Three photos, three entrances in order: a man dances alone on the forecourt, a second man films a handheld selfie, then a woman steps out of the car to close. No location shoot, no prompt.
PNG / JPG / WebP · up to 10 MB · at least 512px on the short side
0/3 photos
Three photos in order of appearance: 1st, the man who dances alone; 2nd, the man filming a selfie; 3rd, the woman who steps out of the car. Full-body photos suit 1st and 3rd.
Video settings
No prompt needed. Only upload photos of yourself or people who agreed to be featured; celebrity photos and film stills aren't allowed.
Three people, in order of appearance
It's an AI remake of a viral three-shot clip from an empty gas station forecourt under a golden-hour sky. Three people appear one after another: a man dances alone, a second man films a handheld selfie, and a woman steps out of a car. Each of your three photos takes one of those entrances, in the same order, while the template keeps the place, the camera and the cuts. It has nothing to do with “gas station AI” software for pumps, payments or station cameras.
1st on screen · Photo 1
The clip opens on one man alone on the asphalt in a white polo and light jeans: a few swaying steps, then both arms thrown at the sky while the camera circles. It's the widest of the three shots, so the dancer is seen head to toe, and his face, build and outfit all come from photo 1.
2nd on screen · Photo 2
The camera whips round to a second man in a black bucket hat and black tee: one arm stretched toward the lens, a grin and a hand sign, with the gas station canopy and a low sun behind. His face fills the frame for a couple of seconds, so photo 2 needs a clear face more than a full outfit.
3rd on screen · Photo 3
After a hard cut, the finale: the door of a grey coupe opens and a woman with messy dark hair in a black track jacket steps out, leans into a close-up and walks toward the camera between the cars. The shot moves from full body to face and back, so photo 3 works best as a standing photo with the face clearly visible.
Doing the Gas Station Dance for real means getting three people, a car and a quiet forecourt together in the right light, then cutting three shots. The template only needs three photos, and your cast can live in three different cities.
The real thing, with your own moves, if you can get everyone to the pumps.
Quick to try, but it’s still AI: faces and moves can drift between runs, so label it when you post.
Upload your three photosThe three people are filmed differently, from head-to-toe dancing to a face-filling selfie. Pick each photo for the shot it plays, in order of appearance, with one person per photo.
1st Dancer: full body, standing
Head to shoes in frame, arms free. Shot 1 is wide, so the outfit and build matter as much as the face.
2nd Selfie: a clear, front-facing face
Both eyes visible, looking near the lens, in even light. The selfie is a close-up, so the face carries the shot.
3rd Car exit: standing, knees up or full length
Shot 3 goes from stepping out of the car to a close-up, so show the face and most of the body.
One photo of all three of you
A group photo leaves the model guessing who plays which role. Crop each person into a photo of their own.
A phone or sunglasses over the face
Mirror selfies and dark glasses hide the features that the selfie and the closing close-up depend on.
Dark, blurry or tiny images
Night shots and heavy filters come out soft or unlike the person. Use photos at least 512px on the short side.
Playing every role yourself? Put the same photo in all three slots. It works, but one face across three shots tends to vary more than three different people.
How it works
Three steps from photos to a clip you can download. Nothing is charged until you press Generate.
Add one photo each to 1st Dancer, 2nd Selfie and 3rd Car exit, in the order they appear. To change who plays what, remove a photo and add it to another slot.
The default is 15 seconds in 9:16 at 768p, enough room for all three shots. For a cheaper test of your casting, lower the resolution rather than the length; the price on Generate updates as you go.
Press Generate; there’s no prompt to write. When the video is ready, download it, and add the song from your app’s music library if you want the original track.
Before you post
Your version puts real faces into a video nobody filmed. Four checks keep it a joke everyone is in on.
Everyone in the video should agree before their photo goes in. Cast adults who can say yes for themselves: no strangers, no celebrities and no photos of children.
Switch on your platform’s AI-generated label or say so in the caption. Many platforms ask for it on realistic AI video and rules differ by country, so viewers shouldn’t have to guess.
Your video comes with generated audio, not the song most posts credit as Akcent’s “On & On (Stay With Me)”. Oeneo AI has no licence for that track; add it from the TikTok, Instagram or YouTube music library when you post.
Videos made on a free account are public in the Oeneo AI gallery; on a paid plan they are private by default. Decide before you upload a friend’s face.
Every Gas Station Dance AI video takes three photos and costs the price shown on the Generate button; the values below come from the live generator.
Every template works the same way: add your photos, press Generate and get a shareable video without writing a prompt.
View all effectsFAQ
Browsing, adding photos and checking prices are free. Each video costs 268 to 954 credits depending on length and resolution, and 358 credits for the default 15-second 768p clip. New accounts get 10 free credits, which isn’t enough for a full video on its own.
Trend guides and reposts credit Russian blogger Mikhail Litvin and his friends Ravshan and Adel, who reportedly filmed it at a roadside gas station on a summer 2024 road trip; in Russian it spread as “Delai!” (“Do it!”). The first post is hard to trace, so treat this as the widely repeated story rather than a verified fact. AI versions took off in September 2026.
Neither. It’s an independent, fan-made AI template inspired by the trend, not affiliated with or endorsed by Mikhail Litvin, Ravshan, Adel, Akcent or TikTok. Your video is newly generated from your photos, not an edit of the original footage.
No. The model generates new audio for each clip, and it won’t be Akcent’s recording. If you want to post with the real track, add it through your platform’s music library.
Yes. All three slots need a photo, but the same photo can go in more than one: one person can play every role, or two friends can split the three shots. Expect more variation between shots when one face appears twice, and results differ from run to run.
Wide shots, fast arm movements and the jump from full body to close-up make faces and outfits more likely to drift, and small, side-on, group or covered photos make it worse. One clear, evenly lit photo per person, full length for the 1st dancer and the 3rd woman from the car, gives the most recognizable result, but every run is a little different.
Not quite. A face swap pastes a new face onto the original footage, so the original bodies and outfits stay. Here the model generates a new video that follows the reference clip, using what each photo shows: face, hair, build and clothes.
No. On this page “gas station AI” means the Gas Station Dance video effect. Oeneo AI makes images and videos; it has nothing for pumps, payments, inventory or station cameras.
Yes. Videos are 9:16 by default, the vertical shape TikTok, Instagram Reels and YouTube Shorts use, and run 12 to 15 seconds in 480p, 768p, 2K, or 4K. Download the video and post it with an AI label.
The credits for that video are refunded automatically, and you can try again with the same or different photos.
Gas Station Dance AI video generator
Pick who dances first, who films the selfie and who steps out of the car, check the price and generate. Fan-made; not affiliated with Litvin, Ravshan, Adel or Akcent.
