Google announces new and updated AI video and image generation tools
Google is leaning harder into the AI-driven image and video generation world. The company announced new version of Veo and Imagen, its respective AI video and image generators.
Veo 2 is a new video generator that Google claims has "an improved understanding of real-world physics and the nuances of human movement and expression." The AI video generator can create videos (like similar tools) from a prompt. Interestingly, though, Veo 2 can also receive prompts for specific lenses and can simulate their usage.
Videos can be created with resolutions up to 4K and lengths "extended to minutes," according to Google. Google also claims Veo 2 inserts oddities (like extra fingers or bizarre objects) with less frequency than other models. Veo 2 will also roll out to YouTube Shorts sometime in 2025.
Alongside Veo 2, Google DeepMind's AI image creation model got an update. Imagen 3 is claimed to "render more diverse art styles with greater accuracy." The generator can receive prompts that include art styles, like anime or photorealism. Google claims the images are more accurate to the prompt and include greater detail. Like Veo 2, Imagen achieved "state-of-the-art results" when compared against human-made images, according to Google.
Lastly, a new experiment from Google Labs allows users to edit images, both input and prompted, using AI. The tool lets users mix-and-match subjects, scenes, and styles in a drag-and-drop interface to prompt it to create something new. Users can also give text prompts to further refine the output.
These models and tools launch today, although they may require getting through a waitlist prior to use.
Source(s)
