How to Find and Delete Duplicate Photos (Free Tool)

Why duplicate photos accumulate faster than you think

Duplicates do not just come from consciously copying files. They accumulate through a dozen invisible channels. Every time a photo syncs from your phone to iCloud and then to your Mac, you may end up with two or three copies in different directories. Backup software creates archives that overlap with live libraries. Messaging apps save received photos to your camera roll, creating copies of images you already have from other sources.

Professional photographers deal with a different but equally common problem: burst shots. Press the shutter in burst mode and you might have 15 nearly identical frames of the same moment. Only one or two of those will be keepers — the rest are storage waste.

The result is photo libraries where 20–40% of the storage space is occupied by redundant images. For a 100GB library, that could be 20–40GB of recoverable space — and dozens of hours of wasted time scrolling through near-identical photos.

How duplicate photo detection works: exact vs near duplicates

There are two fundamentally different types of duplicates, and they require different detection techniques.

Exact duplicates: cryptographic hashing

Exact duplicates are files where every byte is identical. Even if the filenames are different (photo.jpg vs photo-copy.jpg vs IMG_4721.jpg), the underlying image data is the same. Detecting these is straightforward with cryptographic hashing.

A cryptographic hash function (like MD5 or SHA-256) takes any file as input and produces a short fixed-length output called a hash or digest. The same file always produces the same hash. Two different files with even a single changed byte produce entirely different hashes. If two files share the same hash, they are byte-for-byte identical — guaranteed.

This approach is fast and certain, but it only catches true exact duplicates. A photo that has been re-compressed, resized, cropped, or had its EXIF metadata modified will not match even though it looks visually identical. That is where perceptual hashing comes in.

Near duplicates: perceptual hashing

Perceptual hashing is one of the most elegant algorithms in computer vision. Instead of hashing the raw file bytes, it hashes the visual content of the image in a way that is tolerant of minor variations. Two images that look the same to the human eye will produce very similar perceptual hashes — even if one has been resized, lightly edited, or saved at a different compression level.

The most widely used algorithms are:

• dHash (Difference Hash): Detects differences in adjacent pixel brightness. Very fast, excellent for finding near-duplicates in large libraries.
• pHash (Perceptual Hash): Uses a Discrete Cosine Transform (DCT) to analyze frequency components of the image. More accurate but slightly slower than dHash.
• aHash (Average Hash): Compares each pixel to the average brightness of the image. Fastest but least accurate.

The similarity between two perceptual hashes is measured by their Hamming distance — the number of bit positions where the two hashes differ. A Hamming distance of 0 means identical images. A distance of 1–5 indicates very similar images (often the same scene with minor variations). A distance above 10 typically indicates different images.

This is exactly how SammaPix TwinHunt finds both exact duplicates and near-duplicates in your photo library. All processing happens in your browser — no image data is ever transmitted to any server.

Step-by-step: finding and deleting duplicate photos with TwinHunt

Step 1 — Open TwinHunt

Go to sammapix.com/tools/twinhunt. No account required, no file size limits, no watermarks. The tool runs entirely in your browser using the File System Access API.

Step 2 — Select your photo folder

Click the “Select Folder” button and choose the directory containing your photos. TwinHunt can process entire photo libraries, including nested subdirectories. For large libraries (10,000+ photos), the initial hash computation takes a few minutes. Progress is shown in real time.

Alternatively, drag a folder directly onto the drop zone. Both methods give TwinHunt read access to the files — no modifications are made during the scanning phase.

Step 3 — Choose your sensitivity level

TwinHunt offers three detection modes:

• Exact only: Finds byte-for-byte identical files using cryptographic hashing. Zero false positives. Safe for automated deletion.
• Similar (recommended): Finds exact duplicates plus near-duplicates with a Hamming distance of 5 or less. Catches re-compressed copies, lightly edited versions, and screenshots of photos.
• Very similar: Hamming distance up to 10. Finds burst shots and photos taken within seconds of each other. Requires manual review — this mode can surface groups that are similar but not actually duplicates.

