Gas Station Dance Meme: The Original Video, Song and AI Remakes
Where the three-person dance came from, who appears in the original, and how to tell a character swap from a newly generated video.
Last updated: 2026-10-08
The gas station dance is a three-person video format: one person dances, another films, and a third gets out of a car. AI remakes change the cast or generate a similar scene. If you want your friends in the version you remember, first check which of these methods a tool uses.
Where did the original video come from?
Memepedia's September 2026 report traces it to a video posted by Adel Veigel on July 1, 2024. Its account names Ravshan Dzhulpaev as the dancer, Mikhail Litvin as the person filming, and Adel as the person getting out of the car. It identifies the music as Akcent's On And On (When the Lights Go Down).
The report links to Adel's Instagram reel and describes a new wave of AI remakes in September 2026. The Russian name “Делай” comes from a shout in the source video. T—J's September 25 article about AI video editing also connects the trend to the three creators' 2024 video.
The original date and song are reported by Memepedia; we have not independently verified the Instagram post. The sources do not establish the first AI remake or the trend's total reach.
Why does changing the cast work so well?
Our reading of the format is that it gives the viewer a recognizable setup and a small reveal. The opening dancer establishes the scene. The person filming makes it feel like a friend recorded it. The car door brings in someone else.
Keeping that structure while changing the people creates a simple joke: the viewer knows what is coming, but not who will appear. A three-person cast also makes it easy to assign parts within a friend group. One person gets the opening, another appears on camera, and someone else gets the entrance.
This is our interpretation of the format. A clip can work as a private joke without becoming a widely shared post.
“Same video” can mean three different things
Editing a finished clip changes captions, length or soundtrack. A CapCut-style project can help with editing, but a timeline alone does not put three new people into the footage.
Replacing characters in a reference video uses the source clip as an input. The important question is what survives the edit: camera movement, body motion, background, clothing and audio. Do not assume that every “face swap” replaces the whole person.
Generating a new scene from photos makes another video with a similar story. It can follow the dancer → selfie → car-exit → returning-selfie sequence without copying the original choreography.
Choose the method that matches your goal: editing a clip, preserving reference motion, or making a new scene.
Which one does this website demonstrate?

The car-door reveal in our generated example, not a frame from the original creator's video.
The October 5 complete input-to-output example used three fictional adult portraits without an original dance-video input. It is a newly generated scene with generated audio. This historical test does not represent the current reference-guided workflow.
Upload 1–3 photos to choose who dances, takes the selfie and gets out of the car. Empty roles keep the default characters. The fixed reference and scene instructions are built in; no prompt writing or editing is needed. The current output setting is 15 seconds. See the reference-guided trial.
To make your own version, follow the creation guide with 1–3 photos. If exact original choreography or music is essential, inspect the full reference-guided trial and its audio before deciding whether it meets your needs.
Questions before making your Gas Station Dance video
Can I make the Gas Station Dance video with only one photo?
Yes. Upload one clear solo photo to the dancer, selfie or car-exit role. Leave the other roles empty to use default characters. You do not need to write a prompt or film a dance.
Will it keep the same dance moves, camera turns and original song?
The fixed video reference helps recreate the familiar dance moves, camera turns and music while the photos guide your cast’s appearance. Watch the complete generated example to compare the people, motion and soundtrack. Some visual or audio details may vary slightly. See the reference-guided trial
What should I check before choosing a tool?
Watch a full result, not just the opening frame. Check whether the same person returns at the end, whether the car-exit character is recognizable, and whether the sound is the track you expected. Look for an input/output pair so you can judge resemblance yourself.
For a friend-group clip, choose the roles before uploading. For a clip that must reproduce the original, ask whether the tool actually takes that reference video as input. The label “gas station dance AI” does not answer that question.
Research checked October 5, 2026. Found an error in the background or a better primary source? Send us the link.