Surface image pairs that look broadly similar
Similar Image Finder helps narrow a large image set to candidate pairs that may deserve closer review. It uses a lightweight perceptual hash, which compares coarse visual structure rather than requiring byte-for-byte identity.
Similarity is not duplication
A small perceptual-hash distance means two images have similar overall structure. Crops, text overlays, color changes, retouching, product variations or small revisions can still make them meaningfully different.
Why this differs from an exact file hash
Two visually identical-looking files can have different bytes because of metadata, compression or export settings. Conversely, two files with similar perceptual hashes are not guaranteed to be the same asset. The finder is therefore a review tool, not an automatic duplicate remover.
Use a pairwise compare when needed
If a candidate pair may represent two revisions of the same asset, continue with Image Version Compare for dimensions and normalized pixel-difference evidence.
Nothing is deleted automatically
The tool only surfaces candidate pairs and never removes or changes images. Confirm whether the files are redundant, intentional variants or separate assets before deleting, consolidating or replacing anything.