The alors on danse ai trend is a silent 12-second video in which one person plays a small keyboard in a white studio while a second person, or a pet, dances beside them and ends with a huge grin into the camera. videosi (videosi.io) makes it from two photos, or one photo of you both, so you choose the photos and the format and never write a prompt.
Key takeaways
- Photo 1 is always the keyboard player and photo 2 is always the dancing hype-man, so the role order is the main creative choice you make.
- The clip is silent. Add Stromae's Alors on danse from the platform's music library when you post.
- Use only yourself, adults who agreed, or your own pets. Photos of Stromae, Jamel Debbouze or any public figure are not accepted.
- A standard 12-second video (Standard model, 720p) costs 10 credits. There is no free generation, and a failed render returns its credits.
- Most failures come from the photos, not the tool: tight crops, sunglasses, a third person in frame or a dark image.
In this guide
- How do I decide who plays the keyboard and who dances?
- What actually happens in the 12 seconds?
- Why a white studio and a tiny keyboard?
- Where does the scene come from, and is it a real dance trend?
- Which photos work best?
- How do I shoot the photos for a pet version?
- Why does a general video model often fail at this?
- How do I add the song when I post?
- What about copyright, celebrities and labels?
- What does one finished video cost me?
How do I decide who plays the keyboard and who dances?
Think of the video as a two-person comedy sketch with a straight man and a show-off. The keyboard player stays small and focused for most of the clip: head down, hands on the keys, nodding. The hype-man does everything funny. They throw an arm up, lean over the keys, point, snap their fingers, roll a hand in the air and then swing toward the lens for the final close-up.
That close-up is the reason the order matters. Whoever is in photo 2 fills the right side of the frame for the last two seconds, so give that slot to the person with the best grin, the most expressive face or the pet with the most character. The player only needs to read clearly as a calm, concentrated person who ends with a thumbs-up.
If you only have one picture of both of you, switch the Alors on danse AI page to its one-photo mode and stand the keyboard player on the left. Two separate photos usually keep each face cleaner, because each character has its own source.
What actually happens in the 12 seconds?
The video has six beats, and knowing them helps you choose photos that survive the movement.
- 0 to 1 s, arm up: the player is on the keys, the hype-man throws one arm over their head.
- 1 to 4 s, producing: the hype-man leans over the keyboard, pointing and directing with quick flicks.
- 4 to 5.5 s, snap: they straighten up, snap their fingers and bounce on the beat.
- 5.5 to 9 s, hand roll: a rolling hand in the air and a proud, knowing face.
- 9 to 10 s, turn to camera: the hype-man swings toward the lens.
- 10 to 12 s, grin and thumbs-up: a big close-up, with the player small behind them.
The most demanding moment is the lunge at 9 seconds. The face goes from medium-shot size to filling a large part of the frame, so a sharp, front-facing source photo pays off exactly there. Mouths stay closed or smiling throughout, because nobody sings or talks in the clip.
Why a white studio and a tiny keyboard?
The set is deliberately plain: a seamless white photo studio with a soft grey floor curve, a small white plinth, a compact black-and-white 25-key MIDI keyboard and an open silver laptop with no logo. That is the visual language of the joke, a producer's desk reduced to its simplest props, and it is also practical.
A blank background gives the model nothing to confuse with the two characters, so faces stay stable and the motion reads clearly. A small keyboard also limits how much finger detail the model has to invent. Hands on keys are where AI video shows its weakest work, so a compact instrument, a fixed hand position and a rule of five fingers per hand keep the result believable.
Where does the scene come from, and is it a real dance trend?
It is a comedy remake, not a choreography challenge. The scene recreates a sketch from Made in Jamel, the Jamel Debbouze DVD released on 1 December 2010, in which he pretends to help Stromae write Alors on danse. The Wikipedia article on Jamel Debbouze confirms the DVD and its release date and lists Stromae among the performers. The joke is that Stromae wrote and produced the song himself.
