Author: Soumyadeep Sarkar
Source: thetechportal.com
Google has introduced Veo 2, its advanced video generation model, alongside updates to Imagen 3 and a new experimental tool, Whisk. Veo 2 improves on its predecessor by delivering lifelike 4K videos, enhanced real-world physics, accurate human motion, and fluid dynamics, such as simulating the pouring of liquids. It also refines subtle human expressions and offers creators advanced tools to replicate cinematic effects like lens variations and lighting techniques.
Compared to OpenAI’s Sora, which generates 1080p videos of up to 20 seconds, Veo 2 stands out with its superior 4K resolution and video lengths of up to two minutes. However, Google acknowledges ongoing challenges like occasional unnatural motions, though it has made strides in minimizing hallucinations. Veo 2 is currently accessible through Google Labs’ VideoFX platform and positions itself as a significant competitor in AI video generation.