This announcement comes three months after OpenAI revealed its competing product, "Sora," with Google surpassing expectations by unveiling "View" through its Vertex AI platform in a private preview.
"View" stands out for its ability to generate high-quality videos in 1080p resolution, with the flexibility to vary visual and cinematic styles based on text or images. When the model was first announced, it was expected that the generated videos would be limited to one minute in length. However, the current preview release does not impose any length restrictions.
In the latest demonstration, Google showcased impressive videos created using "View," which are difficult to distinguish from traditional videos, highlighting the significant advancements in AI within this field.
Google also announced the expansion of "Imagen 3" – its model for generating images from text – making it available to all Google Cloud customers via Vertex AI starting next week. This follows its previous limited release in the U.S. through "AI Test Kitchen" in August. New features will allow users to modify images based on prompts and incorporate branding, logos, or specific themes into the generated content.
To enhance security, Google emphasized that both "View" and "Imagen 3" include safeguards to prevent the creation of harmful content or copyright violations, although reports suggest that some restrictions might be bypassed. Additionally, these new technologies incorporate DeepMind's "SynthID," a digital watermark designed to combat misinformation and ensure proper content attribution. This system mirrors Adobe's "Content Credentials" for AI-generated content.