Clean up small surface damage
Describe the fine lines and dust marks you want repaired. Start with a scan or an evenly lit photo of the print so new glare does not become part of the edit.
AI photo scratch remover
Remove Scratches from Old Photos
Clean up fine scratches, dust and narrow fold marks in a scanned photograph. Start with a gentle repair that keeps the original black-and-white or color appearance.
Upload your photo
JPG, PNG or WebP. Drop a photo here or use the + button.
Starting point
Target edit
The editing goal
Repair surface marks without automatically changing the photo’s character.
Start with a focused change and tell the editor which details to keep:
- Original black-and-white or color appearance
- Recognizable faces and clothing
- Composition and photographic grain
Compare fine details with your original after generating. Results vary with the source photo and instruction.
Keep the original color treatment
Repair a monochrome portrait without requesting colorization. For color prints, focus on the scratches first and handle fading separately if needed.
Retain the character of an old print
Ask to keep period grain, clothing details and the original scene. A restrained restoration can make the picture easier to enjoy without giving it a modern studio look.
Made for your photo
Repair the surface marks that distract from the memory
Fine lines on a family portrait
Thin marks surrounded by visible detail are a useful starting point. Look closely at eyes, hair and clothing after repair. If a face changes, retry with a gentler instruction rather than accepting a sharper but less familiar person.
Dust on a scanned photograph
Small specks can be confused with freckles, stars or patterned fabric. Use the dust preset and compare those details at full size. Cleaning the scan should not erase meaningful parts of the original image.
A narrow fold across clothing
A thin crease may be repaired by continuing the nearby texture. Inspect stripes, seams and the outline of the garment. Wide cracks or missing paper involve a different level of reconstruction and may not yield a dependable result.
A starting prompt for your photo
Use a prompt below to fill the editor, then adjust it for your picture. Your uploaded photos stay in place; selecting a prompt does not start generation.
Fine scratches
A restrained cleanup for thin marks; retain the original color treatment.
Gently repair fine surface scratches in this old photo. Keep undamaged detail and photographic grain. Preserve the original people, facial identity, objects, composition, photographic grain and original color or black-and-white appearance. Do not colorize, beautify faces or invent large missing details.
Dust and small spots
Focus on isolated dust spots rather than broad face enhancement.
Remove small surface dust specks and isolated scanning spots. Preserve freckles, fabric patterns and photographic grain. Preserve the original people, facial identity, objects, composition, photographic grain and original color or black-and-white appearance. Do not colorize, beautify faces or invent large missing details.
Narrow fold lines
For thin crease marks with visible detail on both sides.
Repair narrow crease and fold marks using surrounding visual context. Leave undamaged areas and original tones unchanged. Preserve the original people, facial identity, objects, composition, photographic grain and original color or black-and-white appearance. Do not colorize, beautify faces or invent large missing details.
Make it your own. Separate repair from restyling: ‘Remove the fine white scratches in the background. Keep the face, film grain and original black-and-white tones.’
How to remove Scratches from Old Photos
Step 01
Scan or photograph the print
Capture the print straight on with even lighting. Upload a clear JPG, PNG or WebP copy and keep the original scan untouched.
Step 02
Identify the surface damage
Choose a prompt for scratches, dust or narrow folds. Describe their location and ask to retain the original grain and color treatment.
Step 03
Inspect the repaired areas
Sign in to generate. Compare eyes, mouths, hands and writing near the damage before downloading a restored copy.
Pick a repair method for the damage you have
AI scratch cleanup
- Best for
- Trying repairs across several fine marks in a scanned image.
- Consider
- Check identity and texture where marks cross important details.
Manual healing or cloning
- Best for
- Repairing a few small defects with direct control over each patch.
- Consider
- Repeated marks need individual attention.
Specialist restoration
- Best for
- Assessing severe damage or a fragile physical print.
- Consider
- Requires specialist review; missing details may remain uncertain.
Get more from your original photo
Use a clear scan or a straight-on photograph of the print in soft, even light. Keep the entire photograph in frame and avoid glare from a glass frame or album sleeve. Save an untouched copy first. Start with gentle repair when a scratch crosses a face, lettering or another distinctive feature.
Keep a black-and-white photo black and white
The default instructions preserve the original color treatment. Scratch cleanup does not need to include colorization, face beautification or a modern-looking background. If you later want color, make that a separate edit so you can compare it with the repaired original.
Thin scratches differ from missing parts of a photograph
Narrow surface damage leaves visual information nearby. A wide gouge through a face or a missing corner removes more evidence. The model can fill a gap plausibly, but it cannot establish exactly what was recorded there. Keep your original alongside the edit.
Judge a repair by identity and texture, not sharpness alone
A result can look cleaner while also changing an expression, button or handwritten note. Review the repaired region and a few untouched areas. Retaining period grain and slight softness can be more faithful than making every part of the image look newly photographed.
Remove photo scratches: frequently asked questions
Can I remove scratches without colorizing the photo?
Yes. Every preset on this page asks to preserve the source’s black-and-white or color appearance. Review the result because an instruction cannot guarantee that no color or tone will shift.
Will scratch removal change faces?
It can, especially where a scratch covers an eye or mouth. Start with a gentle instruction and compare the person’s features with the original. A clearer source or another photograph may be needed for important details.
Can this remove fold lines too?
Choose Narrow fold lines for thin creases. Broad folds with missing emulsion or paper are closer to reconstruction. The quality depends on what detail remains visible around the damage.
Should I scan the photo or photograph it with a phone?
Either can provide an input. A clean scan is useful; a phone photo should be straight, well lit and free of glare. Keep the largest available source and upload a supported JPG, PNG or WebP copy.
Can it recover handwriting or a missing face?
Do not rely on generated detail as factual recovery. Missing writing or facial features may be invented from surrounding context. Preserve the source and use documented references when accuracy matters.
Do I need to sign in, and can I use starter credits?
Sign in to generate. Eligible new accounts receive starter credits, and all models are available when your balance covers the displayed cost. Nano Banana 2 is selected by default. Check the total in the editor and purchase more credits if needed.
Which files can I upload, and are my results private?
Upload JPG, PNG or WebP photos. Free generations use public mode and results may be selected for public display. Watermark-free downloads require purchase access. Review the account options on the pricing page and keep a copy of your original photo.
Make the photograph easier to see again
Upload your photo, make the instruction your own and sign in when you are ready to generate.
Edit your photoContinue with a related edit
Choose a focused tool with a ready-to-edit instruction. Upload your photo, review the settings and compare your result.