How to Make a Raindance AI Video From Two Photos

What you are making

The Raindance AI trend recasts a short seaside scene set to "Raindance" by Dave featuring Tems. One person waits on the end of a wooden pier, the other walks up the pier and sings to the camera, and the clip ends with both of them sitting on the sand. The generator uses one fixed reference cut of that scene, a community swap by Tyler (@tyleeeee15) on TikTok, so every video follows the same shots, camera moves and timing. Only the two people change.

The first 10 seconds of the reference clip. Your video keeps these shots and this audio.

1. Pick one photo per role

The two slots are roles, not left and right. Each photo stays with its role through every cut, even when the two people cross the frame.

Each slot takes one photo of one person. A group photo of both of you won't work, because the model can't tell which face belongs to which role. If you put the photos the wrong way round, Swap roles flips them without uploading again.

The photo decides most of the likeness. Front-facing, evenly lit, head to waist, nothing over the face. The full list is in Best photos for Raindance AI.

2. Choose model, length and resolution

The credit price for exactly these settings shows above the button before you start, and the cost table lists every model and length. Raindance AI videos use purchased or subscription credits; signup gift credits don't cover them.

A cheap way to start: make a 5-second Seedance 2.0 Mini video at 480p to see how your photos hold up, then make the long version on the model you prefer. On Seedance 2.0 Mini, Fast and 2.0 a 30-second video is made as two 15-second parts and joined; Seedance 2.5 makes all 30 seconds in one pass. Either way you pay once for the whole video.

3. Generate and follow the progress

Press Generate my Raindance AI video. If you're not signed in, you sign in first and your photos and settings are still there afterwards. The credits are charged when the video starts.

The progress bar moves through three stages:

1. Checking photos (Seedance 2.5 only): the model reviews both photos and the reference clip, usually a minute or two.
2. Putting both of you on the pier: the model makes the video. Most videos are ready in a few minutes.
3. Adding the reference audio: the video is cut to the exact length and the audio is laid in.

You can leave the page. The video keeps going and shows up under Your recent videos when you come back. If it fails, the credits go back automatically and the page tells you whether the photos were refused or the service broke.

4. Check the video before you post

Watch it to the end and look at:

• The wide opening shot. Faces are smallest here, so this is where a likeness slips first.
• The cut after the close-ups. Check that person 1 is still person 1 when both of you are on screen.
• Hands and overlaps. When the two people cross or touch, check for extra fingers or merged shapes.

If one face drifts, a sharper, front-facing photo usually helps more than a different model. A finished video can't be edited; change the photo or the model and generate again. Only failed videos are refunded.

5. Download and post to TikTok

Press Download MP4. The file is 4:3 with the reference audio already in it, so it plays correctly anywhere. On TikTok you have two choices:

• Post it as it is. The sound is part of the video.
• Use the official sound. Turn the original sound down in the editor, tap Add sound and pick "Raindance" by Dave and Tems. Your post then appears on the trend's sound page. More on this in Raindance AI video with the song.

TikTok shows a 4:3 video with bars above and below. Videos made before your first purchase carry a small moving watermark; videos made after any purchase don't. Only post people who agreed to it, and follow TikTok's rules for AI-generated content.

Troubleshooting

Ready? Open the Raindance AI generator.