(Credit: PCMag/Michael Kan)
OpenAI’s Sora is gone, but Google is filling the void with its own AI-powered video-generation function called Gemini Omni.
At Google I/O, the company debuted Gemini Omni, a tool for creating AI-generated video clips from your existing photos, selfies, or videos. In a demo, Google's AI chief Demis Hassabis showed how you can use the new AI model to drastically change your surroundings while taking a video of yourself. This might include placing yourself on Mars, in a lush forest, or adding a disco ball in the background.
(Credit: PCMag/Michael Kan)Omni isn't just an AI filter, though. It actually represents a step to build a "world" model designed to accurately simulate real-world physics as part of Google's effort to create artificial general intelligence. The same capability enables it to create realistic-looking videos across a wide variety of styles and topics. In another short demo, Hassabis showed Omni can create an educational video using claymation, breaking down scientific concepts for kids.
(Credit: PCMag/Michael Kan)Google plans on making the Omni Flash model available today through the Gemini App, Google Flow, and on YouTube Shorts. But for now, the company notes you'll need a paid subscription through Google AI Plus, Pro or Ultra to access it on the Gemini app or Google Flow. Users on YouTube Shorts and YouTube Create App will get access this week at no extra cost.
The feature arrives after OpenAI officially discontinued both the Sora app and web experience last month to use the company’s AI computing power for other projects. Sora faced plenty of controversy and legal action over the technology creating AI-generated videos featuring characters from popular franchises and dead celebrities.
In contrast, Google is framing Omni as a tool to reimagine your own personal photos or videos by adding fictional AI elements, which might help sidestep potential legal battles. Still, we could also see the capability unleashing deepfakes that could fool the public. In response, the company says: "All videos created with Omni include our imperceptible SynthID digital watermark. You can easily verify that videos were generated with Gemini Omni through the Gemini app, Gemini in Chrome and Google Search."
Google adds that Omni will initially focus on creating video outputs. But the company will eventually expand it to include image, audio and text.