For most users, the “Similar” mode is the right starting point. It catches the vast majority of real duplicates while keeping false positives manageable.

Step 4 — Review the duplicate groups

TwinHunt presents results as groups of similar images, displayed side by side. Each group shows the file name, file size, creation date, and pixel dimensions for each image. The recommended “keep” candidate (typically the highest resolution or most recently modified version) is highlighted automatically.

You can click any image to view it at full size before making a decision. This is especially important for near-duplicates in the “Very similar” mode, where you want to confirm that the images are genuinely equivalent before deleting.

Use the “Select all duplicates” button to auto-select the recommended deletion candidates across all groups, or review and adjust each group manually. TwinHunt never pre-selects files for deletion without your explicit confirmation.

Step 5 — Delete selected duplicates

Once you have reviewed and confirmed your selections, click “Delete Selected”. Deletions move files to the Trash (on macOS and Windows) rather than permanently deleting them immediately. This gives you a safety net if you change your mind after the operation.

After deletion, TwinHunt shows a summary: total files deleted, total storage recovered, and a breakdown by group.

Exact vs near duplicates: how to decide what to keep

For exact duplicates, the decision is easy: keep one copy, delete the rest. All copies are identical so there is no quality consideration. Keep the one in your primary, organized library location and delete copies in backup folders, downloads, or synced directories.

For near-duplicates, use these criteria to decide which version to keep:

• Higher resolution wins. If two images show the same scene and one is 4000×3000 pixels while the other is 1200×900, keep the higher resolution version.
• Larger file size often means better quality. Between two otherwise equal images, the larger file typically has less compression, meaning less quality loss.
• Prefer originals over edited copies. Keep the RAW or unedited original. Edited JPEGs can always be regenerated from the original; the reverse is not true.
• Check EXIF metadata. The original photo preserves EXIF data (camera settings, GPS, timestamp) that an edited copy may have stripped.

Preventing duplicate accumulation going forward

Cleaning your library once is satisfying. Keeping it clean over time requires a few systematic habits.

Establish a single source of truth.

Decide where your canonical photo library lives — whether that is Apple Photos, Google Photos, Lightroom, or a folder structure on an external drive. All other locations (phone camera roll, cloud syncs, backup folders) feed into this one library and are cleared regularly.

Cull on import.

The best time to remove near-duplicate burst shots is immediately after an import session, while you still remember which frame was best. Letting these accumulate means doing the decision-making work later when context is lost.

Run TwinHunt quarterly.

Even with good habits, duplicates accumulate. A quarterly deduplication scan catches what slips through. With TwinHunt running entirely in the browser, it takes less than five minutes for a library under 5,000 photos.

FAQ

Will TwinHunt find duplicate photos even if they have different filenames?

Yes. TwinHunt uses perceptual hashing which analyzes the visual content of the image, not the filename. A photo named IMG_4721.jpg and its copy named vacation-photo.jpg will be detected as identical regardless of the name difference.

Can TwinHunt find duplicates across different formats (JPEG and PNG of the same image)?

Yes. Perceptual hashing operates on the decoded visual content of the image, not the encoded bytes. A JPEG and a PNG of the same photo will produce very similar perceptual hashes and be grouped as near-duplicates. Cryptographic hash matching (for exact duplicates) requires byte-identical files, so it would not catch cross-format copies — but perceptual hashing does.

Are my photos sent to any server?

No. TwinHunt processes all images entirely within your browser using JavaScript. No image data, no thumbnails, and no hash values are transmitted to any external server. Your photos never leave your device.

How large a photo library can TwinHunt handle?

TwinHunt can process libraries of tens of thousands of images. For very large libraries (50,000+ photos), processing time increases but the tool remains stable. Processing speed depends on your device's CPU and the image resolutions in the library. Most libraries under 10,000 photos complete in under two minutes.

What happens to deleted files?

Deleted files are moved to your operating system's Trash (Recycle Bin on Windows, Trash on macOS). They are not permanently deleted immediately. You have a recovery window to restore anything that was deleted by mistake before emptying the Trash.