes or smile can come out subtly different.
UnBlur offers two modes that fit old photos. Unblur is meant for softness and blur; Restore Old Photo also targets scratches and fading, so it is usually the better pick for a damaged print. Either way, you choose the scanned image from your library and get a side-by-side ORIGINAL and ENHANCED comparison before saving. Severe blur may not be fixable, and the result is an estimate. See how to fix a blurry face in a photo for how to check faces, and which UnBlur mode to use.
Checking the result
- Compare eyes, teeth, ears and hairlines with the original at the same size.
- Check text on signs, uniforms and badges: AI often invents letters.
- Show the result to someone who knew the person; they will notice changes you miss.
- Label AI-sharpened copies when you share them, and keep the original scan.
Common questions
Why does my old photo look blurry even after a high-resolution scan?
Because the blur is in the print itself. Scanning captures what is there; past a certain resolution, more dpi only adds file size and dust. 2
Is sharpening or AI better for old photos?
Sharpening is safer because it only boosts existing edges, but it cannot add detail. AI can look much sharper but may change features, so compare it with the original. 3,4
Should I use an AI-sharpened photo for genealogy records?
Keep the original scan as the record. If you share an AI version, label it as AI-enhanced so others know details may have been predicted.
Sources and further reading
- Federal Agencies Digital Guidelines Initiative. Technical Guidelines for Digitizing Cultural Heritage Materials, 3rd edition (2023), section 3.8 Prints and Photographs (PDF). digitizationguidelines.gov
- Library of Congress, The Signal blog. Personal Digital Archiving: The Basics of Scanning (2014). blogs.loc.gov
- GIMP 2.10 documentation. Sharpen (Unsharp Mask). docs.gimp.org
- Blau Y, Michaeli T. The perception-distortion tradeoff. IEEE CVPR. 2018. arxiv.org
- Wang X, Li Y, Zhang H, Shan Y. Towards real-world blind face restoration with generative facial prior. IEEE CVPR. 2021. arxiv.org
- Apple Support. Edit photos and videos on iPhone. support.apple.com
- Apple Support. Sharpen a photo or video on Mac (Photos User Guide). support.apple.com
- Apple Support. Adjust definition in a photo or video on Mac (Photos User Guide). support.apple.com