Google has released updated versions of its AI video and image generation models, Veo 2 and Imagen 3.

Veo 2 produces higher-quality videos with improved realism and cinematic capabilities, demonstrated through various example videos showcasing detailed scenes and effects.

Imagen 3 generates brighter, more accurately composed images across diverse art styles. Both models achieve state-of-the-art results in human evaluations and incorporate safety features like SynthID watermarks.

The improved models are being rolled out to Google's VideoFX and ImageFX tools, with planned expansions to other platforms.