A video, not a selfie
Ask a chatbot how to do this on your phone and it may suggest uploading a selfie to make an AI photo of you in a nice café. That produces a still image with no joke in it. The trend is motion: the perch, the person walking in, the realization and the climb down.
Fancy Cafe AI renders that as a ~15-second vertical clip, ready for the same phone you made it on.
Taking a selfie that works
Stand facing a window or step outside. Hold the phone at eye level and include your shoulders and top. Take off sunglasses and turn off beauty filters. Portrait mode blur is fine, because the background comes from the cafe scene anyway. If you want a friend in the clip, take a separate photo of them rather than a two-person selfie.
Posting from your phone
In TikTok, tap the plus button and upload the MP4. In Instagram, create a Reel from your library. The template sound is already on your clip, so you only need to add audio if you chose Silent. Keep the POV caption and turn on the AI-generated label where it is offered.