Back in May, Google first previewed Veo, its first video generation model. The new Veo model had the capability to generate 1080p resolution videos in a wide range of cinematic and visual styles. Google also claimed that Veo can accurately understand a text prompt"s tone and render details, even in longer prompts.
Today, Google announced the private preview of Veo for Vertex AI customers. Using Veo on Vertex AI, customers can generate videos from simple text or image prompts, allowing them to quickly prototype and iterate on video content. With this, Google Cloud becomes the first hyperscaler to offer an image-to-video model for external customers.
Along with the private preview of Veo, Google also announced the general availability of Imagen 3, its latest text-to-image model. Google claims that Imagen 3 can generate images with great detail that are photorealistic and feature fewer distracting visual artifacts compared to Imagen 2.
The Imagen 3 text-to-image model will be available next week on Vertex AI for Google Cloud customers. Google is also introducing the following features that will allow Imagen 3 customers to further customize the AI-generated images:
- Imagen 3 editing offers an easy way to refine and tailor images with text prompts. It will also allow users to edit only parts of an image, like updating product backgrounds and more.
- Imagen 3 Customization will allow customers to steer the model to generate images based on specific characteristics. Customers can make Imagen 3 use their own brand, style, logo, subject, or product features when generating new images.
Google highlighted that both Veo and Imagen 3 come with safety features. Both models prevent the creation of harmful content and adhere to Google"s Responsible AI Principles. Also, SynthID invisible watermarks are embedded into every image and frame that Imagen 3 and Veo create.
With the launch of Veo and Imagen 3, Google Cloud is establishing itself as a leader in the field of AI-powered content creation among the hyperscalers.