Veo 2 video generation added to Gemini Advance - you can create cinematic quality clips up to 8 seconds
Google has launched a video creation feature inside Gemini Advanced. Users can generate short videos using the new Veo 2 model, which works on the basis of text queries. To do this, just select a model from the list and describe the scene - the system will create an MP4 file with an 8-second video clip in 720p landscape orientation.
Veo 2 is distinguished by its increased attention to motion physics, scene details, and cinematic style. Users can experiment with visual genres, from realism to fantasy, and even describe specific details such as lighting, style, characters, or settings. The more precise the query, the more accurate the video.
Each generated video is marked with a SynthID digital watermark that is embedded in each frame and signals that the content was created by AI. Google notes that the feature will gradually become available to all Gemini Advanced subscribers worldwide - both on the web and on mobile devices.
Examples of prompts
A wide, slow-motion panorama of a huge ice cave at dusk. A pale bluish light shines through from above, illuminating the ice figures frozen in the icy walls. Two people in white space suits are walking in the centre, the lights on their helmets shining. You need to show how big and quiet the cave is.
An animated shot of a small mouse in big glasses reading a book in the light of a shining mushroom in a cosy forest hole.
An aerial shot of a grassy rocky cliff above a sandy beach with waves crashing on the shore. Near the beach, a majestic stone cliff rises from the ocean, illuminated by the warm, golden light of dawn or sunset, conveying the serene beauty of the Pacific coast.
Source: Google
