Phone Cleaner
memFree scans your photo library on-device and surfaces the duplicates, near-duplicates, and clutter that are actually safe to delete. Nothing is uploaded. No account is required.
Free to use On-device AI No account needed
What it does
Most phones accumulate thousands of photos over years — duplicates from burst mode, near-identical shots from trying to get the right angle, screenshots that were useful once and forgotten. By the time storage fills up, manually reviewing that library is not realistic. memFree does it automatically, on your device, without any of it leaving your hands.
The app is for anyone whose phone has thrown up the "Storage Almost Full" warning one too many times. It works on iPhone, iPad, and Android. No subscription, no account, no sync to a server. Scan, review suggestions, delete what you don't need.
The design principle is restraint. memFree doesn't show you every photo in your library or ask you to curate an album. It surfaces what's waste, lets you confirm, and gets out of the way.
Key features
Each feature solves a specific kind of storage waste — no filler.
Perceptual hashing groups exact duplicates and near-identical shots together. You see the groups, pick the best, and confirm deletion.
Screenshots accumulate fast and are rarely revisited. memFree isolates them, shows you total storage consumed, and lets you bulk-delete with one tap.
The heaviest files sorted by size. Photos and videos consuming the most space surface immediately — share before deleting if you need to keep a copy somewhere.
Three compression levels let you reclaim video space without deleting. See the projected size before committing. Compression runs natively — no upload, no waiting on a server.
All scanning and analysis runs locally. No photos are uploaded, no account is created, no server sees your library. The analysis happens between the app and your device.
Under the hood
Duplicate detection uses perceptual hashing — each photo is converted to a compact hash that captures visual structure rather than pixel-exact content. Two shots of the same scene taken seconds apart will produce hashes close enough to be grouped together, even if they differ in compression, exposure, or minor framing. The comparison is entirely synchronous and runs in a background isolate, keeping the UI thread free throughout a full-library scan.
Video compression is handled through video_compress, a native
wrapper around platform video processing APIs on both iOS and Android.
Three quality tiers map to real target bitrates rather than arbitrary
percentage sliders. On iOS, the app uses Apple's AVFoundation
pipeline under the hood; on Android, it delegates to MediaCodec.
The size estimate shown before compression reflects the actual output
parameters — it's a calculation, not a guess.
The app ships with 12 localizations including right-to-left Arabic. Layout
responds correctly to both text direction and font scaling. State management
runs on Riverpod 2.x with an async notifier pattern — data loads at startup
in background isolates, so every screen opens instantly on already-computed
state. The photo library access layer wraps photo_manager with
platform-specific guards: on iOS, iCloud assets that return zero byte-size
are detected and skipped before they reach the UI, preventing NaN crashes
in storage calculations.
Download
Free to download. No subscription. No account required.