You can put yourself into a music video by pairing a reference performance with portrait photos. In faceswap, the video workflow analyzes an uploaded MP4, lets you assign a portrait to each segment, and assembles the processed clips with the original audio.
Open MV Motion Transfer to begin. Although people often search for this as music video face swap, our workflow uses character motion transfer. It can change clothing, body details, and backgrounds as well as facial appearance. Review the full result rather than expecting a face-only edit with every other detail preserved.
Plan the scenes before generating
A music video may include solo performers, crowd shots, titles, and scenery. Decide which scenes need character generation and which should remain intact. Keeping an original shot can be a useful creative choice, especially when it contains no clear subject.
Here is an illustrative plan, not a record of a completed generation:
| Segment | Duration | Choice | Portrait |
|---|---|---|---|
| Opening title | 3 seconds | Keep original | None needed |
| Solo performance | 6 seconds | Generate | Portrait A |
| Audience cutaway | 3 seconds | Keep original | None needed |
| Second solo shot | 6 seconds | Generate | Portrait B or the same A |
This plan contains 18 seconds of footage and 12 seconds selected for generation. Assigning two portraits to different scenes does not select two individual people within the same shot. The current tool cannot precisely target a named or clicked person in a group scene.
Prepare the MP4 and portrait files
Use a standard MP4 up to 100 MB. A different file renamed with an .mp4 extension is not a format conversion. If your source is another format, export a standard MP4 from your editor before uploading.
Portraits can be JPG, PNG, or WebP, up to 10 MB each. Choose a clear face with limited obstruction and a useful angle for the performance. Start with a short solo clip to judge whether the reference works before committing to a longer sequence.
Use footage you own or have permission to edit. A video available to watch online is not automatically available to download, modify, or republish. For preparation advice, see the portrait quality checklist.
Upload and wait for scene analysis
Sign in and upload the video. The first stage only requires the MP4; you do not need to decide all the portraits in advance.
When the task reaches the review stage, inspect the time ranges and original previews. Check whether the person is visible throughout each clip. A scene that begins clearly may include a sudden turn, a hand over the face, or another person entering later.
If the analysis does not produce the scene structure you want, prepare a more suitable reference clip in your video editor. The current website does not provide a manual timeline editor for moving every cut point.
Assign portraits in the canvas workspace
The workspace has a portrait library, a preview canvas, and a scene timeline. The timeline selects analyzed scenes; it is not a video editor for dragging cut points or changing the order of the source footage.
- Select Add portrait in the library and upload the face you want to use. Add other portraits if different scenes need different identities.
- Click a scene in the timeline to inspect it in the preview canvas. Choose a portrait, then select Apply to this scene.
- For repeated identities, select several scenes and use Apply to selected scenes. Use Apply to all generated scenes when all scenes marked for generation should use the same portrait. You can also drag a library portrait onto a scene.
- Choose Keep original for a scene you want to leave unchanged. For several scenes, select them together and use Keep selected originals. Kept scenes do not need portraits.
- Wait for assignments to finish saving. Check the scenes marked Needs portrait and resolve them before generation.
Adding a portrait to the library alone does not assign it to every scene. Confirm the assigned portrait in the current scene preview. If a save fails, retry the affected assignments rather than assuming the visible library selection has already reached the service.
Before submitting, review transitions between generated and original scenes. Keeping a difficult shot may preserve local quality but can create a visible identity change beside a generated scene.
Generate, inspect, and download
Submit the complete plan once. The service queues the segments, processes them, then merges the full video in the original order with the source audio track. It does not compose new music or synthesize a replacement singing voice.
Inspect the segment results and the merged video. Watch normally for overall continuity, then inspect rapid turns, glasses, hands, and transitions more closely. Check that the original audio and scene sequence make sense together.
Use individual segment regeneration when one generated clip needs another attempt. A failed task also has a retry option. Completion time varies with the footage and queue, and a new attempt is not a guarantee that a difficult scene will improve.
Download the completed MP4 after review. Identify AI editing when sharing a performance that could otherwise be mistaken for real footage.
What does the video trial cost?
As of October 18, 2026, video generation, failed-task retries, and individual segment redos deduct 0 credits during the trial. The proposed displayed rate is 1 credit per generated second, with the combined generated duration rounded up once. Original clips are excluded.
For the illustrative plan above, the proposed estimate is 12 credits and the current trial charge is 0. This is separate from photo generation, which costs 4 credits per result. Check the current tool notice before submitting; the trial is not a promise of permanent unlimited free rendering.
For a comparison of photo credits, video trials, and proposed duration estimates, read AI video face swap: free trial, costs, and limits.
Questions about music and dance video face swap
Can I create a dance video from just one photo?
This tool needs a reference video as well as a portrait. The video supplies the motion. It is not a text-only or photo-only dance generator.
Can I change different characters in the same music video?
You can assign different portraits to different segments. Selecting a particular performer inside a multi-person shot is unsupported. Choose solo footage or keep the group scene unchanged.
Can I make a Jensen Huang world tour-style video?
The Jensen Huang dancing video tutorial covers that creative reference and its limits. The original creator's video is not a faceswap output or a guarantee of the result you will receive.
What if I only need one still image?
Use the photo face swap tutorial instead. It explains the target picture and portrait reference without a video timeline.
When your footage is ready, upload your music video and review the scene plan before starting generation.
