Three Meters: Balcony and Dance Floor
Two photos, a shared look across the club and the original soundtrack. 14 seconds, 9:16.
Upload two photos to star in a club scene: one character looks down from the balcony, then the camera shifts to the dance floor and finds the other among the dancers. Your images define the leads’ appearance and visual style, while a prepared reference guides motion and framing. Background clubgoers stay varied. The original soundtrack is added after generation. The result is a 14-second portrait video in 9:16, at 480p or 720p.
How it works
- 1Choose the balcony character
Upload a clear photo of one person or character. They will stand at the upper balcony rail and look down. Waist-up or full-body references with a visible face work best.
- 2Choose the dance-floor character
The second photo identifies the person the camera finds on the dance floor. They are a separate lead; the surrounding crowd does not become copies of them.
- 3Get the clip with its original soundtrack
Choose 480p or 720p. After preparing both characters, the club scene is generated and the original soundtrack is added. The result is a 14-second portrait video in 9:16.
FAQ
Why do I need two photos?
The first photo is the balcony lead; the second is the lead on the dance floor. Both roles are required. Background clubgoers remain varied and do not copy either lead’s face.
Which photos work best?
One person or character per photo, with a clear face and no heavy shadows or obstructions. Waist-up or full-body photos help preserve clothing. Use images you are allowed to use for this generation.
What controls the motion and camera?
A prepared reference guides character positions, camera motion and timing. Appearance comes from your photos: the balcony lead first, then the dance-floor view and the second lead below. Motion accuracy depends on the generation.
Can I choose landscape or a different duration?
This scene uses a fixed 14-second duration and portrait 9:16 format to preserve the reference composition and pacing. You can choose 480p or 720p.
Does the model regenerate the music?
No. Seedance generates the visuals without audio. The original 14-second track is added afterwards, so the finished result already includes sound.
Is the character’s visual style preserved?
Preparation and video prompts preserve each reference’s face, clothing and style. An illustrated character should remain illustrated, and a photo should stay photographic. Exact likeness depends on the model and source quality.
How much does the clip cost?
496 credits at 480p or 1084 credits at 720p. The price includes both character sheets, reference-guided video generation and adding the original soundtrack. Credits are returned if generation ultimately fails.