Google unveiled its advanced AI image and video generation models, Imagen 3 and Veo, at Google I/O. These models, which integrate into Google’s managed machine learning platform Vertex AI, mark a significant leap forward in generative AI technology.
More than six months after their initial announcement, the Mountain View-based company recently made these models accessible to enterprise clients on Vertex AI.
Imagen 3, previously integrated into various Google platforms and tools like Google Docs and GenChess, will now be available for standalone use in Vertex AI starting next week. It is designed to generate photorealistic images from natural language prompts without requiring users to input technical details.
Veo is now available to Vertex AI customers in private preview, making @GoogleCloud is the first cloud provider of its scale to offer an image-to-video model. Our text-to-image model Imagen 3 will be available to all Vertex AI customers starting next week https://t.co/iWA79MoiVp
— News from Google (@NewsFromGoogle) December 3, 2024The Veo model, developed by DeepMind, allows businesses to create high-quality videos from text or image prompts. Available in a private preview on Vertex AI, Veo can produce videos in diverse cinematic styles while maintaining high fidelity to the original prompts.
Both models come equipped with advanced editing tools, including inpainting and outpainting, allowing companies to tailor images and videos to feature their brand’s colours, styles, and logos. To address privacy and safety concerns, Google has incorporated SynthID watermarking technology into every image and video frame produced by these models. This initiative aims to prevent the misuse of AI-generated content in creating deepfakes and spreading misinformation.
Google also ensures that AI models adhere to stringent data governance and privacy controls, confirming that they do not train these models on customer data.
