Does my video leave my device?
No. Face detection, crop planning and export run in your browser. A small face model and processing libraries download when needed, but your video is not uploaded. No account or DojoClip watermark is required.
Turn a landscape recording into a vertical video that follows your chosen person. Compare the original and crop, correct the framing and export on your device. Free, with no upload, account or added watermark.
Processed on your device · No account · No added watermark
Choose a video
Your video stays on this device. A small face model and processing libraries download when needed. No account, upload or watermark.
Use smart cropping for an interview, presentation, lesson or talking-head video when you already know which recording to publish. A moving crop can follow the person you choose instead of leaving them at the edge of a fixed center crop. One person is followed at a time; this is not a two-person split-screen layout.
A face can stay visible while a useful gesture, product or slide disappears outside the frame. Check those details along with entrances, exits, crossings and scene changes. Use less zoom or a wider aspect ratio when context matters. No fixed duration or file-size cap means capacity depends on your device, not that every recording will process successfully.
To publish only one passage, trim the video first. For a stationary area use the video cropper; to fit the whole picture into another shape, try the social media resizer. Add subtitles after framing so the text fits the final composition, or explore all free tools.
No. Face detection, crop planning and export run in your browser. A small face model and processing libraries download when needed, but your video is not uploaded. No account or DojoClip watermark is required.
Yes. Choose a detected face for the whole shot or from the current playback position. You can switch people later. Tracking uses position rather than persistent identity, so review crossings, head turns and every new shot.
No. You select the person. This tool does not analyze speech, identify active speakers, generate subtitles or choose highlights from a long recording.
The crop holds its last reliable position when tracking is missing or uncertain. Review the marked moments, select a person again or set a manual position. Manual positions apply until the next person selection or shot; undo and redo let you compare changes.
Choose 9:16, 4:5 or 1:1. Input must have a video codec your browser can decode. There is no fixed duration or file size limit; available storage, memory and browser codec support determine practical capacity.
No. A vertical crop keeps only part of the original image. Source-detail output avoids upscaling; 1080 output may enlarge the crop without adding detail. Review on-screen text, hands and objects near the edges before export.
Yes. Keep the original audio or export muted video. The tool prefers MP4 when compatible encoding is available, with WebM as a fallback. Export re-encodes the picture; it is not a lossless crop. Output is limited to 30 fps, a 1080-pixel short edge and a 1920-pixel long edge.
Yes. Build final preview renders the actual export, which can take time. The result player and download use the same file. Changing the framing or output settings clears the old result; build it again to check the new version.