The Trouble Maker AI dance is the two-person partner routine to Trouble Maker by HyunA and Hyunseung, remade from photos instead of filmed. On videosi you upload two photos, one per dancer, and get a 15-second duet with the trend sound mixed in for 13 credits at standard 720p.
This guide goes deeper than the FAQ on the trend page. It covers the six beats, which photo lands on which side, why two faces can drift in the dips, what the AI version keeps fixed and how to post the result.
Key takeaways
- The routine has six beats in 15 seconds: walk-in, pose, dip and lean, shoulder swivel, wide-arm sway and a back-to-back finish.
- Photo 1 is the left dancer and photo 2 is the right dancer, and either spot can be anyone.
- The Trouble Maker sound is mixed in, so the clip is ready to post.
- Place, outfits and choreography stay as in the reference; only the people change.
- A standard 720p video costs 13 credits, and packs start at $6.99 for 10 credits.
In this guide
- What happens in the 15 seconds, beat by beat?
- Where does the song and the trend come from?
- Which photo becomes which dancer?
- Can I use one photo of both people instead of two?
- Why do the two faces sometimes look alike or drift?
- What kind of photo works for a head to toe duet?
- Is the trend sound included, and is it a lip-sync?
- Can I change the choreography, the place or the outfits?
- How does it work under the hood, without sharing the prompt?
- How do I post it and what should I label?
- What does it cost, and is the second dancer extra?
What happens in the 15 seconds, beat by beat?
The routine has six beats. In the first two seconds the two dancers stride toward each other from opposite sides of the frame. From 2 to 4 seconds they stand together, one hand raised and knees bent, and hit the first pose.
The middle is the part that makes it a partner dance. Between 4 and 7 seconds the left dancer dips the right dancer back and they lean into each other. From 7 to 10 seconds they swivel shoulder to shoulder and trade places on the beat. From 10 to 13 seconds they face each other and sway with arms stretched wide, and in the last two seconds they end back to back, looking over their shoulders.
The AI version keeps all six beats in order, in the same sunny old-town square, so what you get is the viral routine with your people in it, not a new choreography.
Where does the song and the trend come from?
Trouble Maker is a duet by HyunA and Hyunseung, released on 1 December 2011. The dance trend is named after it. The routine keeps coming back: the earliest duo cover we found is dated 24 August 2023, and the most-watched one, a duo in an old-town square, was posted on 16 October 2023 with about 4.9 million views.
The first duo routine is not identified, so we do not credit one creator. What we can say is that new covers and AI dance clips set to the song were still being posted on 30 October and 3 November 2026, which is why the page is live now.
Which photo becomes which dancer?
Photo 1 becomes the dancer on the left and photo 2 becomes the dancer on the right. The positions come from the reference routine, so they do not swap by themselves. If you want to appear on the right, upload your photo second.
The original cover has a man on the left and a woman on the right, but the spots are not tied to gender. Two women, two men, a couple, two friends or two siblings all work, and each dancer keeps the face, hair and outfit from their own photo. The Trouble Maker AI dance page shows the same layout in its examples.
Can I use one photo of both people instead of two?
Yes. If you only have a single picture of the two of you, the page accepts it. A clear photo of you both side by side, both faces sharp and facing the camera, works best.
Two separate photos are still the safer choice. With one photo, the model has to decide which face goes on which dancer, and it can get that wrong. With two photos each face has its own source, so the faces are much less likely to mix.
Why do the two faces sometimes look alike or drift?
The dips and leans are the hardest seconds. The dancers tilt, turn and come close, and the model has to keep two identities apart while their heads are nearly touching. That is when one face can drift toward the other.
Likeness is not guaranteed, and the page says so. The fixes are practical: use one person per photo, keep both faces front-facing and sharp, avoid sunglasses and heavy filters, and retry with a different photo if the result is off. The video keeps your hair and outfit from the photo, so a plain, evenly lit shot helps more than a stylish one.
What kind of photo works for a head to toe duet?
The dancers are seen head to toe, so a photo from head to hips or head to toe gives the best likeness. A close selfie works for the face but gives the model less to go on for build and clothes.
Soft daylight, hair visible and no group crops are the basics. What you wear in the photo is what you dance in, because there is no choice of outfit or place. Use adults only, and if the second person is not you, get their permission before you upload and again before you post.
Is the trend sound included, and is it a lip-sync?
The sound is included. The model renders a silent video, and the Trouble Maker audio from the reference is mixed in afterwards, so the 15-second clip is ready to post as is. Your own sound does not need to be added in an editor.
It is a dance video, not a lip-sync. The dancers move to the beat and are not asked to mouth the lyrics, which also keeps mouths from looking odd in the close moments.
Can I change the choreography, the place or the outfits?
No, and that is deliberate. The moves follow a 15-second motion reference of the real duet closely, which is how the walk-in, dip and shoulder swivel land on the beat. The place stays the sunny old-town square and the outfits come from your photos.
The reason is what happens without a reference. A typed description alone gives a generic couple dance, with no dip and no swivel on the beat, and the pair can drift to one or three dancers. Locking the motion is what keeps the result recognisable as the Trouble Maker dance.
How does it work under the hood, without sharing the prompt?
videosi uses ByteDance Seedance 2.0 or 2.5, depending on the tier you choose. The input is your two photos plus the motion reference, and the output is a 15-second video in 9:16, 1:1 or 16:9.
The instruction is built from five parts: a replace line saying which reference dancer becomes Image 1 and which becomes Image 2, an identity lock that makes each photo the only source for its dancer, a timeline of six short beats, a few rules such as exactly two adult dancers and five fingers per hand, and the audio step. We do not publish the full prompt, but the trend page lists the same parts under How it is made.
How do I post it and what should I label?
Pick 9:16 for TikTok, Reels and Shorts, or 1:1 and 16:9 for other feeds. The free preview is watermarked, so a clean download is what you post. Add the AI label when you publish on TikTok or Reels, since the clip shows real-looking people dancing.
Posting it as a duet idea tends to work best with a caption that names the partner, because the whole point is the pair. If you made it of a friend or sibling, tell them first.
What does it cost, and is the second dancer extra?
A standard 720p, 15-second video costs 13 credits, and the second dancer does not cost extra. Credit packs start at $6.99 for 10 credits and $17.99 for 30 credits, credits never expire, and a failed render returns its credits.
Because close partner moves sometimes need a second try, it is worth planning for a retry with a different photo. If you want a solo routine instead, the AI dance page turns one full-body photo of an adult into a short dance.
Ready to try it? Open the Trouble Maker AI dance page, add your two photos and generate the duet.
