Tech Desk 
Dec 17, 2024
Google DeepMind, Google’s flagship AI research lab, has officially unveiled Veo 2 on Monday to outdo OpenAI's Sora. Veo2 is the latest version of its video generation model, pushing the generative artificial intelligence (AI) boundaries towards excellence by generating real-time videos.
Veo 2 features an enhanced comprehension of real-world physics and human movement to streamline the workflow.
Veo 2 was previously announced in May. It enables users to specify genres, cinematic effects, and resolutions up to 4K, in line with its improved understanding of cinematography.
It’s important to note that Veo 2 will feature Google’s SynthID watermark on its videos to indicate AI generation. Despite significant upgrades, Veo 2 still has some errors, such as hallucinating additional fingers.
In internal tests, Veo 2 surpassed OpenAI’s Sora and other AI models in "overall preference" and "prompt adherence." Veo 2 is set to surpass Sora, which is currently available to OpenAI’s paid subscribers.
Imagen 3 updates
Alphabet-owned Google has brought several significant upgrades to its renowned image generation model, Imagen 3, making it more prompt-precise. It is currently available through Google’s Gemini chatbot and ImageFX platform. Imagen 3 caters to users' diverse needs with high versatility, from photorealism to anime, while generating top-visual images.
Veo 2 availability
Veo 2 is available on Google Labs’ VideoFX platform and expanding the “number of users who can access it” via a waitlist. It is set to expand on “YouTube Shorts and other products next year,” according to Google.
