How to Restore an Old Photo Without Changing It

Old and scanned photos degrade in a specific, recognizable way — grain from the original film, fading and yellowing from age, and scan artifacts (moiré patterns, JPEG blocking) added when the print was digitized. Restoring one means reconstructing that lost detail, not reinventing the photo.

imgmend uses Real-ESRGAN, a neural network trained to tell the difference between real image detail and degradation. It removes the noise and scan artifacts while reconstructing the underlying detail — the faces, clothing, and scene stay exactly who and what they were. It does not generate new content or alter identity, which is the main worry people have with AI photo tools.

When a face is present, the model also runs a dedicated face-restoration pass alongside the general cleanup — sharpening eyes, skin texture, and expressions that blur or fading have softened. It works within the same rule as the rest of the photo: recovering detail that's degraded but still there, never inventing features that weren't in the original.

One upload does all of this in a single pass — noise and fading removal, face restoration, and a built-in 2x resolution boost — free, with no extra steps.

Common Old Photo Problems This Fixes

• Film grain and noise — visible speckling from the original film stock, especially in prints from before the 2000s.
• Fading and low contrast — colors and blacks that have washed out over decades of light exposure.
• Scan artifacts — moiré patterns, blocking, and softness introduced when a physical print was scanned or photographed.
• Blur from the original camera — soft focus or slight motion blur common in older, slower cameras.
• Light dust and speckling — minor surface noise picked up during a scan. For a specific, visible scratch or crease, the dedicated Repair Scratch tool targets just that area for a cleaner result.

Restoration vs. Generative "AI Enhance"

Frequently Asked Questions

Related Guides