Quick answer
Starrd's Kakalika Dance template puts you in a one-take dance to DopeNation's 'Kakalika'. Upload one photo: filmed from a phone resting low on the ground under a huge blue sky, you step back from the lens, throw three big kicks, lean far back, shoot both arms overhead, and finish pointing straight up at the sky. It's one continuous take, and every move is copied from the original clip as a depth map so it lands at the same moment. You keep your own face, hair, top and shoes, and everyone gets very wide baggy cargo pants, because the dance is built around them. Pets work too, dancing upright in the pants. The song is on it. 23 seconds, vertical 9:16, Seedance 2.5, 480p, 250 credits (about $12.50).
The same dance, any subject — tap one to make it with your photo.
A phone on the ground, a huge blue sky, and pants wide enough to have their own weather. One photo puts you in the Kakalika dance, three big kicks and all. Here it is:
Kakalika Dance
One photo. Three big kicks, a deep lean back and a point at the sky, in very wide baggy pants. 23 seconds.
What You Get
- One unbroken take. No cuts: the whole dance plays in a single shot from a phone resting low on the ground.
- The low angle. The camera looks up, so you tower over the frame and the sky fills everything behind you.
- Three kicks. At about 5, 13 and 15 seconds, and the last is the biggest, with the leg thrown out nearly level with the hips and one arm straight up.
- The lean back. At 8.5 seconds you turn side-on and lean far back from the ankles.
- Your own look. Your face, your hair, your top and shoes, in very wide baggy cargo pants.
- People or pets. A dog or cat dances upright in the pants.
- 23 seconds, vertical 9:16, with the song, ready for TikTok, Reels and Shorts.
What the Video Actually Is
The song is "Kakalika" by the Ghanaian duo DopeNation, at about 119 BPM. The clip this template copies is from TikTok creator @itsxxmellerie, posted on October 24, 2026 with a "dc @Fidelis" credit in the caption. It's a small post, with just over 500 plays when we built this, so the draw is the dance itself. She sets the phone low on a dirt patch in a park under an autumn tree line, then dances toward it and away from it in a black crop top and enormous camo cargos, one continuous take with no cuts.
Why it suits AI: the whole look comes from the low angle and the pants, and you don't need either. No park, no phone balanced in the dirt, no 23 seconds of choreography to learn. One photo covers all of it.
The 23 Seconds, Beat by Beat
| Time | What happens |
|---|---|
| 0.0s | You start close to the lens, towering over the camera, knees wide, forearms up |
| 1.0s | The head tips back, then you start stepping away from the camera |
| 2.7s | One knee lifts and both hands come up to the face |
| 4.4s | You're at your furthest from the camera |
| 4.9s | First kick, the foot out to the side |
| 7.0s | Back close to the lens in your widest stance, one hand on the chest |
| 8.5s | Side-on, at the deepest point of the lean back |
| 10.0s | Quick dusty steps, and the arms open out to both sides |
| 11.2s | Both arms shoot up overhead |
| 12.0s | Hands drop to the waist, feet wide, hips rocking |
| 13.0s | Second kick, high to the side |
| 15.1s | Third kick, the biggest, with one arm straight up |
| 16.0s | The arm at its highest, with a wide laughing grin |
| 18.7s | The widest lunge, kicking up dust |
| 20.3s | Both hands go to the head |
| 22.9s | One finger pointing straight up at the sky, held to the last frame |
Your Look, Their Moves
- Your own top and shoes. Whatever's in your photo stays. If your photo stops at the waist, the shoes are filled in to match its style.
- Baggy pants for everyone. Whatever you wear on your legs becomes very wide cargo pants, coloured to go with your outfit.
- Or your own outfit idea. Put it in the Director's Notes. The demo's silver parachute cargos and cropped black puffer came from one line of notes.
- Your own hair. The original dancer's cap, headphones and long hair are smoothed out of the depth map, so they stay out of your video.
A clear, well-lit, full-length photo works best. If you write your own outfit in the notes, keep the legs wide: the dance's biggest shapes come from those pants, and a slim pair has nothing to fill them with.
The Fastest Way
Kakalika Dance
One photo, one take: three big kicks and a point at the sky. 23 seconds.
One photo, 250 credits (about $12.50), no prompt. It renders on Seedance 2.5 at 480p, vertical 9:16, 23 seconds, with the song on top.
Build It Yourself
1. Turn the clip into a depth map. Run the clip through a depth-estimation model (Depth Anything V2 works). The depth map keeps the moves and drops the face, the clothes and the colour, so the original dancer doesn't show up in your video. Smooth the cap and headphones into a plain rounded head, and trim the long hair back to the head.
2. Make one character sheet. For a continuous take, use a sheet that shows what the person looks like instead of a storyboard of moments. Storyboard panels can pull the video toward their framing and cause jumps. Put the person in full outfit in the set at the top, and a four-view turnaround below.
A photorealistic still on a flat patch of dry tan dirt in an open park on a clear sunny late afternoon, autumn trees along the distant horizon and a huge deep-blue sky with thin streaks of white cloud. The person from reference image 1, full figure from head to feet, standing with the feet apart and facing the camera, seen from a phone resting low on the ground and looking up, in their own top and shoes with very wide baggy cargo pants that pool over the shoes. No text, no logos.
3. Drive it with the depth map. Feed the sheet, your photo and the depth video to a model that copies motion from a reference video (we recommend Seedance 2.5). Then write what it can't read from a depth map:
- Sources: the sheet decides what you look like and wear, and the depth map decides only how you move.
- The take: a timed list of the kicks, the lean and the arm moves, and a fixed camera that never moves.
- The face: a relaxed smile and the grin at 15 seconds, because small faces carry no expression through a depth map.
Lay the song back on top at the end: Seedance can drop or change music on its own, so mux the track yourself.
Common Mistakes
Driving with the raw clip. The original dancer's face and clothes come straight through. Use the depth map.
Fighting the pants. A depth map of huge cargo pants turns slim jeans into something baggy anyway. Give the subject wide pants in the description so the sheet and the shapes agree.
A storyboard for a one-take. Panels of key moments can make the video jump to a panel's framing. A character sheet has no moments to jump to.
Asking for lip sync. The original dancer isn't lip-syncing the song, so a depth map gives the model no mouth to copy. Say plainly that nobody sings, or the mouths flap at random.
Normalizing a hot master without a limiter. The song's file peaks above 0 dB. Set it to -14 LUFS with loudness normalization alone and the peaks stay too high. Add a limiter first.
Related Reading
- How to Make the Wipe Me Down AI Dance Video: another one-take dance from a phone propped low, with a friend
- How to Make the Lonyo AI Dance Video: lead a whole crew, filmed from above
- The Jawn Dance (Jerri): How to Make the AI Version From One Photo: a dance copied move for move from one photo
- How to Make the 'I Need A Girl' AI Dance Video With a Friend: two friends and one take in a parking lot at night
- Viral AI Video Trends 2026: every trend we've turned into a template this year
