What happens in the What If You Fly video, second by second?
Your video copies the original clip shot by shot, and your photo takes the creator's place. The room, the camera, the fire and the sky come from the clip:
| Time | What you hear and see |
|---|---|
| 0 to 2 s | "What if you fall?" Your chair tips far back, arms windmilling, then you grab the desk and save yourself. |
| 2 to 3.5 s | "Twin!" Both hands on your head, grinning. |
| 3.5 to 4.8 s | "What if you..." you rock back again, arms wide, and a giant "FLY" fills the screen. |
| 4.8 to 7.4 s | The floor drops away in sunrise light, the webcam tips to look down, and you fall on your back into the clouds with the papers and the furniture. |
| 7.4 to 11 s | Fire roars from your feet and you shoot forward low over the clouds and the sea. |
| 11 to 15 s | You roll onto your back over the shining sea, arms wide, then blast up and rise above the sea of clouds. |
How it works: the best AI video models available edit the clip itself: they swap the person for the one in your photo and keep everything else. You get 15 seconds at 720p in the clip's own 4:3 frame, with the original sound put back after the run, so the lines and the music play over your video. The original is 17 seconds; your version ends as you rise above the clouds, before the last two seconds of the far shot. A run takes 5 to 15 minutes, and the exact price shows on the Generate button before you spend anything.
The words on the T-shirt: in the original, each line is printed on his T-shirt as he says it. The AI usually keeps them, but it can misspell a word or leave the shirt plain. The joke still works without them: the lines are in the sound.






