AI picture mills have been all the craze in 2023, however now firms are shifting focus to the following frontier — AI video technology. With OpenAI unveiling its AI text-to-video generator, Sora, in February 2024, it was solely a matter of time earlier than Google did the identical.
On Tuesday, at its annual Google I/O developer convention, Google unveiled Veo, its most superior text-to-video generator, able to producing movies with 1080p decision which might be over one minute lengthy.
Along with the high-quality output, Google says that Veo offers customers with an “unprecedented stage of inventive management.” The AI generator’s deeper understanding of pure language permits Veo to ship extra particulars from longer prompts and to know cinematic phrases like “timelapse” or “aerial pictures.”
Moreover, the video generator can sort out a standard drawback with video technology — the fluidity of pictures. In keeping with Google, Veo can create constant footage, with totally different topics comparable to individuals, animals, and objects transferring realistically within the pictures.
Google is not new to video technology. The corporate famous that this mannequin builds on all its prior video-generating tasks, together with Imagen-Video, VideoPoet, and Lumiere.
Like OpenAI’s Sora, Google’s Veo just isn’t out there to the general public but. Slightly, Google is sharing Veo first with choose creators in a personal preview inside VideoFX. Google does, nevertheless, invite that you just be part of a waitlist to ultimately attempt the mannequin.
Moreover, Google unveiled Imagen 3, its highest-quality text-to-image mannequin thus far. Imagen 3, which boasts improved picture high quality and fewer visible artifacts, can be restricted to a personal preview inside ImageFX for choose creators and has its personal waitlist.
