A fingerprint is not the movie
A file fingerprint is a short value calculated from selected bytes. Subtitle databases can store the same value alongside subtitles uploaded for a particular release. When your calculated value matches a stored value, the database can return those subtitle records without receiving the underlying video.
A fingerprint is useful for lookup; it is not a playable excerpt and cannot be transformed back into the movie. SubtitleFinder performs the calculation with browser file APIs after you deliberately choose a local file.
Why SubtitleFinder calculates more than one hash
Subtitle providers use different fingerprint formats. SubtitleFinder calculates the OpenSubtitles movie hash from the file size and the first and last 64 KiB, the SubDB MD5 fingerprint from the first and last 64 KiB, and the NapiProjekt fingerprint from the first 10 MiB. These calculations all happen locally.
The browser then sends the resulting hash strings, file size, filename, and selected language to the SubtitleFinder API. The API queries the supported providers and combines their available results. The raw chunks used during calculation are never included in that request.
Why a hash match helps synchronization
A movie title describes the work; a file hash describes a particular binary file. Exact binary matches normally share the same edit, duration, and encoded timing, which makes subtitles associated with that file much more likely to line up than an arbitrary title match.
Hash matching is not a guarantee of subtitle quality. A submitted subtitle can contain transcription mistakes or incorrect timestamps, and provider records can be incomplete. It is best understood as strong release-identification evidence rather than a quality score.
What changes the fingerprint
Re-encoding, trimming, remuxing, adding tracks, or changing metadata can produce a different binary file and therefore a different fingerprint. Two visually identical videos may not hash alike. Conversely, renaming a file does not change its contents, so content-based hashes remain the same.
The privacy boundary
SubtitleFinder does not upload or store the selected video. Processing begins only after file selection, and the browser reads the required portions locally. The API needs lookup information, so fingerprints and basic file metadata do leave the device. That distinction is why the interface describes the process as local fingerprinting rather than claiming that no information is sent.
- Stays on your device: video contents and locally read chunks.
- Sent for the search: fingerprints, file size, filename, and language.
- Sent for product analytics: filename, coarse file-size bucket, extension, language, result count, and provider. Funnel events do not contain the file hashes.