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.
Perceptual hashing analyzes image content rather than raw bytes to find visual duplicates - Photo by Luke Chesser on Unsplash
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. Find Duplicates 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 Find Duplicates read access to the files- no modifications
are made during the scanning phase.
Step 3 - Choose your sensitivity level
Find Duplicates 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
Find Duplicates 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. Find Duplicates 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, Find Duplicates 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.
A systematic approach to photo management keeps your library clean long-term - Photo by Clement Helardot on Unsplash
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 Find Duplicates 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 Find Duplicates find duplicate photos even if they have different filenames?
Yes. Find Duplicates 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 Find Duplicates 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. Find Duplicates 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 Find Duplicates handle?
Find Duplicates 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.
Originally published at sammapix.com
Try it free: SammaPix — 27 browser-based image tools. Compress, resize, convert, remove background, and more. Everything runs in your browser, nothing uploaded.
