Why costumes work in this template
The Come Back and Back template has no fixed costume. Each dancer wears whatever the photo shows, so a photo of you in costume puts the costume in the video, through every bounce, kick and arm swing.
Video trend
In the second half of October 2026 couples were posting to the Come Back and Back sound in paired Halloween costumes, captioned with the characters they dressed as. Because the Oakgen template takes each dancer's outfit from the photo, you can make the same kind of video from two costume photos without filming a single step.
The Come Back and Back template has no fixed costume. Each dancer wears whatever the photo shows, so a photo of you in costume puts the costume in the video, through every bounce, kick and arm swing.
The couple posts we saw on the sound used classic pairs, such as a gothic husband and wife and an animated bride and groom. Any pair that reads at a glance works, because the dance itself is identical for both dancers. Pick costumes that look clearly different from each other so the two never blur.
Take one full-length photo per person, head to shoes, face uncovered, in even light. Masks, veils and face paint that hides the face make the final step toward the camera, where faces get bigger, hard to get right; a costume with a visible face is safer. Capes and long dresses are fine, but expect some movement detail to be invented.
The Left dancer ends slightly in front when both step toward the camera, so put the costume you most want to show off, a crown, big hair or a dramatic collar, in the Left slot. The Right dancer stays on the right the whole time.
Watch that the costumes stay the same from start to finish, that nothing from one costume jumps to the other dancer, that the crossed arms at about 10 seconds stay separate, and that both faces hold up in the final step-in.
Name the characters in the caption, the way the couple posts do, add "come back and back" and the sound title, and label it as AI-made. Use only photos of people who agreed to appear.
Add a photo for the left dancer and right dancer.



Upload a clear photo for the Left dancer and the Right dancer. A full-length photo, face to shoes, works best. Use only photos of people who agreed to appear.
Choose 480p, 720p or 1080p and review the quote. It includes the full video, both photo preparations and three scene references.
Watch that the two faces never swap, the feet land on the beat and both faces hold up as you step toward the camera at the end, then download your portrait MP4 with the original sound.
Each photo casts one dancer: face, hair, build and the outfit in the photo. The left dancer stays on the left and the right dancer on the right for the whole take.
All 20 seconds from one low camera: the bouncing jersey club steps, the hand rolls, the crossed-arm pose, the wide arm swings and the step toward the camera at the end, on the original sound.
There is no costume. Each dancer wears the outfit from their photo, so full-length photos give the most faithful result.
The preview is the reference dance clip the template follows. It is the motion and timing reference, not an Oakgen result.
It is a portrait AI clip of about 20 seconds in which two people dance the same jersey club routine side by side to the slowed sound "Wyw my sweetie #JerseyClub - Slowed": bouncing footwork, hand rolls at the chest, a crossed-arm pose, wide arm swings and a final step toward a camera on the ground. Oakgen's template puts you and a friend in the two dancers' places and keeps the alley, the moves, the timing and the sound.
Because many TikTok posts of this dance are captioned "come back and back", and Google suggests the phrase together with "dance trend" and "dance tutorial". The sound itself is titled "Wyw my sweetie", so you will see both names; they are the same trend.
"Wyw my sweetie #JerseyClub - Slowed" by nissi & 25Hzz. It is the slowed version from their EP "Wyw my sweetie #JerseyClub", released on 28 August 2026, which also has the original-speed track. Your video keeps the sound from the reference clip.
Yes. The two dancers in the reference are real people, so each role needs a photo and neither reference face is ever kept.
The Left dancer stays on the left of the frame and ends slightly in front when both step toward the camera; the Right dancer stays on the right. They never swap sides.
What you wear in your photos. There is no template costume, so full-length photos show the AI your real outfits; with a head-and-shoulders photo the AI has to fill in the rest of the outfit.
Your own, or photos of people who agreed to be in the video. Do not upload photos of celebrities, public figures or anyone who has not consented, and label AI videos clearly when you share them.
No. Nobody lip-syncs; mouths stay closed, relaxed or smiling while the sound plays, as in the reference.
About 20 seconds in portrait 9:16, one fixed low camera from the first bounce to the final step toward the camera.
No. It is the reference dance clip the template follows, not an Oakgen result. Review both faces, hands and feet before sharing.
4,132 credits per 720p video. The quote includes both photo preparations, three scene references and the complete video. Check it before generating.
Plan a story, keep your character recognizable, and make the finished clip your own.
Review your finished Come Back and Back AI video by timestamp: no swapped faces, feet on the beat, separate crossed arms and both faces in the final step-in.
Make the Come Back and Back duo dance with AI, step by step: pick two photos, choose who dances on the left, read the quote, check the result and post it.
Upload 2 photos. Everything else is set.