The song is older than the sketch. According to the Wikipedia page for Alors on danse, it was written by the Belgian artist Stromae, released in Belgium in September 2009 and in the rest of Europe in February 2010. The AI remakes that swap in other people and pets began to appear on TikTok in early October 2026. We have not confirmed who made the first one, so treat any claim about an original author with caution.
Which photos work best?
Four habits make the biggest difference.
- Show the upper body. Both characters are seen from the waist up, so shoulders, arms and clothes in the photo keep the outfit and build consistent.
- Use clear, front-facing faces in even light. Sunglasses, a hat that covers the face or a tight crop around the eyes make the likeness drift once the camera pushes in.
- One person per photo when using two uploads. A group photo can lead the model to add a third character, and the clip has exactly two.
- Avoid dark or blurry images. Soft light from a window works better than a flash or a nightclub shot.
Likeness is never guaranteed in a 12-second generated clip, but good inputs raise the odds a lot. If the first result misses, change the photo before you change anything else.
How do I shoot the photos for a pet version?
Pets were among the first versions people shared, and they work well because a dog or cat standing upright at a keyboard is funny on its own. The goal is to keep the animal the same animal: same species, breed, size, coat colour and markings, with natural paws and a natural face.
Photograph the whole animal, sharp and well lit, with the face and markings visible. A sitting-up pose is ideal because it suggests the upright stance the clip needs. Do not crop to the face only, since the model then has no body to build from.
Pairing is part of the humour. A dog as keyboard player and a cat as hype-man, or a person with a pet as the showman, gives the final close-up a lot of personality. Just remember the pet must be your own, since the photo rules are about consent and ownership of what you upload.
Why does a general video model often fail at this?
You can try to build this by hand in a general-purpose video tool, but several things tend to go wrong at once. Without a motion reference, the hype-man just sways, and the pointing, the hand roll and the lunge into the camera lose their timing. With two people standing close behind a keyboard, models often swap the faces, blend features or invent a third person. Hands on keys sprout extra fingers unless the instructions pin them down. And every attempt costs credits and several minutes, so a handful of tries adds up.
The videosi page removes those variables. It uses ByteDance's Seedance video model, a built-in motion reference that supplies the gestures, timing and camera push, and a written prompt that locks each face to its own photo, sets the white studio and forbids text, logos and cuts. You supply two photos and press Generate. That is the whole difference, and it is why the same idea feels easy on the Alors on danse AI generator and fiddly elsewhere.
How do I add the song when I post?
The finished MP4 is silent, with no speech, no music and no sound effects. That is intentional: the audio belongs to the platform, not to the generated file.
Open the clip in TikTok, Reels, Shorts or CapCut, and add Alors on danse by Stromae from the built-in sound library, or reuse the sound from a clip you liked. Most posts use the original song, often the Extended Mix or the Radio Edit. Drag the section you want so the louder moments land on the snap at 4 seconds and the grin at 10 seconds. Then switch on the platform's AI-generated label, which is the honest thing to do for synthetic video and keeps your post within the platform rules.
What about copyright, celebrities and labels?
There are three separate questions here, and they have separate answers.
First, the song. Alors on danse is a copyrighted recording and composition. Using the platform's licensed music library is the safe route, because the platform handles the licence for in-app use.
Second, the people. videosi does not accept photos of celebrities, public figures or minors, so you cannot place Stromae or Jamel Debbouze in your video. Use yourself, an adult friend, partner or sibling who agreed, or your own pets. Users must be 18 or older, and photos and results are deleted within 30 days.
Third, the brand. videosi is independent and is not affiliated with TikTok, CapCut, Stromae or Jamel Debbouze. The sketch is a reference for the format, and your video should be presented as your own AI remake, with the AI label on.
What does one finished video cost me?
There is no free generation and no free trial. One standard 12-second video (Standard model, 720p) costs 10 credits. A Single pack is $6.99 for 10 credits, which covers exactly one standard video, and monthly plans start at $29.90 for Basic. Better models and higher resolutions cost more credits per clip.
If a render fails, times out or is blocked, the credits come back automatically, so you only pay for a clip you receive. You can start from the Alors on danse AI video page whenever you have two good photos ready.
