Source: https://restorebunny.com/guides/compare-duplicate-family-photos/

# How to Compare Duplicate Family Photos Before Deleting Anything

Separate exact file duplicates from alternate scans, crops and restorations. Use a four-file comparison example and a reversible review process.

Family photo projects5 min readUpdated 13 September 2026

Quick answer

## The short version

Separate identical files from different versions of the same photograph. Compare file content, pixel dimensions, crop, visible detail and metadata before choosing a preferred source. Keep unique scans, annotations and restoration history; a larger file or sharper-looking face is not automatically the best original.

## There are three different duplicate problems

An exact file duplicate contains the same bytes under another name or in another location. An image variant shows the same scene but may have different compression, metadata, edits or dimensions. A separate physical print may look similar while carrying a different reverse note, crop or printing history. Only the first category is a straightforward file-redundancy question; the others require looking at what each version contributes.

Begin with a backed-up review copy and one manageable folder. Do not let a similarity tool auto-delete across your entire archive. Two adjacent exposures can have nearly identical composition but different expressions or people looking towards the camera. A colourized version and a black-and-white original are related works, not interchangeable copies. Keep these distinctions visible in your review notes.

## Use hashes for exact duplicates, not visual similarity

A cryptographic file hash is a compact value computed from a file's contents. Microsoft's Get-FileHash documentation explains its use for comparing file contents and uses SHA-256 by default. Matching SHA-256 values are strong practical evidence of identical content. Different values mean the bytes differ; they do not tell you whether the visible image differs or which file is better.

A metadata update can change a hash without changing the displayed pixels. Conversely, two independently scanned prints will have different hashes even if they depict the same exposure. Treat an exact-duplicate report as one list and a visual-similarity report as another. Preserve source-folder information before consolidating: knowing that one copy came from a particular relative or album can matter even when the file bytes match.

## Worked example: four files, four different jobs

Imagine FAM-0140 has four candidates. File A is a 1200×800 messaging JPEG. File B is a 3600×2400 direct scan including the border. File C is a 7200×4800 AI enlargement of A. File D is a 3000×2400 alternate print scan with a tighter crop and a separately scanned handwritten reverse. These hypothetical dimensions help structure the comparison; they are not a test result from RestoreBunny or another service.

B is a sensible initial restoration candidate because it captures more of the source at higher native scan dimensions. C has more pixels but cannot be assumed to contain more authentic detail: it descends from A. D may contribute sharper detail or important provenance despite being smaller than B. A can still document the file supplied by a relative. Record the preferred input without treating every other role as disposable.

## Compare versions at the same scale

1. Make a candidate list with filename, source location, dimensions, format, known edits and any associated notes. Mark unknown history as unknown. A name containing 'original' or 'final' is a claim to investigate, not proof.
2. Align the same scene area at a comparable display size. Look for genuine fine detail in hair, fabric and lettering rather than simply judging which file has stronger edge contrast. Upscaling and sharpening can make comparisons at unequal zoom misleading.
3. Inspect the frame edges and background. One copy may retain a person, studio stamp or handwritten border that another crop excludes. Compare reverse captures and metadata as separate information channels, not afterthoughts.
4. Check faces for altered structure and text for newly plausible characters. Prefer documented captured information for historical use. A smooth, high-resolution restoration may be a useful display copy while remaining a poor replacement for the source.
5. Assign roles: preferred restoration input, alternate source, display derivative, exact redundant copy or unresolved. Keep a short reason with each decision so a later family archivist can understand it.

## Reasons to keep a second version

- It contains a different crop, reverse note, mount, inscription or exposure. Preserve the unique information and connect it to the shared scene ID.
- It has useful metadata or documented provenance absent from the preferred file. Export or record that information before considering consolidation, and verify that it remains connected.
- It is the unaltered input to a restoration you intend to keep. The ability to compare or redo the edit is a separate value from immediate visual appeal.
- Its origin is unresolved. Put it in a labelled review group and decide what evidence would settle the question; temporary uncertainty is not a reason for irreversible deletion.

## Consolidate reversibly and verify the result

For confirmed redundant copies, move candidates to a review holding folder rather than immediately emptying the trash. Keep a manifest mapping the old path to the retained file. A safe retention period depends on your backup history and whether other people still use those paths; there is no magic number of days that makes deletion risk-free. Do not remove files from a synced library without understanding whether deletion propagates.

Before final cleanup, open retained files, check the inventory links and verify a backup independent of the working folder. Search for a sample using an old filename and a photo ID. The Library of Congress's personal-archiving guidance emphasises organising material and protecting originals during review. Your success measure should be a smaller, understandable collection with no unique information lost—not the largest possible deleted-file count.

## FAQs

Is the biggest photo file always the best one?

No. It may be an upscale, a less compressed export or a different format. Compare captured detail, crop, metadata and source history, not just bytes or dimensions.

Do different file hashes mean different photographs?

No. Metadata, compression and other encoding differences can change a hash. Hashes help identify identical file contents; visual variants need separate comparison.

Should I delete a low-resolution copy after finding a better scan?

First check whether it has unique provenance, notes or a role in edit history. If it is genuinely redundant, consolidate only after preserving that context and verifying a backup.

## Sources

Preservation and technical guidance reviewed for this article.

- Microsoft: [Get-FileHash: compare file contents](<https://learn.microsoft.com/en-us/powershell/module/microsoft.powershell.utility/get-filehash?view=powershell-7.5>)
- Library of Congress: [Your personal archiving project: where do you start?](<https://blogs.loc.gov/thesignal/2016/05/how-to-begin-a-personal-archiving-project/>)

## Related

- [Retrieve an unmodified digital original](<https://restorebunny.com/guides/get-original-photo-files-before-restoration/>)
- [Name and organise scan files](<https://restorebunny.com/guides/name-and-organize-photo-scans/>)
- [Back up your family archive](<https://restorebunny.com/guides/back-up-family-photo-archive/>)

## Ready to restore your photos?

Upload an image, choose your settings, and preview results with a before/after slider.

[Get Started](<https://restorebunny.com/signup/>)[View Pricing](<https://restorebunny.com/pricing/>)

RestoreBunny Team

Photo restoration editors

Guides follow our [editorial policy](<https://restorebunny.com/editorial-policy/>) for sourcing, product claims, AI limitations, updates, and corrections.
