CEO Neil Mohan announced on Wednesday at the Cannes Lions International Festival of Creator 2025 that YouTube will soon connect a powerful video generation model with Google’s latest artificial intelligence (AI). It will support better capabilities of video generation driven by Mountain View -based Tech Dev’s recently released VEO 3 video Generation AI model. It is said that it enables users to make AI infield backgrounds and video clips for shorts, and provides better video quality as well as further upgrades.
Google VE 3 AI Model on YouTube shorts
In a blog post, CEO Neil Mohan said that YouTube is already providing creators to powerful AI tools like Dreamscreen, which allows superchargers to create AI background and video clips for the shorts platform. The video streaming company will add to the recently introduced VEO 3 video generation model by Google Deep Mind.
It is expected that audio support will improve the quality of the video created as well. YouTube says it will be introduced “this summer” even though there is no definite timeline for its arrival.
Introduced in Google I/O 2025 in May, VEO 3 supports local audio generation, which includes acne sounds, background noise and dialogue. Moreover, the video generation model is also said to improve the adherence to the real world physics, and lips.
The company official also highlighted how YouTube’s AI offer is helping to eliminate obstacles. Auto dubing, which translates to the video into nine languages, has been used to increase the access to more than 20 million videos since the arrival of this feature on the platform six months ago.
In particular, Canawa also recently added a new text -to -video tool, which is powered by Google’s VEO 3 model. Create a video clip, this feature allows users to produce a video clip with cinema quality video rendering and ancestral audio generation. Once the video is produced, it can be made properly using several tools in Canawa’s video editor.
Create a video clip is a part of Canawa AI Sweet and thus, available only to paid users. Currently, the company has set a preliminary limit for five video races every month, but it is working to enhance its capabilities.
