How to make a video with the Tesla Roadster AI Video Generator
The Tesla Roadster AI Video Generator turns one or two photos into a short fictional flying-car clip. You do not write a prompt. The scene you pick already carries the motion, and your photo replaces the person or vehicle in that reference.
Start from a scene, not a blank prompt
Open the workspace and choose a scene first. Each card is a finished reference clip: a ramp flip, a lift over traffic, a cabin shot, or a pet on a skateboard. The generator sends that clip with your photo, so the result follows the reference instead of inventing a new stunt from text.
Traffic? What Traffic? is the first scene because it is the simplest people upload: one adult portrait replaces the driver, and the car’s lift-off stays with the reference.
Match the photo to the scene
Car scenes want the whole vehicle, usually a side or rear three-quarter view on plain ground. People scenes want a clear adult face and outfit, without sunglasses or a heavy filter. City in Flight is the exception that needs two portraits: one passenger and one driver.
A cropped bumper, a group selfie, or a pet that is only a head will fight the reference. If the photo cannot show what the scene asks to preserve, the model has to invent it.
Set length, frame, and credits before you submit
Length runs from 4 to 15 seconds. Frame is 9:16, 1:1, or 16:9. Quality is 768P or 2K. The estimate shown in the workspace is the amount that will be charged. It includes the output length and the reference clip, not the output alone.
A 15-second 768P video costs at most 10 credits on every current scene. Shorter clips cost less. 2K costs more, so check the number before you create.