How to make a Raindance AI video

What is the Raindance AI workflow?

The Vismuse seaside preset combines a wooden pier, golden-hour light, ocean waves and relaxed singing-style gestures. Your photo supplies the performer’s face, hair and visible outfit; a detailed prompt supplies the scene, movement and camera.

The current workflow uses Seedance 2.0 Fast to generate new footage. It does not send the original performance video to the model or reproduce a fixed sequence of its shots. The locally bundled original soundtrack is added after generation, so choreography and mouth timing can vary.

1. Choose Solo or Duet

Solo needs one clear photo of the performer. Duet needs two separate photos: upload the left performer first and the right performer second. These are appearance references, rather than the first and last frames of your video.

Start with the cast you want in the final scene. Duet keeps both performers visible and lets them alternate small gestures without crossing positions. You do not need a photo of the two people standing together.

2. Prepare faces, hair and outfits

Choose sharp, well-lit portraits with one intended subject per image. Include the hair, shoulders and enough of the outfit to show its shape and color. A tiny face crop gives less clothing information, and a group image makes identity selection ambiguous.

• Keep eyes and face visible; avoid dark glasses, blur and heavy filters.
• Use distinct images for the two duet roles.
• Keep hands away from the face in the input photo.
• Use your own photos or photos you have permission to use.

3. Choose 5 or 10 seconds

Raindance supports 5- and 10-second clips, subject to account access, with Seedance 2.0 Fast output at 480p or 720p. Landscape 16:9 is the default. A short take is useful for checking the cast before trying a longer version.

Generation requires available paid credits through a subscription or credit pack and enough credits for the selected settings. Free trial credits alone do not qualify. Review the estimated credit cost and current settings; Generate opens subscription and credit-pack options if your eligible balance is insufficient.

4. Start with restrained seaside movement

The preset uses one continuous medium-wide shot: relaxed head nods at the opening, a small singing-style gesture below the chin in the middle, and a settled camera-facing pose at the end. A duet alternates the lead without changing sides. The camera makes a slow push-in.

You can add scene, outfit or camera changes, and your written direction takes priority. Keep the sunset pier for a first take so you can assess likeness before adding rain, a new location or more complex movement.

5. Review the result with sound

Submit once and wait for completion. Watch both faces throughout the clip, especially when a person turns or raises a hand. Check that the hair and clothing remain recognizable and that the two identities stay separate.

With audio enabled, the export pipeline adds the matching opening segment of the locally bundled original soundtrack. This applies to the current Solo and Duet workflows. The recording is added after video generation, so it does not clone your voice or guarantee precise singing lip-sync. Listen to the downloaded MP4 before sharing.

Is the example above a current-model result?

The video above is an earlier generated sunset-pier duet with fictional performers. Its orientation was corrected in editing. It illustrates the scene and two-person composition, rather than demonstrating the current Seedance 2.0 Fast workflow.

Current generations use your photos and written scene direction to create a fresh performance. Do not expect the earlier example’s exact camera movement, face quality or timing from every new take.

Why do faces or clothes change during movement?

Start with clearer portraits and fewer gestures. For clothing mismatch, include more of the intended outfit and remove conflicting instructions. Keep hands below the chin and avoid asking performers to cross or overlap.

For duet identity mixing, check the Left and Right upload labels and explicitly assign the two photos. Higher resolution alone may not fix identity drift. Change one input or direction at a time; a new video generation uses credits.

Can I make a vertical Raindance video?

You can choose another available aspect ratio to change the composition. A portrait frame has less room for a side-by-side duet, so check that neither face or outfit is cut off. Solo is simpler to frame tightly.

An editor is optional for placing a landscape result on a vertical canvas, adding captions or trimming. You do not need CapCut or a source video to begin the current photo-and-prompt workflow.

Can I make it for free or use my own voice?

Reading this guide and watching the example are free. Personalized generation requires paid credits from a subscription or credit pack. This entry creates a short performance, not a free full-length music video.

An uploaded portrait does not supply a voice recording, and the bundled soundtrack does not become your voice. Use photos and music you have permission to publish, and review the actual output before posting.