The competition in the multimodal field of large models is rapidly evolving towards higher resolution and longer video generation. On August 27 local time, Google officially launched the new Gemini Omni 1.1 Flash video generation AI model. This model not only supports up to 4K ultra-high definition resolution, but also achieves comprehensive upgrades in scene expansion, camera movement control, and generation efficiency.

As a high-performance tool for developers and professional creators, Gemini Omni 1.1 Flash brings several impressive features in its functional design. In terms of scene expansion, users can seamlessly continue generating subsequent frames directly from the end of an existing video, with each expansion step being 10 seconds, allowing a maximum total length of 40 seconds; in camera movement control, by specifying the starting and ending frames of a shot, it is easy to achieve smooth and natural transition effects and professional camera movements. In addition, the new model supports introducing reference video clips of up to 3 seconds, thus maintaining accurate contextual background and character consistency when building new scenes.

image.png

In balancing image quality and operational efficiency, the model provides multiple resolution options to meet different scenario needs. Among them, the 360p preview mode can increase the generation speed by up to 60% compared to the standard 720p resolution, with a cost only one-third of the original, making it very suitable for rapid prototyping verification, storyboard iteration, and high-speed rendering; while the highest can directly output high-quality final videos with detailed and complete details in 1080p or 4K.