Google’s latest Veo 2 is a video-generative AI model capable of creating 4K videos

Google has unveiled Veo 2, its next-gen video-generative AI model, capable of whipping up videos in up to 4K resolution. Joining the party is Imagen 3, Google’s latest image-generating AI model.

Veo 2, as per Google, delivers high-quality videos across various subjects and styles, all while levelling up its understanding of “real-world physics” and the subtleties of human movement and expression. The result is hyper-realistic footage with up to 4K resolution that can stretch up to two minutes. As per Google, Veo 2 can also grasp cinematic lingo like “low-angle tracking shots” and “close-ups,” while simulating lens effects such as an 18mm lens or a shallow depth of field.

Google also claims Veo 2 tackles the notorious “hallucination” quirks that plague other models—no more creepy extra fingers or surprise floating objects. So the videos look more realistic and polished than what you’d get from rival AI generators. To keep things up and up, every Veo 2 creation will feature both visible and invisible SynthID watermarks, helping curb the spread of misinformation and sneaky misattribution.

ALSO SEE: HP OmniBook 14 Ultra Flip review

Meanwhile, Google’s Imagen 3 takes the reins for still visuals, offering photorealistic image generation that’s already live in over 100 countries via Google Labs’ ImageFX. As for Veo 2, it’s currently rolling out in select markets through Google Labs’ VideoFX, with plans to make its way into YouTube Shorts by 2025. If you’re eager to take it for a spin, you can hop onto the waitlist via VideoFX.