02GOOGLE

Omni The conversational editor.

Omni turns any mix of text, images, audio and video into a finished shot, then takes notes the way a crew would: change the jacket, move her to a neon alley, keep everything else. Every edit builds on the last, so the scene stays coherent from first take to final.

Generate with Omni12 cr / edit
MODEAny input → Video
EDITConversational
LENGTHUp to 8s
RESUp to 4K
PROMPTKeep the same character, change the jacket to red leather, move her to a neon alley

Where it wins.

  • 01Multi-turn edits that keep the scene consistent
  • 02Up to five reference images per shot
  • 03World knowledge, so physics behaves
PROMPTA paper boat drifting down a flooded city street

Prompting Omni, well.

TIP 01

One note per turn

Give Omni a single change at a time, the way you'd give notes on set. Each edit builds on the last, and small steps keep the scene from drifting.

TIP 02

Only describe the change

Everything you don't mention is preserved. Say what moves, not what stays, and the character, lighting and framing carry over on their own.

TIP 03

Feed it references

Up to five images can anchor a shot: a face, a wardrobe, a location, a colour palette. References beat adjectives for keeping a subject on model.

The rest of the lineup.

Compare all five

Your next shot is a sentence away.