iOS/Android Storage Breakdown & Duplicate Photo Perceptual Hash (dHash) Finder (2026)

Cryptographic hashes like SHA-256 exhibit the avalanche effect: changing a single pixel, stripping one EXIF GPS tag, or re-saving an image at 95% JPEG quality changes ~50% of the output bits. Perceptual hashes (pHash, dHash, aHash) hash the low-frequency visual structure of an image so resized or recompressed copies produce identical or near-identical 64-bit integers.

First, the image is resized to 9 pixels wide by 8 pixels high (72 pixels total) and converted to grayscale luminance. Next, each pixel in a row is compared to its right neighbor (9 comparisons become 8 boolean bits per row). Across 8 rows, this produces a 64-bit fingerprint encoding relative brightness gradients.

Hamming distance is the number of bit positions at which two 64-bit hashes differ, calculated in hardware via popcount(hashA XOR hashB). A Hamming distance of 0 means all 64 gradient bits match; 1 to 5 bits typically indicates a JPEG recompression, slight crop, or consecutive burst frame; 20+ bits indicates a completely different scene.

iOS System Data aggregates APFS local Time Machine/OTA update snapshots, Siri neural voice assets, Safari/WebView IndexedDB caches, streaming media ring buffers, and orphaned Spotlight/Photos syndication indexes that have not yet been purged by macOS/iOS cache_delete daemons.

No. Image decoding and 9x8 pixel downsampling happen exclusively inside your browser's local HTML5 Canvas API via FileReader/URL.createObjectURL. Your photos never leave your device.