Choose the people to replace and give each one a face photo. Swap up to 4 people in a short clip, keep the original audio and download the result. Your photos and video stay on your device.
Processed on your device · No account · No added watermark
Small, side-on or hidden faces may be skipped. Check the entire result before sharing.
Checking browser support…
Start with a sharp, well-lit portrait containing one visible face, up to 20 MiB. A fairly frontal view is easier to use than a strong profile, sunglasses or a hand over the mouth. Try a short scene with larger faces before processing a busy group shot. A larger output frame does not guarantee sharper facial detail.
In a two-person conversation, assign one portrait to each selected speaker and check the moments when they turn or cross. A detected person is an approximate match, not a guarantee that every later appearance will be recognized. Review cuts, entrances and occlusions, including frames where the original face remains visible.
For a silent video, use Remove Audio from Video. For face coverage, use Blur Faces in Video. To reduce the finished file size, use Video Compressor. Explore all free tools for other edits.
Yes. Processing and downloading are free, with no account or generation credits. DojoClip adds no watermark. Your browser still needs compatible graphics, codecs and enough memory to finish the export.
No. Face analysis, swapping and export run in your browser. Finding and grouping people first downloads about 175 MB of models. The first face swap downloads another 278 MB; scanning alone does not need that larger model. Processing libraries are additional downloads. Models are reused when the browser retains them, but clearing or losing the cache can require another download.
Yes. Assign a portrait to each person you want to replace, up to 4 mappings. People without an assigned portrait stay unchanged.
Each result covers up to 90 seconds. Files up to 300 MiB need no extra size confirmation. If a video is longer than 90 seconds or larger than 300 MiB, review and confirm the proposed beginning section: the first 90 seconds, or the whole clip if shorter. Your original file stays unchanged. Very large or complex files may still exceed browser resources. Export keeps the aspect ratio within a 1280-pixel long edge, 720-pixel short edge and 30 fps, without enlarging the source.
A desktop or laptop with a current WebGPU-capable browser is recommended. Graphics support, available memory and video codecs all matter. Mobile devices may be slow or run out of memory. Keep the tab open while processing.
The main audio track from the processed section is included automatically when the video has one. If you accept a beginning section, both picture and sound use that same time range. Face swapping does not change voices or dialogue. Audio may be re-encoded for the output format. For silent output, use the separate mute tool before or after this edit.
There is no guarantee. Small faces, profiles, obstructions, cuts and fast movement can cause missed or incorrect matches. Uncertain matches may be skipped, leaving the original face visible. Review the complete exported video, both playing and paused, before using it.