Veo 3, Google's video Ai arrives in Italy: here's what it can do
The new frontier of generative video lands in our country via the Gemini app, but with a compulsory 'trademark'.
2' min read
2' min read
After its international debut in May during Google I/O, Veo 3 - the artificial intelligence-based video generator developed by Google DeepMind - is now also available in Italy within the Gemini app (only for the paid Pro version, however). The tool makes it possible to create videos from simple textual prompts, promising a visual quality ever closer to cinema standards. A further piece in the mosaic of generative technologies, but with a focus on rendering, creative control and integration into the Google ecosystem.
The Veo 3 is capable of generating video up to 1080p, with smooth sequences, realistic transitions and a variety of visual styles, from documentary to surreal. The user can simply describe the desired scene ('a sailboat on the sea at sunset, shot from above') and the system returns a coherent, edited and animated clip. Compared to other tools such as OpenAI's Sora, Google emphasises the ability to better control narrative and stylistic consistency over time, a key point for those working with professional visual content.
The technology is the result of integration between Gemini 1.5 Pro language models and DeepMind's advances in video generation, and can already be tested in a demo version within the Gemini app, available for Android, iOS and web. An enterprise account is not required, but to access the advanced features, the Gemini Advanced plan, included in the Google One AI Premium subscription (at a cost of €21.99 per month), must be activated.
In addition to generation from text prompts, Veo allows more sophisticated interaction: one can specify camera movements, angles, or edit the aesthetics of the image ('vintage style, warm light, 35mm film effect'). Functions for text-to-edit and frame-by-frame control have already been released to improve the artistic direction of the results.
Google also emphasised the responsible approach to development: every video produced is labelled with SynthID metadata invisible to the human eye but detectable by software. It is a step towards greater transparency in the distinction between real and synthetic content, which in an era of disinformation and deepfake content becomes central.
