Google introduces Gemini Omni 1.2 Flash

Google has introduced Gemini Omni 1.2 Flash, an artificial intelligence model designed for video creation. According to Google's official blog, the new model is capable of generating videos up to 15 seconds long, editing existing clips through conversation, and maintaining consistent details within images.

The new model can create longer videos compared to the previous Omni 1.1 version. While the previous version was limited to a 10-second duration, it is now possible to set video lengths from 3 to 15 seconds in one-second increments.

Key capabilities of the model:

• Video editing: Users can make changes to an existing clip during a conversation without having to recreate it from scratch. For example, it is possible to change the background, adjust lighting, or adapt the image format.

• Detail consistency: The model strives to keep characters, clothing, product packaging, and brand colors consistent from the beginning to the end of the video.

• Natural speech: Improvements have been made in distributing voices correctly in scenes with multiple characters, syncing lip movements to speech, and creating more natural-sounding conversations.

• Camera movements: Effects such as continuous camera pans, zooms, and pull-backs can be created with fewer interruptions.

Google has also reduced the price of the new model. The cost to generate a one-second video in 720p quality using Gemini Omni 1.2 Flash is $0.07. This is 30 percent cheaper than the previous version.

The model has been made generally available for production on Gemini API and Google Cloud platforms. It is also being gradually integrated into Google Vids, Flow, the Gemini app, and Google Ads services.