Why album exports repeat media
A photo can belong to several Google Photos albums. The export may include copies under album folders in addition to a chronological folder, so raw file counts can be much larger than the number of unique images.
Those repeated paths carry useful album relationships even when their bytes are identical. Flattening folders before inventory can destroy that context.
Use full content hashes for automatic decisions
A SHA-256 hash compares file contents rather than names, timestamps or thumbnails. Matching hashes are strong evidence that two exported paths contain the exact same bytes.
Hash deduplication should not rely on file size alone. A collision-resistant full hash also needs to be computed from the entire file, not a prefix chosen for speed.
Do not delete visual or named similarities automatically
An edited crop, transcoded video, recompressed JPEG or changed description may share a basename and look similar while containing different data. Those variants should remain separate unless the user makes an explicit review decision.
Likewise, camera filenames such as IMG_0001.JPG repeat across devices and years. Filename equality is useful for sidecar matching context, not proof of duplicate media.
Preserve albums without multiplying storage
After exact duplicate paths are reconciled, album output can use hard links when the filesystem supports them. Several album paths then reference the same underlying data without creating several physical copies.
When hard links are unavailable, safe copies preserve the visible album structure. The report should identify which method was used.
Verify the deduplicated result
Compare unique-media counts, album membership and the duplicate report. Inspect files with numeric suffixes, edited variants, Live Photo resources and repeated camera names before deleting any extracted workspace.
Keep the original ZIPs so a mistaken manual decision can be reversed.