Not everything related to AI lives below the Gemini level at Google. At I/O, the company will unveil several exciting features related to video and image generation, as well as improvements to a tool aimed at distinguishing AI from reality.

At Google I/O, Google is showcasing several new developments in its quest for ever-increasing AI capabilities. These developments are (for now) independent of Gemini and show where generative AI is heading.

AI cameraman

This is how Google displays Veo. Veo is the successor to Lumiere and, like its predecessors, can create moving images. Veo is capable of creating 1080p clips lasting longer than a minute. The model does this using text prompts. Veo is not yet available to the general public, although in some countries you can register for a waiting list through VideoFX.

Veo is Google’s direct answer to OpenAI’s Sora. This model amazed the world in February with good and elaborate video clips based on prompts, where the AI ​​model seemed very aware of the characters, the setting and their relationship to each other throughout the video.

Better pictures

Shouldn’t the images move? Then there’s Imagen 3. Imagen is the latest generation of Google’s genAI image generator and is intended to deliver higher quality images with more detail. This also includes an expanded understanding of user guidance. Patience is also a virtue here because, like Veo, Imagen 3 is not publicly available, but you can sign up for a waiting list. Google publishes examples of the new photo and video functions on its blog.

Are you worried you’ll soon no longer be able to tell the difference between a real photo or video and an AI-generated image or clip? This fear is justified and Google is aware of it. SynthID ensures that generated images contain watermarks so that they are easily recognizable when generated. All Veo videos contain such a watermark.