How to Restore Old Photos for Ancestry Websites

Restore photos for your family tree: Old Photo Restoration — $4.99 one-time, no subscription. Preview free before downloading.

Genealogy platforms have transformed how families document and share their history. Ancestry.com, FamilySearch, and MyHeritage now hold billions of photographs from family trees around the world, and their face recognition and matching tools connect photographs across different family branches that would never have been linked manually. AI photo restoration makes these tools significantly more effective — and it makes your family's photographic record dramatically more usable for future generations.

This guide covers the complete workflow for restoring old photographs specifically for genealogy platform use: preparation, restoration, metadata embedding, upload format requirements, and the face recognition benefits of uploading a restored rather than original image.

Why Does Photo Quality Matter for Genealogy Platforms?

Genealogy platforms are no longer passive photo albums. Ancestry's ThruLines and MyHeritage's Theory of Relativity tools use photograph matching to suggest DNA connections and tree overlaps. FamilySearch's indexing volunteers use photograph quality to transcribe names and dates from documents and portraits. These algorithmic tools fail when photograph quality is too low for their underlying detection models.

The face recognition system that powers MyHeritage's Smart Matches and Ancestry's photo discoveries requires detectable faces to function. A blurry, faded, or scratched portrait that a human viewer recognizes immediately as a face may register zero detected faces in an automated system — making it invisible to matching algorithms and generating no family connection suggestions.

AI restoration changes this. Running a damaged portrait through ArtImageHub's Old Photo Restoration — which uses GFPGAN for face detail reconstruction and Real-ESRGAN for overall upscaling — produces an image that face detection algorithms can process reliably. The same photograph that generated zero face matches as a damaged original may generate multiple high-confidence matches after restoration.

What Are the File Requirements for Each Major Platform?

Each genealogy platform has different technical specifications. Uploading the wrong format or an oversized file wastes effort.

Ancestry.com:

• Accepted formats: JPEG, PNG, GIF, BMP, TIFF
• Maximum file size: 25MB per image
• Recommended: JPEG at 90-95% quality
• Display resolution: resized to approximately 1200-1600px wide for viewer
• Storage: full-resolution original preserved

FamilySearch:

• Accepted formats: JPEG, PNG, GIF, TIFF, PDF (for document scans)
• Maximum file size: 15MB for photos, 30MB for documents
• Recommended: JPEG for restored photographs, TIFF for document preservation
• FamilySearch actively encourages highest-resolution uploads for preservation

MyHeritage:

• Accepted formats: JPEG, PNG, GIF, BMP
• Maximum file size: 10MB per image (standard accounts)
• Note: MyHeritage does not accept TIFF directly — convert to high-quality JPEG first
• Face recognition runs on all uploaded photos, regardless of privacy setting

Findmypast:

• Accepted formats: JPEG, PNG
• Maximum file size: 10MB
• Primarily used for document images, not portraits

For all platforms, upload the full-resolution restored output from ArtImageHub rather than a downsized version. Platforms resize for display but preserve originals — giving collaborators access to full-resolution downloads.

How Should You Embed Metadata Before Uploading?

Metadata embedded in the image file creates a permanent, platform-independent record of what the photograph documents. When you upload to a genealogy platform, that embedded data pre-fills caption and date fields, saving re-entry time and ensuring the information survives if the photograph is downloaded and uploaded elsewhere.

Key EXIF fields to complete before uploading:

• DateTimeOriginal: The date the original photograph was taken. Use YYYY:MM:DD format. If only the year is known, use 1945:00:00. If approximate, use the midpoint of the estimated range.
• ImageDescription: Free-text caption including full names of all identified individuals, their relationship (e.g., "maternal grandmother"), location, and occasion. This is the field most platforms display as the photo caption.
• Artist: Name of the photograph's owner or submitting family member — useful for tracking source when the photo circulates across trees.
• UserComment: Additional notes that do not fit in ImageDescription — photographer name if known, negative number, source archive.
• GPSLatitude/GPSLongitude: Approximate coordinates of where the photograph was taken. A town center is sufficient — precision to the meter is unnecessary and potentially a privacy concern for living individuals' addresses.

