The skull cap trend on TikTok has two halves. First you look ordinary and a bit goofy in a bright room. Then the camera tips up to the ceiling, the room goes dark, and you rise back into frame in a black skull cap and hoodie.
I wanted to generate that from a single photo, and one model call wasn't enough. A photo of someone in a T-shirt gives the video model nothing to work with for the second half. So my generator runs in two steps: first it makes a portrait of the same person in the cap and hood, then it makes the video from both images, following the timing of a real clip from the trend.
Rendering is slow. My 10-second test videos at 480p took between 196 and 296 seconds, so people can close the page and come back for the result.
If you want to see the output or try it on your own photo, it's here: Skull Cap AI. The steps for doing the trend, filmed or generated, are in this guide: how to do the skull cap trend.
