Dust spots are the small stuff. Individually they're specks, but on a scan of an old print you can count hundreds of them — most in the sky, some on skin, a stubborn cluster anywhere there was static. They're not chemistry. They're particles.

What causes dust spots on old photographs

Dust on a photograph has two origins: contaminants deposited during the print's life, and contaminants deposited at scanning time. NEDCC's surface-cleaning leaflet explicitly separates surface dust and loose particulate — which can in principle be removed mechanically — from embedded staining, which is a chemical problem that cleaning can't touch (NEDCC 7.2).

The Library of Congress photograph care leaflet names dust, dirt, and fingerprints as the top category of handling contamination, both because it's the most common kind and because fingerprints combine with dust to etch into the gelatin over time (LoC care leaflet). The Preservation Self-Assessment Program's print identification guide catalogs "accretions" — dust, lint, insect debris, adhesive flecks — as a distinct deterioration category from binder cracks and emulsion loss (PSAP photo prints).

Visually, dust spots look like: small dark or light points, roughly round, scattered without pattern but with local density (more in the corners, more where the scanner glass was dusty), and uniform in brightness regardless of what's behind them.

How AI handles dust spots — and where it can't

This is one of the easiest damage classes for a modern restoration model, because the spots are small, dense, and have very high local contrast against the background. PhotoFlip's pipeline:

• Detects specks as high-frequency point defects with round support — distinct from legitimate image detail like eyes, freckles, or stars.
• Fills each spot with context from its immediate surround, using texture-matching rather than uniform color fill.
• Preserves intentional grain. A bad filter smooths grain off along with dust; a good one removes the speckle but leaves the emulsion's native texture, which is what the pipeline is trained to do.

Honest limit: where dust happens to lie on top of a real small feature — a person's eye across a dense scan, a pinpoint star in a night shot — the model has to decide what's feature and what's contaminant. On faces this is why running face-restore as a second pass is often a good idea, since the facial prior protects eye geometry from being eaten by the despeckler.

Example restorations

• Basement-stored B&W print. Thousands of small specks across a plain sky. AI removes essentially all of them and the sky regains a smooth graded tone.
• Scan of a print with dusty scanner glass. A repeating horizontal band of specks caused by a hair on the scanner bar. The model treats these as normal speckle and removes them cleanly.
• Detailed group photo with light dusting. Specks across faces and clothing. Here you want the face-restore pass because general despeckling can't distinguish eye pupils from dust at small scales.

How to restore a photo with dust spots

1. If you're scanning yourself, clean the scanner glass first with a lens cloth. Eliminating scanner-introduced dust before it ever reaches the model is free and perfect.
2. Upload at photoflipai.com/restore. The dust pass runs automatically as part of the standard restore.
3. For portraits with dust across faces, chain the restore output through face-restore for better feature protection.

Related: dust is often accompanied by creases on handled photos, and dust scans can look like fade on low-contrast prints — see fade. Credits never expire at pricing.