From a homemade costume to the reveal nobody saw coming

Plan the contrast before generating

Choose what changes between the opening and ending: the outfit, hair, makeup or props. Select a matching template and preview all three stages. A clear face and an unobstructed outfit give the model useful references. You can also choose custom mode and describe the look yourself.

How the slideshow is put together

A confession template generates two images and places a blurred-background caption between them. A progressive template generates a separate middle image, so each of the three stages changes the look. Version 1 assembles the images into a nine-second video. Version 2 generates motion; choose its quality and duration before submitting. See the creation walkthrough.

The sound behind the trend

The trend audio is usually “Coming Up Roses”, credited to Harry Styles; on TikTok it commonly appears as “original sound - secret_alt6741” or, in CapCut templates, “original sound - harrison.saltzman”. The slideshow video this site produces is silent on purpose so you can add that sound inside TikTok or Reels yourself. Adding audio through a platform does not grant you rights to use it anywhere else.

Why a familiar face makes the joke work

The reveal only hits when the dream and the truth are recognisably the same child. That is why the best source photo shows one clear face and, if you have it, the whole costume. The AI redraws the costume, not the memory: it keeps the person's face, pose and framing and swaps in the outfit they were promised. Results can drift, so review the face before you post.

Choose the photo before the costume

Pick the costume you were promised first, then the photo that makes the contrast obvious. A butterfly, a vampire, a witch — the costume sets the first caption, while the photo supplies the truth. Only use photos of yourself or your own child, and label the result as AI-generated when you share.

Explore the maker, costume packs, creation walkthrough and photo checklist.