How to use this tool
- Select one MP4 or MOV up to 50 MB with one non-cover video stream and at most one audio stream. Files may contain up to 32 total tracks.
- Run tag removal and review the before/after counts and common tag families. Original values remain in a closed local-view panel.
- Download the remuxed video and check playback and orientation. The optional audit contains technical counts and common-family flags without original tag values or filenames.
Check the built-in example
The fictional sample is a one-second blue video, 160 × 90 pixels, with H.264 video and AAC audio. It carries example title, artist, recording date, location and camera make/model tags. The actual browser export changes 20 reported tag entries to 9 technical entries. The checked common families are no longer reported. Independent ffprobe inspection finds identical compressed video and audio packet hashes in this sample. Nine remaining entries do not mean a metadata-free file, and the count reduction is not a privacy score.
Processing method
A locally served FFmpeg WebAssembly engine reads container and track information with ffprobe. It disables automatic global and per-stream metadata mappings, excludes chapters and copies the selected media streams into a new container. This uses stream copy rather than a video encoder. The linked FFmpeg reference explains negative metadata mappings and stream selection. A second probe inspects the result; an output with a reported common location, date, camera, title or artist tag is withheld. The original media file is not modified.
Reference: FFmpeg: metadata mapping and stream copying
What the result leaves out
The first run loads about 31 MB of engine code from this website. Input is limited to 50 MB, ten minutes, 32 tracks, a longest video side of 4096 pixels and 8.85 megapixels. Exactly one non-cover video and at most one audio track are supported. Chapters, data, subtitle and cover tracks are omitted. Total processing is limited to 90 seconds. Technical headers, orientation, color information, timing and data embedded inside codec packets can remain. Picture and sound are not redacted. Unrecognized proprietary information may not appear in the probe report. This is tag cleanup, not forensic anonymization, and some codecs or browser previews may be unsupported.
Checking local processing
Files are processed on your device without a source upload. Only engine assets are fetched from this site. The original-tag panel shows values locally; tables display up to 50 entries with values limited to 500 characters. The audit intentionally omits original filenames, original tag keys and original tag values. It retains technical properties, counts and fixed common-family flags. Cancel stops the engine. Reset clears the current view and selected file, while downloaded copies remain on your device.
Common questions
Will the output contain no metadata at all?
No. A usable container retains technical information, and codec-embedded or unreported data can remain. Inspect what the report covers.
Does this blur faces, remove speech or erase embedded telemetry?
No. Compressed video and audio are copied; their audiovisual content and embedded data are not sanitized.
Will all original tracks be preserved?
No. One video and optional one audio track are kept. Additional video/audio tracks are unsupported; chapters and other track types are omitted.