DanseAI is an AI video generation tool built around the "Alors on Danse" white studio keyboard meme. Its purpose is clear: use two photos of people to replace the two performers in the reference clip, automatically reenacting the white studio scene, keyboard playing, crowd-hyping actions, and close-up shots, and pairing it with the soundtrack from the original reference clip to output a short video ready for TikTok, Reels, or Shorts.
Main capabilities
- Photo character mapping: Photo 1 replaces the performer on the left standing by the table playing the mini keyboard and laptop; Photo 2 replaces the hype partner on the right who points, raises their hands to the beat, and then leans toward the camera for a close-up. You can click Swap photos to swap the roles.
- Parameter selection: Duration can be 5, 10, 15, or 30 seconds; aspect ratio can be vertical 9:16 or horizontal 16:9; resolution up to 1080p. The default settings are Seedance 2.0 Mini, 10 seconds, vertical 720p.
- Model selection: Seedance 2.0 Mini (default, most credit-efficient, capped at 15 seconds, weaker likeness preservation), Seedance 2.0 Fast (30 seconds is stitched from two 15-second segments, capped at 720p), Seedance 2.0 (capped at 1080p), Seedance 2.5 (best likeness, checks both photos and the reference clip first, 30 seconds can be rendered in a single pass), MiniMax H3 (balances likeness and price, only 720p and 1080p).
- Soundtrack: The generated model outputs a silent video, and a final finishing step then edits the soundtrack that comes with the reference clip into it according to the selected duration, so the downloaded MP4 has sound. It does not generate a new song or clone voices; the music is Stromae's commercial recording, and platforms may treat it as an ordinary clip containing commercial music.
- Reference clip: The template is the vertical Kratos and Zeus version of the white studio scene; the 5, 10, and 15 second options are the beginning of the same clip, while 30 seconds adds the ending where the two walk down from the studio and linger for about one second.
How to use
- Upload two photos: Put the keyboard player in Photo 1 and the hype partner in Photo 2, with one person per photo.
- Choose duration, model, and resolution: You can choose 5, 10, 15, or 30 seconds, as well as Seedance or MiniMax H3 models and the corresponding resolution. The credit consumption is updated and displayed before generation.
- Generate, check, and download: Most videos take a few minutes, and you can leavepage, and when finished, download the MP4 with music in the history. Photo requirements: JPG, PNG, or WebP, max 20 MB each, short side at least 300 px, only one person per photo. It is recommended to put the more expressive face in Photo 2, because that character has a large close-up, and a clear frontal photo is most important; a half-body photo can provide clothing information, while a tightly cropped face means the clothing can only be guessed by the model. Only upload photos of people who consent to appearing, and label them as AI-generated when posting.
Pricing and Credits Generation uses on-site credits, and the total price depends on duration, model, and resolution. It is shown before confirmation, and one deduction is made per completed video. A 5-second video starts at 6 credits. Eligible new accounts currently receive 3 credits, which can be used for Seedance 2.0 Mini; other models require purchasing credits. Signup credits expire after 7 days. Videos from free accounts have a moving watermark; any purchase unlocks saved watermark-free originals, including already generated videos. Deducted credits are refunded when a generation task fails or photos are not accepted, but completed videos may still have visual flaws, so check before posting.
Use Cases
- Couples, friends, or family recording short videos together, casting the two people into the roles of keyboardist and hype partner.
- Content creators who want to combine any two people, such as a tech CEO or film and TV characters, to reenact the meme.
Limitations and Notes
- Exact likeness is not guaranteed; faces, hairstyles, and clothing may change, especially in close-ups and cuts, and the two identities are occasionally mixed. On models other than Seedance 2.5, 30 seconds is stitched from two 15-second segments, and faces may drift slightly at the seam.
- Generation usually takes a few minutes, and longer for 30 seconds or during busy periods.
- This is an independent video tool, not affiliated with Stromae, Jamel Debbouze, their record labels, or the reference remakers, nor with TikTok or CapCut.
- There is no real-time filter version, and it is not face swap: the model regenerates the entire performance from the photos, including body, clothing, and camera movement.
- The site's default prices are for reference only, and the exact total price is shown before generation; specific price information for each package and credit pack is not provided with specific amounts.
Background This format originates from Jamel Debbouze's Made in Jamel comedy sketch: Jamel walks into Stromae's white studio and pretends to direct him in writing "Alors on danse." It is often taken as behind-the-scenes footage of the song's production, but Stromae wrote it alone, and the sketch is just a joke. AI remakes swap new combinations into the two roles, making the trend go viral on TikTok. Titled in French, meaning "so we dance"; Stromae is Belgian, and the song was released in 2009.
