AI Raindance: make a 15-second music video from one or two portraits

I built AI Raindance, a web app that puts people from your photos into a short video inspired by Dave and Tems' Raindance music video.

You choose who appears in the scene. The app supplies the reference, the waterfront setting and the original audio. There's no prompt to write: upload one or two portraits, choose the roles, and generate a 15-second MP4.

Here's how it works, and what to check before spending a video credit.

Start with the cast

There are two modes:

Mode	Photos	Who appears
Solo	One portrait	The person in your photo becomes the singer; the companion is removed.
Duo	Two separate portraits	Photo 1 is the companion, and photo 2 is the singer.

For Duo, you don't need a picture of both people together. Use a separate portrait for each person. If you've put the singer in the first slot, use Swap roles before generating.

Both modes cost one video credit. Adding a second person doesn't cost extra.

Choose a photo where the face is easy to see

A front-facing portrait or a slight angle is a good starting point. Keep the face large enough to see clearly, with even lighting and no sunglasses or heavy filters. Group shots and faces far from the camera are harder to work with.

The app accepts JPEG, PNG and WebP files, up to 10 MB per photo. Use your own photos, or photos you have permission to use with the subject's consent.

Sign in before selecting them. The uploaded photos aren't saved as an unfinished draft, so leaving the page before submitting can lose your selection.

Make and download the video

1. Sign in with Google. If your balance is empty, buy a one-time video pack.
2. Choose Solo or Duo, add your portraits and check the role labels.
3. Click Generate to submit the task and use one credit.
4. Open My videos to follow its progress, play the result and download the MP4.

Once you've submitted the task, you can close the page. Return with the same Google account to find it again. You don't need to keep a browser tab open throughout generation.

The finished video is 15 seconds, 768P, in a 4:3 frame, with the original reference audio included. Your portrait changes the person on screen; it doesn't turn your voice into the song.

Watch the whole clip before sharing it. Facial details, expressions and singing timing can vary between AI results.

Look at the sample before buying

The examples page separates the original reference footage from a generated Solo sample.

That sample is a short 4.5-second clip made from a fictional AI portrait, with its model-generated soundtrack. It isn't a full preview of a purchased video. Purchased videos use the fixed 15-second template and original Raindance audio.

The template is deliberately specific: you can choose the portraits and roles, but you can't change the song, scene or duration.

Cost and file storage

Generation is paid. The current packs are $9.99 for one video, $24.99 for three, or $69.99 for ten. There is no subscription, and purchased credits don't expire. If a task is confirmed as failed, its credit is returned automatically.

Photos are kept for seven days and finished videos for thirty days. Download the MP4 while it's available. Uploads and finished videos stay in your account; they aren't automatically added to the public examples page.

AI Raindance is an independent project, with no affiliation with Dave or Tems.

You can try the flow at airaindance.online. I'd welcome feedback on the Solo/Duo controls, especially whether the companion and singer roles are clear before you upload.