Raindance AI: Plan an Original Music-Inspired Video

The Dave and Tems reference

Dave's “Raindance,” featuring Tems, has an official music video filmed in Lagos, Nigeria. Universal Music Canada's announcement dates its release to January 9, 2026. That is the music-video reference discussed in this article; it does not establish a particular AI trend or a built-in generation template.

Take a broad creative cue, such as relaxed interaction or a shared performance, and turn it into your own scene. This article provides a Raindance creative-direction preset requesting the song, vocals, lip sync and relaxed performance movements. The song recording and music video are not supplied as references. It is not affiliated with Dave, Tems or their labels.

Direct the seaside sequence

The Raindance reference is a coastal music-video sequence: the partner sits on a wooden pier, the singer walks toward camera, sings in shallow water, and joins the partner on the sand. Use photo 1 for the seated partner (Dave role) and photo 2 for the singer (Tems role). Those roles stay fixed through the scene changes.

The article entry supplies a seaside direction designed for this site’s two-photo generator. The first upload remains the seated partner and the second remains the singer through each scene. Follow those role labels when preparing the references. The prompt asks for a pier entrance, singing close-ups and a shared beach scene, with a consistent cast and soundtrack direction throughout. Review the generated sequence before sharing it.

Raindance performance direction

Recreate the seaside sequence from Dave - Raindance (ft. Tems) with the uploaded subjects. Photo 1 is the seated partner (Dave role); photo 2 is the singer (Tems role). Keep each face, hair, body and outfit consistent across every shot; never swap roles. Warm coastal daylight, wooden pier, open sea and sandy beach. Sequence: wide pier shot with the partner seated as the singer walks toward camera; cut to the singer in shallow water, singing directly to camera with restrained hand gestures and gentle swaying; finish with both seated together on the sand, relaxed and interacting naturally. Cinematic wides and singing close-ups; subtle sea breeze. Soundtrack: Dave - Raindance (ft. Tems), with its music and vocals. The singer lip-syncs Tems' corresponding vocal phrases precisely; the partner lip-syncs only when Dave's vocal part is heard. Match mouth shapes, gestures and cuts to the same song passage. No microphone, studio, rain-dance choreography, extra people or text.

Use this Raindance prompt →

This entry fills the prompt and selects the Raindance seaside scene, Seedance 2.5, 15 seconds, 16:9 framing and 480p. Check the credit charge before submitting. The scene direction requests an AI interpretation; the original video and song recording are not supplied as references, so exact choreography, music and lip sync need review.

Prepare the references and first attempt

Use two separate, clear photos, with one consenting adult in each. Avoid blurred faces, heavy filters, crowded backgrounds and tightly cropped heads. JPG, PNG and WebP files are accepted up to 10 MB per reference. The two-photo guide explains preparation in more detail.

In the studio, keep the Raindance seaside scene or edit the direction, framing and duration. Portrait 9:16 suits vertical edits; landscape 16:9 gives a duo more horizontal room. Review the displayed credit cost before submitting. Browsing the examples is free, while new personalized videos require paid credits.

Music and movement need separate checks

The preset explicitly requests Dave - Raindance (ft. Tems), its music and vocals, and synchronized lip movements. These are generation instructions; the original recording is not attached and accurate song reproduction or lip sync is not guaranteed. The studio has no song upload, song selector, precise lip-sync control or choreography reference upload. Seedance 2.5 requests generated audio, which can vary; H3 has no audio-generation setting.

If a selected soundtrack is essential, plan an external editing step using music you are authorized to publish. Adding a track later does not make the generated mouth movements match its lyrics. For exact dance timing or a faithful multi-location recreation, this studio workflow may not meet the brief.

Before sharing, inspect facial consistency, hands, feet, rhythm and the soundtrack. Simplify the movement if either performer changes appearance or moves unnaturally. Clearly describe generated material as AI-created when presenting it to viewers.

Assign the singer before choosing your photos

The seaside sequence has two different jobs. The seated partner establishes the opening composition, while the singer carries the walk and the close-up performance. Put the person you want singing in the second upload slot. This assignment describes the role in the video, not the person's gender or their relationship to the other subject.

Choose the clearest face reference for the singer because close-ups expose small changes in eyes, mouth shape and facial proportions. For the seated partner, a waist-up or wider photograph can provide more information about posture and clothing. Both people should be recognizable without relying on the backgrounds of their original photos.

If you swap the photos, review the role descriptions before generating. The prompt must still refer to photo one as the partner and photo two as the singer. Do not ask the same image to become both characters, and avoid directions that exchange their identities between the pier, water and beach shots.

Keep the coastal sequence readable

The preset uses fifteen seconds and landscape framing as a starting point. That gives the opening pier view, singing close-up and shared sand scene more room than a very short attempt. It does not guarantee that the model will allocate an equal amount of time to each shot. Review the displayed settings and cost if you change the duration or resolution.

Describe one clear action per scene. On the pier, the partner sits while the singer approaches. In the shallow water, the singer performs toward the camera with restrained gestures. On the sand, the two subjects share a relaxed moment. Extra running, spinning or elaborate dance steps can distract from that sequence and make continuity harder to assess.

Keep clothing and identity consistent across the cuts. A scene change should alter the location and framing, not create a new person. If the water shot loses the singer's likeness, use a clearer reference or simplify the requested movement before adding more camera instructions. The prompt requests coastal daylight; it does not need a storm or literal dancing in rain.

Review the song passage and performance together

Listen to the full generated soundtrack before judging the mouth movements. Check whether the requested song and vocal parts are actually present, whether the singer's visible delivery follows the heard phrase, and whether the partner performs only during the relevant vocal part. Treat the instructions as a brief for the model rather than evidence that those requirements were fulfilled.

Next, inspect the cuts without sound. Check facial continuity, walking posture, hands and the transition onto the sand. If you keep a useful visual result and add authorized music later, that editing step can improve the soundtrack choice, but it cannot automatically repair the generated mouth timing.

Choose the concept that fits

The seaside preset follows a different scene from the Hotel Lobby studio examples; those examples do not demonstrate a tested Raindance output. For the orange-stage format, read Quavo & Takeoff AI; for settings and submission steps, use the creation guide.