How to Fix Creased and Folded Photos

Why Creases Are So Common

Creases are the most frequent damage type in old family photos because:

• Photos were folded to fit wallets, letters, small frames
• Piles of photos in drawers create pressure creases
• Prints curl as they age, causing uneven creases
• Old album adhesives pull prints into specific shapes
• Kids folded photos for craft projects (yes, really — happens often)

The good news: creases are one of the easiest damage types for AI to handle. A visible crease line is treated the same way as a scratch and reconstructed convincingly in most cases.

Step 1: Flatten the Physical Photo First

Before capturing:

For simple creases

Place the photo between two sheets of clean paper under a heavy book for 24-48 hours. The weight helps flatten the crease line somewhat, making capture easier.

For severe folds

More aggressive flattening can sometimes work:

• Place between two paper towels
• Mist paper towels very lightly with distilled water (the paper towels, not the photo)
• Press under heavy books for 48 hours

Caution: moisture risks damaging the photo. Only try this for photos that are already severely creased and have no other damage to worry about.

Don't try to "iron" creases

Heat damages emulsion permanently. Ironing photos will destroy them.

Step 2: Capture the Photo

Follow the iPhone digitizing guide. For creased photos specifically:

• Capture after flattening attempts
• Use indirect daylight to avoid harsh shadow along crease lines
• HDR helps even out tonal differences along the crease
• Multiple captures at different angles if the crease creates uneven lighting

Step 3: Apply Restory Workflow

Open Restory.

Standard workflow for creased photos

1. Remove Scratches (5 coins) — treats the crease as a linear damage pattern and fills it
2. Restore Faces (5 coins) — rebuilds any facial detail disrupted by the crease
3. Enhance Details (4 coins) — recovers sharpness along the former crease line

Total: 14 coins, about EUR 1.75.

For severe folds

Re-run Remove Scratches a second time if the first pass leaves visible crease remnants. The stochastic AI produces different results on each attempt; occasionally the second attempt catches what the first missed.

Step 4: Specific Crease Types

Straight creases (most common)

Photo was folded for storage. The crease is a sharp straight line. Remove Scratches handles these excellently.

Curved creases

Photo was stored rolled or curved. The "crease" is actually multiple parallel bends. Usually restores well with standard workflow.

Intersecting creases

Multiple folds creating intersection points. The AI handles linear damage but intersection points can be more challenging. Results are usually good but may show slight imperfections at the crossing.

Creases through faces

The crease runs directly across someone's face. Restore Faces handles this case specifically — facial structural priors rebuild the disrupted area convincingly.

Creases with emulsion damage

The fold has cracked the emulsion, creating a raised line or missing paint. Remove Scratches typically handles surface cracks; for missing emulsion, you may need Recreate (6 coins) for generative fill.

Step 5: Verify the Result

After processing:

• The crease should be invisible at normal viewing distance
• Zoom in 100% to see any residual imperfection
• Faces affected by the crease should look natural and symmetrical
• Background along the crease should match surrounding content

If a faint line remains, re-run Remove Scratches. For most family photos, a single pass fully eliminates visible creases.

A Realistic Example

Consider a typical case: a 1963 family wedding photo that was stored folded in a wallet for decades. The photo has:

• A sharp crease line across the bride and groom's faces
• Slight yellowing
• Some additional surface scratches

Workflow:

1. Flatten under books for 48 hours
2. Capture with iPhone (1 minute)
3. Remove Scratches (5 coins) — handles crease and additional scratches
4. Restore Faces (5 coins) — sharpens the bride and groom
5. Enhance Details (4 coins) — corrects yellowing, improves overall

Total: 14 coins, ~EUR 1.75. Time: about 5 minutes plus flattening.

Result: a clean restored wedding photo suitable for framing as an anniversary gift.

Cost Comparison

Approach	Cost per creased photo	Time
Restory AI	EUR 1.75	5 minutes
Manual Photoshop work	Software + 1-2 hours	1-2 hours
Professional restoration	EUR 75-250	1-3 weeks

For typical family creased photos, AI restoration is the right balance of cost, time, and quality.

For broader context, see our Restory vs Remini comparison.

Related Guides

• How to fix a torn photo
• How to restore a torn wedding photo
• The ultimate guide to photo restoration

Frequently Asked Questions

Will a restored crease look obvious if I zoom in really close?

Usually no, with modern AI. The Remove Scratches feature uses surrounding texture to fill the former crease line, producing results that are generally invisible even at 100% zoom. Exceptions: creases that intersect with complex patterns (detailed backgrounds, fabric textures) may show slight imperfections on close inspection. For normal viewing and printing, AI-restored creases are indistinguishable from undamaged photos.

Should I physically flatten the photo before scanning or let the AI handle the crease?

Flatten first when possible. Physical flattening reduces the shadow line created by the crease, giving the AI better source data to work with. A photo with a sharp visible shadow from a crease is harder to restore cleanly than one where the fold has been compressed. 24-48 hours under a heavy book is usually enough for most creases.

Can AI restore a photo that's been folded into quarters multiple times?

Yes, though results vary with severity. Simple fold patterns (folded once or twice) restore to completely invisible results. More complex fold patterns (folded into quarters with multiple intersection points) may show slight imperfections at intersection points but are still dramatically improved. For severely folded photos, run the restoration twice and accept that the result is excellent even if not absolutely perfect.