Walkthrough

How it works

Google Flow makes 8 or 10 seconds at a time. We write one prompt per stretch. Cuts stay in the timestamps. Faces are meant to be locked with stills you attach in Flow.

1. 01

   Paste a link or upload a file

   Public YouTube goes straight to Gemini. Instagram, TikTok, and Facebook are fetched first, then watched. If a link won’t open, upload the MP4.

2. 02

   We watch picture and sound

   Camera, clothes, light, quoted speech, sound effects, and music, as they actually appear. We don’t invent dialogue.

3. 03

   One prompt per 10 seconds (or 8)

   A 10-second Short is one card. If the camera jumps, those jumps are timestamps inside that card, so Flow isn’t asked for a 2-second clip.

4. 04

   Free, for now

   Free reads the first 60 seconds, 3 times a day. Pro isn’t out yet. Generate the cards in order, and reuse the same face stills on every one.