Sunday, May 19, 2024

Forget Sora: Veo is Google's most advanced text-to-video generator

Must read

Screenshot by Sabrina Ortiz/ZDNET

AI image generators were all the rage in 2023, but now companies are shifting focus to the next frontier — AI video generation. With OpenAI unveiling its AI text-to-video generator, Sora, in February 2024, it was only a matter of time before Google did the same. 

On Tuesday, at its annual Google I/O developer conference, Google unveiled Veo, its most advanced text-to-video generator, capable of generating videos with 1080p resolution that are over one minute long.

In addition to the high-quality output, Google says that Veo provides users with an “unprecedented level of creative control.” The AI generator’s deeper understanding of natural language enables Veo to deliver more details from longer prompts and to understand cinematic terms like “timelapse” or “aerial shots.”  

Also: Everything announced during Google I/O 2024: Gemini, Search, Android 15, and more

Additionally, the video generator can tackle a common problem with video generation — the fluidity of shots. According to Google, Veo can create consistent footage, with different subjects such as people, animals, and objects moving realistically in the shots. 

Google isn’t new to video generation. The company noted that this model builds on all its prior video-generating projects, including Imagen-Video, VideoPoet, and Lumiere.

Like OpenAI’s Sora, Google’s Veo is not available to the public yet. Rather, Google is sharing Veo first with select creators in a private preview inside VideoFX. Google does, however, invite that you join a waitlist to eventually try the model.

Additionally, Google unveiled Imagen 3, its highest-quality text-to-image model to date. Imagen 3, which boasts improved image quality and fewer visual artifacts, is also limited to a private preview inside ImageFX for select creators and has its own waitlist

Latest article