Remove scratches and damage from old photos
AI inpainting that understands the difference between a scratch across a face and one across a lawn — and fills each with texture that belongs there, not a blurred compromise.
No credit card required. Your photo is deleted after 30 days.
Scratches on photographs are a record of their physical history. The wallet photo carried for thirty years through generations of pockets. The portrait stored face-down in a wooden box. The family album with metal corner mounts that rubbed against every print for decades. The negative processed through a dirty film carrier. Each handling occasion leaves its mark, and by the time a photo is scanned, those marks are often inseparable from the memory it contains. Physical scratch removal — attempted with cotton swabs, cleaning solutions, or varnishes — almost always worsens the problem. Scratches on photographic prints are often in the emulsion layer itself, not on a surface coating, meaning any physical cleaning attempts can abrade further, spread contamination, or leave residue that photographs just as badly as the original scratch. The correct approach is to digitize first, then repair digitally. Digital inpainting for photographic scratches works by identifying the damage geometry — usually linear patterns of higher or lower density than the surrounding areas — and filling them with synthesized texture sampled from adjacent regions. The key challenge is that different subjects require different filling strategies: skin texture differs from fabric, which differs from foliage, which differs from architectural surfaces. Modern AI inpainting models apply region-specific synthesis that produces repairs indistinguishable from the surrounding image at normal viewing distances. First restoration is free.
Physical damage AI can remove from old photographs
Fine surface scratches that catch scanner light and appear as bright white streaks across the image
Deep creases where the paper folded, creating a valley that reflects as a bright line and a ridge that shadows as a dark line
Ink or ballpoint pen marks left by someone labeling the photo decades ago
Tape residue from repairs attempted in the past that left ghost rectangles in the scan
Abrasion patterns from contact with rough album pages or cardboard box interiors
Scratched negatives that print fine detail correctly but show processing damage on clear areas
Missing corners or edges where the print was damaged from improper cutting or framing
Wax crayon or grease pencil marks sometimes found on the backs of professional prints
How AI scratch removal and inpainting works
Scan without physical cleaning attempts
Place the photo directly on a clean scanner glass without rubbing, wiping, or attempting to clean the scratches. Use a lint-free cloth on the glass, not the photo. For photos in frames, remove from the frame if possible — scan face-down on the glass. Scratches often appear more dramatic under scanner illumination than they look with ambient light, which is normal.
AI identifies and classifies the damage
The model runs an initial damage detection pass that maps linear and irregular scratch patterns across the image. It classifies the damage by type (specular scratch, density scratch, mechanical abrasion) and by the subject underneath (skin, hair, clothing, background). This classification determines which inpainting strategy is applied — skin regions get a different synthesis approach than foliage or architecture.
Inpainting and quality verification
Detected damage regions are filled with synthesized texture that matches the surrounding area in grain size, tonal range, color, and spatial frequency. The repair is applied at full resolution. Download the restored image and zoom to 100% to evaluate the quality of the inpainting — for light to moderate scratches, the result should be invisible at normal print viewing distances.
Private by design
Auto-deleted in 30 days
30 seconds or less
Most photos restored instantly
First one is free
No credit card required
Scratched photo restoration — what the AI can and cannot fix
What is the difference between a surface scratch and an emulsion scratch, and why does it matter?
A surface scratch sits on the protective coating on top of the emulsion layer — it appears as a white or bright line because it scatters light differently. An emulsion scratch penetrates the image-forming layer and creates an actual loss of image data — it appears dark because there is literally less silver or dye in that location. Surface scratches are easier to repair digitally because the image data beneath is intact. Emulsion scratches require synthesis of new image data, which is harder but still often achievable for narrow damage. Both types respond to AI inpainting, but emulsion scratches through critical areas (a face, the only copy of a rare subject) may still show subtle traces at very high magnification.
There is a crease through my grandfather's face. Can AI restore it?
Faces are one of the hardest inpainting targets because humans are particularly sensitive to any inconsistency in facial geometry or skin tone. For a crease, the challenge is: the crease itself is a physical deformation of the paper that creates both a bright ridge and a dark valley in the scan. AI must fill both the bright line and the dark line with continuous skin texture that matches the surrounding face. For narrow creases (less than 3mm wide in the scan), results are usually excellent. For wide creases through complex facial features, the AI produces a good foundation that may benefit from additional manual touch-up for critical archival work.
Someone wrote a date on the front of the photo in pen. Can that be removed?
Yes, in most cases. Pen ink on photo surfaces is a masking problem — it covers whatever image data was underneath, and the AI must synthesize appropriate background texture to fill the masked region. For writing over backgrounds (sky, walls, simple surfaces), the repair is usually seamless. For writing directly over a face or important detail, the AI generates plausible texture but cannot recover the actual image data hidden beneath the ink. The result is still dramatically cleaner than the original.
Can AI restore a photo that has been taped back together?
If the torn halves are taped and the tear line is visible in the scan, yes — AI treats the tape residue and tear line as scratch damage and applies inpainting along the join. If the pieces are taped accurately (the image lines up on both sides), the result is often very clean. If the tape is holding pieces that are slightly misaligned, the misalignment itself may be visible in the restoration. Scanning the pieces separately and digitally aligning them first typically produces better results than scanning the taped-together original.
Should I try to clean scratches before scanning?
Never attempt physical scratch removal on photographic prints without professional conservation expertise. Products marketed as "photo cleaning solutions" often contain solvents that attack dye layers. Rubbing with any material — even soft cotton — can create new abrasions, spread existing contamination, or lift softened emulsion in areas that seemed undamaged. Scan the photo as-is and let AI handle it digitally. Digital repair is reversible; physical cleaning damage is not.
How much does scratch removal cost?
The first restoration is completely free with no credit card required. After that, one credit per photo regardless of damage complexity or severity —
Remove the scratches hiding your precious memories
Upload a scratched photo for free. AI inpainting in 30 seconds. No credit card required.
Restore your first photo for freeTrusted by families, historians, and photo enthusiasts worldwide