How to Swap Any Character Into Any Video (Seedance 2.5)

The base video decides how good the swap looks. Choose one with:

• One person, facing the camera most of the time
• A steady camera, like a phone propped on a table
• A visible face and hands, with no long stretches of hands covering the face
• A length of 2 to 30 seconds, and an MP4, MOV or WebM file of 200 MB or less

Talking and singing clips are ideal because the lips carry the performance. A clip with a strong beat is ideal for the final cut, so a song or a rhythmic voiceover works best.

Seedance copies the face and hair from your character photo, so the photo matters more than anything else you add.

• A sharp, front-facing face close-up in even light, with nothing covering the face
• A full-length outfit shot if the outfit should change too
• No other people in either photo

If the character already lives in your library, you can pick it directly in the next step. If not, make the photos first with New photo in Studio, or upload your own.

You do not need to rebuild the first frame of the base video. Seedance keeps the room, the framing and the pose from the base video and only takes the identity from your photos.

1. Open Studio and choose Edit video.
2. Under Reference video, drop your base video, or pick a Studio video you already made. This is @Video1.
3. Open Output settings and choose Seedance 2.5 as the model.
4. Under People, scene or style references, add the character's face photo, either uploaded or picked from your library. This is @Image1. Add the outfit shot as @Image2 if you have one.

The duration and the aspect ratio follow your base video automatically, so the swap comes out the same length and the same shape.

You can do this two ways. The quickest is to assign each reference:

1. Set Reference 1 to Replace a person.
2. In Which person in the clip?, describe the person you are replacing, for example "The woman singing to the camera".
3. If you added an outfit shot, set Reference 2 to Change an outfit and describe the same person.

Or leave the references on Follow my instructions and write the edit yourself under What should change?:

Three things make this prompt work:

1. Naming @Video1 and @Image1 tells Seedance exactly which input supplies the motion and which supplies the person.
2. "Keep everything else exactly" locks the room, the framing and the performance to the base video.
3. Matching the look of the original clip means the swap does not jump out as a different camera.

1. Set Sound to Silent in Output settings. You will use the base video's own audio in the final cut.
2. Check the cost preview, then press Generate edit.

The swap lands in your Studio library when it is ready, and you can download it from there.

Repeat Steps 3 to 5 with your second character and the same base video. You now have two videos with the same timing, which is exactly what the final cut needs.

Download both swaps and the original base video. Place the swaps on two tracks in your editor, one above the other, and use the original base video's audio for the whole timeline.

Because both swaps share the base video's timing, the lips match the audio on every track.

This is the step that turns two swaps into one video people rewatch.

1. Find the beat. Most songs sit between 90 and 130 BPM, which puts a beat every 0.45 to 0.65 seconds. Put markers on the strongest vocal hits or drum hits.
2. Start on character A.
3. At the first marker, cut to character B. Switch back at the next one, and keep alternating.
4. Cut on the hit itself, not just after it. A cut that lands a few frames early feels tight, and one that lands late feels like a mistake.
5. Because both tracks share the same timing, the mouth, the head and the hands keep moving through the cut. Only the person changes.

For a faster feel, cut on every hit in the opening, then switch every second beat once the verse starts. Rapid cuts early hook the viewer, and slower cuts later give them time to take in each character.