Gas Station Dance Prompt: Three Roles and Four Scenes

A Gas Station Dance prompt should name the three photo roles and describe the complete shot order: opening dancer, selfie, car exit, then returning selfie. Reference-capable video tools also need the photos attached in that order. Text alone cannot supply your friends’ identities. GasStationDance.video handles those scene instructions internally; its three-photo workflow does not require you to write a prompt.

See the sequence before writing it

This is our single published 480p result, made from three fictional adult portraits. It shows the four intended moments and generated audio, with no original video supplied. It does not establish a repeatable success rate or exact reference choreography. The case study shows the actual inputs and limitations.

A prompt you can adapt

The following is an editorial example based on our four-scene brief. We have not tested this exact wording across different models. Image labels and supported references depend on the tool you use.

Create a vertical gas-station scene with three distinct adult characters.
Reference photo 1: the opening dancer.
Reference photo 2: the person filming the selfie, who returns at the end.
Reference photo 3: the person stepping out of the car.

Show four moments in order:
1. The first character dances on the forecourt.
2. The camera turns to the second character in selfie view.
3. The camera reveals the third character getting out of a car.
4. Return to the second character for the final selfie.

Keep each character’s face, hair and outfit distinct between shots.
Use coherent camera transitions and natural-speed movement.
If supported, include upbeat instrumental audio and natural scene sounds.
Do not add dialogue, text or logos. Do not end at the car exit.

Request duration, aspect ratio and resolution in the tool’s settings where supported. Writing “20 seconds” in a prompt does not override a model’s duration limit. Likewise, asking for a song does not demonstrate that the tool can reproduce or license it.

Which model can accept this kind of brief?

Our published sample used Wan wan/3-0-video through Kie, with three image references, a 20-second request, 480p, 9:16 and audio enabled. Kie’s model documentation describes its generation interface. This identifies our tested setup; it is not a comparison proving that Wan is the best model.

For another tool, check whether it accepts multiple subject references, how it labels them and whether it supports the required duration. A first-frame-only workflow may not offer a separate reference for each later role. We have not validated this prompt on Kling, Seedance, Veo or other models, so no compatibility is promised here.

Why can roles change or the ending disappear?

• The same person appears twice: check photo order, use one clear person per reference and describe the role separately from the camera action. A prompt cannot fix an incorrect upload.
• The car-exit character replaces the selfie character: state that the second character returns at the end. Avoid describing both as “the main person.”
• The output ends at the car door: request the final selfie explicitly and check the available duration. A longer story may still be shortened by the model.
• Clothing or hands change: reduce conflicting visual instructions and inspect the full motion. Extra adjectives do not guarantee likeness.
• The dance or music differs from the viral clip: this brief makes a new scene. Preserving an existing clip’s motion and soundtrack requires a suitable reference-editing workflow and the right to use that reference.

Change one instruction or input at a time, save your settings and retain failed outputs when evaluating quality. Repeated attempts can cost money; we do not have evidence that rewriting the prompt alone reliably resolves these problems.

Manual prompt or dedicated generator?

A manual workflow is useful when you want to experiment with setting, scene timing or camera language. You must choose the model, attach references, set output options and judge each attempt. GasStationDance.video is designed for the specific three-photo sequence: assign the dancer, selfie and car-exit roles, choose a resolution, then generate.

Neither route guarantees exact faces, original choreography or the trending song. Compare the tool-selection guide before choosing a workflow.

Common questions

Can a text prompt turn a group photo into the three roles?

This site requires three separate role-photo uploads. A group photo does not give its uploader three unambiguous references. Use the photo guide to choose a clear subject for each slot.

Do I need to paste this prompt into GasStationDance.video?

No. Its scene instructions are built in. The prompt above is for understanding and adapting a manual workflow, not a required field in our generator.

Does this reproduce the original song?

No such guarantee is established. Our own example contains model-generated audio. Add a specific track separately only where your publishing tool makes it available for your intended use.

Ready to use three photos without writing a prompt? Follow the creation guide or open the generator. Found an error or an unsupported instruction? Contact us.