Find faces locally, choose which tracks to cover and correct the masks in the preview. Use blur, mosaic or a solid cover, then export a short video without uploading it. Free, with no account or added watermark.
Processed on your device · No account · No added watermark
Choose a video
Detection creates separate tracks for each continuous appearance. All detected tracks are selected initially.
A small face detector and its processing libraries download on first use. Your video and face positions stay on this device.
Use this tool to mask faces in an event excerpt, a street clip or a short video you want to share. Automatic detection provides a starting point. You choose the tracks to cover and repair missed or misplaced masks without sending the original video to a processing server.
Check where a face enters or leaves the frame, turns sideways, passes behind an object or appears after a cut. Short face tracks do not provide persistent identity recognition, and one selected track may not cover every appearance of the same person. Leave useful margin around the face and inspect the motion between your keyframes.
Face blurring changes the covered picture area. It does not remove a name on clothing, an address elsewhere in the image or a recognizable voice. You can cover other visible details with manual rectangles and choose a muted export, but the final file still needs your review.
The video is re-encoded at up to 1080p and 30 fps, in MP4 or WebM when supported. Higher-resolution or faster-frame-rate inputs may be reduced. Detection and export speed depend on your device and browser; keep this tab open and download the result before leaving.
For background softening with the person left visible, use video background removal and replacement. If you only need silence, remove the audio track. You can also compress a finished video; review that final version before sharing. Find more free browser tools.
No. Detection, editing and export run on your device. The browser downloads a small YuNet face detector and processing libraries; those downloads do not upload your footage. No account is required, and exports have no added DojoClip watermark.
Choose a video up to 60 seconds and 100 MiB. It must use a codec your browser can decode. Large source frames are subject to processing limits and may be reduced for export. Trim a longer clip before selecting it.
Yes. Detected tracks start selected; deselect any tracks you do not want covered. A track is a short sequence of face positions, not a recognized person. The same face may appear as several tracks after a cut, an obstruction or leaving the frame. Review every relevant track.
Draw a rectangular mask, then move or resize it over the face. Set keyframes at the current playback time and adjust the mask’s start and end. Remove or reset corrections when needed. Check the frames between keyframes as well as the positions you edited.
Yes. Choose blur, mosaic or a solid cover, and adjust the available padding and strength. Make the covered area large enough for movement and changing head position. The appearance of an effect does not prove that the person is anonymous.
Export is limited to a 1920-pixel long edge and 1080-pixel short edge, with at most 30 fps. MP4 is preferred, with WebM as a compatible fallback. The picture is re-encoded to apply masks, so output quality and file size can change. The original file is not overwritten.
Keep audio is enabled by default when an audio track is available. Turn it off for a silent export. Audio may be copied or re-encoded as needed; face masks do not disguise a speaker’s voice or remove spoken names.
No. Small, side-facing, obscured or fast-moving faces can be missed, and tracks can split or move incorrectly. Review the entire preview and exported video, and correct missed coverage before sharing. Voices, text, clothing and context can still identify someone. Automatic detection does not identify people or guarantee anonymity.