The beat and the routine are the reference. Your portrait brings a new performer into the frame. Watch our real test side by side, then upload your portrait to generate a full 19-second video.
Upload one portrait. Keep the routine. Make it your look.
Credits are reserved when generation starts and returned if it fails. The free duo trial does not cover this template.
Dominican artist Bulin 47 joined Kevin Wai for an impromptu dembow freestyle outside a restaurant in Spain. The playful dance and conversational delivery became as recognizable as the music. AI edits then placed new characters into that performance, making the same gestures feel funny or surprising with a different face.
Think of it as recasting a short scene. A portrait supplies identity; the reference video supplies the performance. The appeal comes from recognizing the routine immediately and seeing an unexpected lead, rather than hearing a completely new song every time.
Watch Kevin Wai’s source post on Instagram ↗The reference supplies the body rhythm, hand gestures, crowd interactions and handheld camera path. Timing matters: the new performer needs to be at the same point in the routine when the original vocal reaches each line.
The portrait guides the face, hair, glasses and outfit. Our sample keeps the replacement recognizable through the later turns, but it is a test of an AI edit, not a promise of a flawless face swap.
Start with one clear portrait showing the face, hair and upper body. Include the outfit you want the lead to wear. Good lighting and a visible mouth give the model more useful detail than a filtered close-up.
Choose Standard 480p for 70 credits or HD 720p for 150 credits. Sign in, confirm permission and click Generate. The fixed 19-second reference guides the routine and camera; you do not need to upload a reference video or write a prompt.
StagePair adds the matching original audio automatically. Play your finished video beside the reference, then download the MP4. It is also saved in My videos. Review mouth timing and identity through the full clip before sharing.
Use one front-facing photo per performer, with good light and room around the shoulders. Avoid hidden mouths, sunglasses and tight face crops. Use people who have agreed to appear.
Start with Street Cypher + Lively. The preset below selects Standard, 720p, 15 seconds and vertical framing. Choose Friends or Couple and add a short personal theme if you want one.
Check faces, hands, mouth timing and the full audio before sharing. Download a result you like from My videos. A fresh generation is a new creative attempt and consumes credits.
Opening the setup does not start a paid generation. Check the credit total in the generator before submitting. Compare credit costs →
The featured performer is recast using a portrait. The reference guides the gesture sequence, framing, camera movement and visible mouth timing. Our current test keeps the supporting cast in place. This is a different workflow from writing and generating a new rap.
Yes. The comparison shows our 19-second, 480p Seedance 2.5 test, with its original reference beside it. The generated picture was silent; the matching source audio was added afterward. This sample is not a 720p render, and it does not guarantee the same result for every portrait.
Yes. Upload one portrait, choose 480p or 720p, confirm photo permission and generate directly here. Sign in first. If your balance is too low, the button takes you to credit packs. Completed videos are saved in My videos for playback and download.
The cost is 70 StagePair credits at 480p or 150 at 720p, using Seedance 2.5 and a 19-second reference. These are product credits, not the provider’s credits. The cost includes reference-video processing; the free five-second duo trial does not cover this workflow.
The matching soundtrack is attached automatically after visual generation, rather than asking the model to sing it again. That keeps the audio itself unchanged. Mouth timing can still drift in the generated picture, so watch the full result with sound. Reuse source footage and audio only with the necessary permission.
Use one sharp, well-lit portrait with an unobstructed face, visible hair and upper body. Include the clothes and accessories you want to see. Side views, tiny jewelry, fast turns and foreground occlusions are harder to preserve. Our test still carries over a few details from the source performer.
Prefer a shared microphone and a studio backdrop? Explore the Hotel Lobby AI format. For a rhythm without lyrics, try the Beatbox session. Both offer a different way to turn your own photos into a performance.
StagePair is independent of Bulin 47, Kevin Wai and their labels. This page shows an independent test of a public trend and is not an official artist product.
StagePair