Google Unveils Veo 2 Video Generation Model, Expands VideoFX Access

image

Google has announced the launch of Veo 2, an enhanced version of its video generation model, alongside updates to Imagen 3 and a new experiment called Whisk, showcasing capabilities with Gemini.

First introduced in May at I/O 2024, Veo 2 builds on its predecessor with advancements in understanding “real-world physics and the nuances of human movement and expression”, resulting in greater realism and detail.

The new model allows users to specify genre, lens type, and cinematic effects within their prompts. For instance:

  • “…low-angle tracking shot that glides through the middle of a scene”
  • “…close-up shot on the face of a scientist looking through her microscope”
  • “…blur out the background and focus on your subject by putting ‘shallow depth of field’ in your prompt.”

If a prompt mentions an “18mm lens”, Veo 2 can produce the distinctive wide-angle shots associated with it. The model also hallucinates “less frequently” and includes the invisible SynthID watermark for added traceability.

Veo 2 is being rolled out via VideoFX (part of Google Labs), with Google expanding access to more users, though it remains on a waitlist for now. The company confirmed that Veo 2 will arrive in “YouTube Shorts and other products next year.”

“We have been intentionally measured in growing Veo’s availability, so we can help identify, understand and improve the model’s quality and safety while slowly rolling it out via VideoFX, YouTube and Vertex AI,” Google said.

Additionally, Google unveiled an updated Imagen 3 model that offers images with “brighter, better composition, richer details and textures” while improving the ability to “render more diverse art styles with greater accuracy.” Imagen 3 is rolling out globally to ImageFX.

Lastly, Google introduced “Whisk,” an experimental feature in Google Labs designed to showcase Imagen 3 and Gemini’s visual understanding capabilities. Whisk enables users to prompt using images—one for the subject, another for the scene, and a third for style. These inputs can be remixed to generate unique creations, from digital plushies to enamel pins and stickers.

image

Harry and Meghan Share Heartwarming Christmas Card Featuring Their Children