Google announced Veo 2, the next-generation video-generating AI, succeeding the original Veo. It can produce video clips exceeding two minutes in length and resolutions up to 4K (4096 x 2160 pixels), which is 4x the resolution and over 6x the duration compared to OpenAI’s Sora. However, in the experimental tool VideoFX, Veo 2's current output is limited to 720p and eight seconds.
Veo 2 enhances video quality across various subjects and styles, demonstrating improved understanding of real-world physics and human movement. It allows users to specify genres, lenses, and cinematic effects, producing tailored shots like low-angle tracking or close-ups. Examples of its capabilities include cinematic scenes featuring a doctor, a cartoon character, and a farmer, all showcasing detailed and realistic video generation.
Veo 2 aims to minimize common issues in video generation, such as unwanted details, leading to more realistic outputs. Google emphasizes a commitment to safety and responsible development, gradually expanding Veo 2's availability while incorporating an invisible SynthID watermark to identify AI-generated content, reducing misinformation risks.
In addition to Veo 2, Google introduced Imagen 3, an upgraded image generation model that produces brighter and better-composed images across diverse art styles. Imagen 3 is now available globally in ImageFX.
Another new tool, Whisk, allows users to input or create images to visualize and remix ideas, combining Imagen 3 with Gemini’s visual understanding capabilities. Whisk is launching in the U.S. and aims to facilitate creative expression through AI.
