Skip to content

Hand it the long cut.Keep the takethat lands.

The frame Gemini picked from the NASA upload

Gemini watches every frame, picks the moment, and hands back a captioned vertical short in 12 languages.

  • 5 free shorts
  • 12 caption languages
  • ~2 min to render

Real output from a 98-second NASA upload. Unedited.

Start with what you have.

One long video, one pass.

Five steps from upload to a short you can share.

  1. 01

    Watch

    Gemini reads every frame and listens to the audio.

    The 16:9 NASA source video
    00:0001:38
  2. 02

    Pick the moment

    It marks the strongest part and says why.

    0:42
    00:0001:38

    The moment, the hook and the reason it works.

  3. 03

    Caption

    Words land on the beat of the voice.

    A captioned frame from the NASA short
    • Clean
    • Bold pop
    • Pop in
  4. 04

    Translate

    The same cut, captioned for another audience.

    • A rocket movesEnglish
    • एक रॉकेट चलता हैHindi
    • Plus Tamil, Telugu, Bengali, Marathi, Japanese, Korean, Chinese, Thai, Indonesian and Vietnamese.
  5. 05

    Review

    Share a link. Notes land on the timeline.

    • 0:03The rocket explanation is clear and the captions are easy to read.
    • 0:08This section feels slow. A tighter cut would keep my attention.
    • 0:14The satellite section follows the rocket section. Please check the transition.

    Mixed overall. Clear explanations, slow pacing.

Same cut, two audiences.

One 45-second test upload, rendered with English captions and again with Hindi. Real output, unedited.

English captions
Hindi captions

What runs underneath.

Each short takes one path through Google Cloud.

1Gemini on Vertex AIWatches and picks
2Speech-to-Text Chirp 3Times every word
3Cloud Translation12 languages
4Cloud RunRenders the 9:16 cut
5Cloud StorageKeeps it private
  • Veo 3.1 prompt to video
  • Google Search claim checks
  • Firestore live progress
  • Firebase sign-in and hosting

Find the short in your long video.

Your first 5 are free. No card needed.

See pricing