Google AI video generation major upgrade! It takes up to 15 seconds to chat to modify the screen.

Gemini Omni 1.2 Flash can generate videos up to 15 seconds long. (Photo/Google)

AI-generated videos are becoming more and more powerful. Google announced the launch of a new generation of AI video generation model, Gemini Omni 1.2 Flash, which supports native 15-second video generation and multi-round conversational editing functions. In addition, the consistency of characters and scenes has been further improved, making AI-generated videos look more natural.

Can generate 15 seconds video

The most impressive change in this upgrade is that AI can generate longer videos at one time. Compared with the previous generation Gemini Omni 1.1 Flash, which can only produce short videos of up to 10 seconds, the new version extends the length by 50% and does not necessarily have to generate a full 15 seconds. Users can freely specify the video length from 3 seconds to 15 seconds according to their needs, adjusting in 1 second increments each time.

For ordinary users, whether they want to turn pet photos into interesting short films, make birthday greeting videos, or shoot social ads for products, there is more room for them to play.

Multiple rounds of conversational editing

In addition to longer videos, another major upgrade is that editing has become easier. It has previously supported conversational video editing, but this time it has further enhanced its functionality, allowing users to continuously modify videos through rounds of conversational commands without having to start from scratch every time.

For example, if you have generated a video of a character standing in a forest, and later want to change the background to a castle, you can directly modify the original scene by entering another command. Google said that during the modification process, AI will try to retain elements that do not need to be changed, such as character movements, clothing, and shot composition, making it easier for creators to produce multiple versions of the same video for different needs.

Follow reference materials more accurately

In the past, AI-generated videos often had problems such as sudden face changes of characters, inexplicable deformation of objects, or lack of detail in different shots. This time, these situations have also been improved. Google pointed out that the new version can more accurately follow the reference images provided by users, allowing characters, products and scenes to remain consistent between different shots.

In addition, the dialogue of the characters in the film has become more natural, and the synchronization effect of mouth shapes and voices has been further improved. In terms of camera movement, AI is also better at producing continuous moving shots, such as first-person shuttle flight, rapid zoom-in, etc., which reduces unnatural jumps in the picture and makes the finished product more cinematic.

Gemini Omni 1.2 Flash is being gradually introduced into many Google applications, but the qualifications for use on each platform are different. In addition to open developers and enterprises, general users can use it through the Gemini App. Initially, it is only available to Google AI Ultra subscribers. As for free users, it can be experienced in Google Vids and Google Ads.