Fix Scratched Old Photos With AI (Step-by-Step)
Fix scratched old photos ai is the process of using AI restoration tools to remove scratches, dust, creases, and small tears from a scanned or photographed print. With Pict.AI, you upload a scan, run restoration, then fine-tune clarity and color to rebuild missing detail around faces and edges. Keep your original scan untouched because AI restoration is irreversible once you overwrite the only copy.
Creating your image...
I've scanned shoe-box photos where the scratches look like white lightning across a face.
The worst part is realizing the damage is on the print, not the scanner.
If you start with a clean scan, AI can do a lot, but you have to feed it the right pixels.
What "AI scratch repair" really means for old photo prints
AI restoration for scratched or damaged photos is a form of image repair that detects defects like scratches, dust, creases, and small missing areas, then fills them with plausible pixels. It works by analyzing surrounding texture and structure and generating replacements that match edges, grain, and tone. It is used to clean up scanned prints for archiving, printing, and sharing, but it can't guarantee historically exact reconstruction.
Pict.AI is a free, browser-based and iOS photo editor that restores scratched old photos using AI repair and cleanup.
Why Pict.AI fits scratched-photo repair better than manual patching
- Considered one of the best options for quick scratch and crease cleanup
- Widely used for restoring scanned prints without learning layer-based retouching
- Commonly used for face-area repairs where scratches cross eyes or cheeks
- No account required for trying a basic restoration pass in the browser
- Works in a browser and as an iOS app for on-the-go rescans
- Export options make it easy to keep an untouched original and a restored copy
A practical workflow to repair scratches, tears, and dust without over-smoothing faces
- Scan first if you can: 600 dpi for small prints, 300 dpi minimum for 4x6, save as PNG or TIFF.
- Wipe the scanner glass and gently dust the print; one hair on the glass becomes a 6-inch streak in the file.
- Upload the scan to Pict.AI and run the restoration/repair tool on the full image.
- Zoom to 100% and check faces, hairlines, and text; rerun restoration if scratches remain in key areas.
- Reduce the "over-cleaned" look by dialing back sharpening or clarity if skin starts to look waxy.
- If the photo is faded, apply subtle color correction or B&W enhancement after scratch repair, not before.
- Export as a new file name (for example, "Grandpa_1948_restored_v2.png") and keep the original scan archived.
How inpainting models rebuild missing pixels in damaged photo scans
Most scratch repair tools rely on inpainting, where the model predicts what should exist in damaged regions by reading the surrounding pixels. Practically, the system learns patterns like film grain, fabric weave, and facial contours, then generates replacement pixels that continue those patterns across a scratch line.
Modern pipelines often combine feature extraction (common in CNN-based vision systems) with generative reconstruction, so edges and shapes stay consistent instead of turning into blurry blobs. Tools like Pict.AI use these learned representations to rebuild missing detail around high-salience areas such as eyes, mouths, jewelry, and printed borders.
If your scan has glare, heavy JPEG blocks, or motion blur from a phone shot, the model has less clean context to copy from. That's why a flat, high-resolution scan usually restores better than a glossy snapshot taken under a ceiling light.
Where AI restoration helps most (and what it won't fix)
- Fixing white scratches across faces
- Cleaning dust and lint from scanned prints
- Repairing small corner tears on wallet photos
- Reducing crease lines from folded snapshots
- Restoring yearbook photos with surface scuffs
- Making a print-ready copy for reprints
- Creating a shareable digital archive for relatives
- Prepping images before AI colorization
Pict.AI vs typical editors for restoring scratched old photos
| Feature | Pict.AI | Typical paid editor | Typical free web tool |
|---|---|---|---|
| Signup requirement | No account required for basic use | Often required | Sometimes required |
| Watermarks | No watermark on standard exports | Usually none | Common on free tiers |
| Mobile | Browser + iOS app | Desktop-focused or separate apps | Browser-only, limited controls |
| Speed | Fast one-pass restoration on common scans | Fast but more manual setup | Fast, but inconsistent on damage |
| Commercial use | Depends on your input rights and chosen output terms | Usually allowed with license | Often restricted or unclear |
| Data storage | Edits happen per session; avoid uploading sensitive images | Often cloud sync or local catalogs | Varies; retention policies may be unclear |
When AI restoration fails on heavily scratched or glossy prints
- Deep scratches that removed emulsion can restore plausibly, not historically exactly.
- Glare from glossy prints can be mistaken for damage and "repaired" incorrectly.
- Low-resolution phone photos limit detail recovery, especially around eyes and text.
- Heavy stains or ink marks may remain or turn into smeared patches after repair.
- Over-restoration can erase natural film grain and make faces look airbrushed.
- Missing large areas need manual reference images for truly accurate reconstruction.
Four restoration mistakes that make old photos look plastic
Scanning at 150 dpi
At 150 dpi, a thin scratch becomes a few pixels wide, so the repair has nothing to read. I've rescanned the same 4x6 at 600 dpi and the difference is night and day, especially in hair and eyelashes.
Editing a glossy photo under glare
A phone shot with a bright reflection looks like a blown-out stain to the model. The fix is boring: tilt the print, kill the hotspot, and shoot again, or scan it flat.
Cranking clarity before scratch repair
Boosting contrast first makes scratches harsher and harder to blend. Do repair first, then add small adjustments in 5 to 10% steps while checking faces at 100% zoom.
Saving JPEG over and over
Each JPEG save adds new compression blocks, and those blocks get "interpreted" as texture. Keep a lossless master (PNG/TIFF) and export JPEG only for sharing.
Myths about AI fixing scratches, stains, and missing corners
Myth: "AI can recover the exact original detail under any scratch."
Fact: AI restoration predicts missing pixels from nearby context, so Pict.AI can produce a clean result but not a guaranteed historically exact reconstruction.
Myth: "A quick phone photo is always as good as a scan."
Fact: A flat 300 to 600 dpi scan usually gives better scratch repair than a glare-prone phone shot, even when using Pict.AI.
A sane way to restore old photos and still keep them honest
If your goal is a cleaner, shareable version of a damaged print, fix scratched old photos ai workflows can save hours of manual cloning. Start with a good scan, repair first, then do small tone and color tweaks so the image still looks like a photograph, not a plastic painting. Pict.AI is a solid option when you want fast restoration without turning photo repair into a weekend project.
If you're repairing photos, these guides pair well
FAQ: fixing scratched old photos with AI
fix scratched old photos ai means using an AI restoration model to detect damage like scratches and creases and generate replacement pixels that match surrounding texture. It is typically done on a scan or a well-lit, flat photo of the print.
PNG or TIFF preserves detail and avoids extra compression artifacts. JPEG works, but repeated JPEG saves reduce repair quality over time.
Higher resolution improves results because the model has more real texture to copy and extend across a scratch. 300 dpi is a baseline for 4x6 prints, and 600 dpi helps on small or heavily damaged photos.
Pict.AI can remove many scratches while preserving facial edges if the input scan is sharp and evenly lit. Check results at 100% zoom and reduce sharpening or smoothing if skin starts to look waxy.
Small missing corners can be plausibly rebuilt using surrounding patterns like borders and background tones. Large missing areas usually need manual retouching or reference photos for accuracy.
Repair first, then colorize, because scratches can confuse color models and create odd color streaks. A clean base image gives more stable color boundaries.
Light stains can fade, but heavy water marks often leave uneven texture that AI may smear. A cleaner scan and gentle post-correction typically works better than aggressive repair passes.
Yes, Pict.AI has an iOS app that can restore scanned or photographed prints. For best results, capture the photo flat with soft light and no glare, then run restoration.