Google has unveiled the preview version of its Veo AI video model, designed to rival OpenAI's Sora, gaining a significant edge in the competition.

Produces Realistic Visuals

Available on Google’s Vertex AI platform, Veo generates cinematic-style 1080p videos and diverse visuals based on text or image prompts. Interestingly, no length limitations are specified for the preview version, making it versatile for various applications. Google’s sample clips are of exceptional quality, making it challenging to discern that they are AI-generated.

Imagen 3 Comes to Google Cloud

Google’s text-to-image AI tool, Imagen 3, will soon be accessible to Google Cloud users. It includes features for customized visuals, logo adjustments, and object placement, providing enhanced creative flexibility.

Both Veo and Imagen 3 integrate safety mechanisms to prevent the creation of harmful content and copyright infringement. However, tests suggest that these safeguards might not be entirely foolproof. AI-generated content from these tools incorporates an invisible digital watermark using DeepMind’s SynthID, ensuring its origins can be verified.

Google’s launch of Veo intensifies pressure on OpenAI to release its much-anticipated Sora model, expected by late 2024.