Old photographs are irreplaceable. Whether it's a grandmother's wedding portrait or a faded snapshot from a family road trip decades ago, these images carry weight that no new photo can replicate. The problem is that physical prints don't age gracefully — they crack, yellow, tear, and fade until the memories they hold become almost unrecognizable. AI-powered restoration has changed what's possible here, and dramatically so. This guide walks you through the types of damage AI handles best, how to prepare your source image for optimal results, and what realistic expectations look like when you're working with a damaged print.
Types of Damage and What AI Can Fix
Not all photo damage is equal. Understanding what you're dealing with helps you set realistic expectations and get the most out of any AI restoration tool — including Fotki, which is built specifically for this kind of work on iPhone.
Scratches and Surface Abrasions
Scratches are among the most common forms of damage on old prints. They appear as thin white or dark lines cutting across the image, usually caused by improper storage, contact with rough surfaces, or simple handling over decades. AI models trained on restoration tasks recognize these linear anomalies as foreign to the natural image structure and reconstruct the underlying content with impressive accuracy — especially when the scratch passes over a relatively uniform area like a sky, wall, or fabric.
Where AI can struggle is with scratches that run across fine detail areas: eyes, hair texture, handwritten text. The reconstruction is still useful, but you may notice slight softening in those zones.
Creases and Fold Lines
Fold lines create a different kind of problem. A crease typically shows up as a bright highlight on one side and a shadow on the other, mimicking a physical ridge in the paper. AI restoration tools are quite effective at neutralizing these because the damage follows a predictable pattern — the model can identify the brightness discontinuity and blend it out using surrounding tonal information.
Deep creases that have actually removed emulsion from the print surface are harder. In those cases, the AI is genuinely inventing content based on context, which works surprisingly well for backgrounds but becomes more speculative in portrait areas.
Tears and Missing Sections
Tears introduce the most complex restoration challenge. When a corner is missing or a section of the image is physically gone, the AI must perform inpainting — generating plausible content for areas it has no reference data for. Modern models handle this remarkably well in sky, grass, and architectural regions. Portrait restoration with missing sections requires more caution: the AI will produce a coherent-looking result, but you should review whether reconstructed facial features align with other photos of the same person.
Fading and Color Shift
Fading is arguably the most universal form of photo aging. UV exposure, chemical instability in older photo papers, and humidity all conspire to drain contrast and shift colors toward yellow, magenta, or cyan casts. AI restoration excels here because color correction is fundamentally a pattern-recognition task. Fotki's restoration engine analyzes tonal distribution across the entire image and applies corrections that are contextually appropriate — restoring neutral whites in clothing without overcorrecting skin tones, for example.
Spots, Foxing, and Water Damage
Brown spots (foxing) from humidity, water stain halos, and mold damage all fall into a category the AI treats as localized noise. Small spots across open areas of an image are cleaned up cleanly. Larger stained regions — especially if they've affected contrast locally — may require the AI to reconstruct significant tonal information, and results vary based on how much of the original detail remains underneath.
Source Quality: How to Prepare Your Photo for Best Results
The single biggest factor in restoration quality isn't the AI — it's the quality of the scan or photo you feed it. A poorly captured source image limits everything downstream.
- Use good, even lighting. If you're photographing a print with your iPhone rather than scanning it, avoid direct flash. Use diffused natural light from a window, positioning the print at a slight angle to eliminate glare from the surface.
- Shoot perpendicular to the print. Even a slight angle introduces perspective distortion and uneven focus across the image. Hold the camera directly above the print, or use a document scanning app to correct geometry automatically.
- Capture at the highest resolution your device supports. More pixels mean more data for the AI to work with. For a 4x6 print, aim to fill as much of the frame as possible.
- Don't pre-edit before restoring. Avoid applying filters, adjusting exposure, or sharpening before running AI restoration. These edits can confuse the model's damage-detection logic. Feed it the raw capture and let the AI work from original data.
- Flatten the photo before photographing. If a print is curled, place it under a heavy book for a few hours before scanning. Curl creates uneven focus and shadow gradients that look like physical damage to the restoration algorithm.
- Clean the surface gently. Loose dust and debris on the print surface show up as spots in your scan. A soft lens-cleaning cloth or air blower removes surface particles without risking further damage to the emulsion.
If you have access to a flatbed scanner, use it. Scanning at 600 DPI is the standard recommendation for restoration work — it captures fine detail without producing unmanageably large files. For heavily damaged or particularly valuable prints, 1200 DPI gives you more to work with.
Practical Examples: What to Expect
Portrait with Heavy Fading
A 1940s studio portrait with severe yellow fading and loss of contrast in shadow areas is a strong candidate for AI restoration. The AI identifies the tonal range that should represent black clothing or dark hair and recalibrates accordingly. Results are typically excellent — contrast is restored, skin tones read naturally, and the overall image reads as a credible black-and-white or early color photograph rather than a deteriorated one.
Group Photo with Fold Damage
A family group photo folded in quarters for decades presents two challenges: four fold lines and potential emulsion loss at the crease intersections. AI handles the linear creases cleanly across most of the image. The intersection point in the center — where all four creases meet — requires the most interpolation. In Fotki, you'd typically see clean results across three-quarters of the image with a small softened zone at the intersection, which is still a dramatic improvement over the original.
Snapshot with Corner Tear and Spots
A 1970s color snapshot missing one corner and scattered with foxing spots is where modern AI genuinely impresses. The spots disappear. The missing corner — if it contains sky or grass — is reconstructed so seamlessly that it's difficult to identify which corner was originally damaged.
Try Fotki for Your Own Photos
Fotki is an AI photo restoration app designed for iPhone that handles all of the damage types covered here — scratches, fading, creases, tears, and spots — in a workflow that takes seconds rather than hours. There's no technical expertise required: import your scanned or photographed print, let the AI process it, and review the result. For photos that matter, it's worth seeing what's recoverable.
Download Fotki on the App Store and restore your first photo today.
Frequently Asked Questions
Can AI restoration fix a photo where faces are severely damaged or partially missing?
It depends on the extent of the damage. AI can reconstruct faces that are faded or partially obscured by scratches and spots, often with very convincing results. For areas where the emulsion is completely gone and there's no underlying information, the AI will generate a plausible face based on context — which may or may not resemble the actual person. For critical family portraits, compare the result against other photos of the same individual to verify accuracy.
Does the resolution of my original scan affect the final quality?
Yes, significantly. Higher resolution scans give the AI more pixel-level information to analyze and restore. A 300 DPI scan produces acceptable results for casual use, but 600 DPI or higher gives the algorithm more detail to work with and generally produces sharper, more accurate restorations — particularly for fine details like hair, text, and fabric texture.
Will AI restoration alter the historical look of the photo?
A good restoration tool removes damage while preserving the authentic character of the image — including appropriate grain, original color palette, and period-accurate tonal qualities. Fotki is designed to restore rather than reinterpret, so the result should look like a well-preserved original, not a digitally enhanced modern image. If you want to preserve the exact aged look but simply remove damage, review the output carefully and compare it against the original to confirm the balance feels right.