Software for metadata embedding:

• ExifTool (free, cross-platform command-line): the most capable option, supports batch processing
• Adobe Bridge (free with Creative Cloud membership): graphical interface for ExifTool-compatible fields
• DigiKam (free, open source): full-featured photo manager with EXIF editing

Embed metadata after restoration but before upload. Some platforms strip metadata on upload; embedding ensures the information exists in your local archive copy regardless.

What Is the Right Restoration Workflow for Genealogy Uploads?

The optimal workflow sequences tools in a specific order that maximizes the value of each processing step.

Step 1: Scan correctly. Scan prints at 600 DPI minimum; small prints (wallet size, cabinet cards) at 1200 DPI; negatives at 4800 DPI. Save as TIFF.

Step 2: Remove compression artifacts if needed. If your source is a JPEG (downloaded from another platform, scanned to JPEG, or photographed with a smartphone), run it through a denoising pass first to remove JPEG blocking artifacts before restoration.

Step 3: Restore. Upload to Old Photo Restoration. The tool applies Real-ESRGAN upscaling and NAFNet denoising automatically. Preview the result for free, then download the full-resolution output for $4.99 (one-time, covers all ArtImageHub tools).

Step 4: Face enhancement if needed. For portraits where face detail is the primary concern — an ancestor portrait you want to identify or match — run through Photo Enhancer, which applies an additional GFPGAN face restoration pass optimized for clarity.

Step 5: Embed metadata. Add EXIF fields using ExifTool or DigiKam before upload.

Step 6: Upload. Upload the metadata-embedded, full-resolution JPEG to your genealogy platform(s).

How Does Restoration Affect Face Recognition Matching Results?

The improvement in face recognition match rates from AI restoration is not subtle. Face detection algorithms require a minimum number of detectable facial landmarks — typically 68 specific points on the face — to generate a usable face descriptor for matching. Degraded photographs frequently fall below this threshold.

GFPGAN, the face restoration model in ArtImageHub's pipeline, was specifically designed to reconstruct the fine facial features — eye corners, nose tip, lip edges, chin contour — that landmark detection algorithms rely on. By reconstructing these features from damaged or low-resolution source images, GFPGAN effectively converts a photograph that was below the detection threshold into one that generates complete landmark sets.

In practical genealogy use, the impact appears most clearly in:

• Group portraits: Individual faces in a group photo are small relative to the frame. Upscaling and face restoration make each face large enough for reliable detection.
• Formal studio portraits: These benefit most from GFPGAN specifically because the faces are the primary subject and damage is visible in the exact regions the algorithm needs.
• Document portraits: Immigration papers, military ID photos, and school yearbook photographs are small by design. Upscaling these before upload dramatically improves match rates.

What Are the Privacy Considerations for Genealogy Photo Uploads?

Old photograph restoration creates a new consideration that does not apply to uploading unrestored originals: restored photographs are significantly clearer and more identifiable than their damaged predecessors.

A group photograph that was too blurry to identify background individuals may, after AI restoration, clearly show faces that can be identified. Before uploading a restored group photograph to a public tree, consider whether any of those newly visible background individuals might be living — in which case, public posting without consent may not be appropriate.

For photographs of deceased individuals only (which covers the vast majority of old photograph restoration use cases), public tree settings are appropriate on all major platforms. The genealogical research exception in most privacy frameworks covers research, archival, and family documentation use of historical images.

Platform-specific note: MyHeritage's face recognition runs on all uploaded photographs regardless of the privacy setting of the containing tree section. If you want to prevent face-matching on a specific photograph for privacy reasons, do not upload it to MyHeritage at all.

The standard advice applies: when in doubt about a living individual's privacy preferences, keep the photograph in a private tree section and ask before making it public.

Your ancestors left you these photographs as a record of their lives. Restoring and properly documenting them at artimagehub.com before adding them to your family tree is the highest-quality contribution you can make to the genealogical record.