Google has introduced a new photo-to-video generation feature in its Gemini platform, allowing users to animate still images into dynamic, eight-second video clips using the Veo 3 model.
The capability is available to Google AI Pro and Ultra subscribers in select regions starting today.
The announcement from David Sharon, Multimodal Generation Lead at Gemini Apps, signifies integration into the existing video tools available in Gemini and Flow, Google’s AI filmmaking suite. Users can upload a static photo, describe the scene and desired audio, and receive an AI-generated video complete with motion and sound, designed to bring images to life in a short cinematic format.
Veo 3, the third-generation video generation model announced in May 2025, underpins this new functionality. The model specializes in generating high-fidelity video content from multimodal prompts and has already seen broad uptake, with over 40 million video creations logged through the Gemini app and Flow within the last seven weeks. Creative applications have ranged from surreal reimaginings of fairy tales to sensory-heavy ASMR clips.
Gemini is Google’s unified generative AI assistant, available on web and mobile, positioned as a successor to Bard. Integrated with Google’s ecosystem of apps and services, Gemini targets both general users and creative professionals looking to explore generative media capabilities.
To access the new feature, users can select “Videos” from the Gemini prompt interface, upload a photo, and provide text describing the desired motion and audio elements. The system then generates an eight-second clip based on the inputs. Google encourages experimentation with personal photos, sketches, or landscapes, suggesting use cases like animating nature scenes or bringing artwork to life.
On the safety front, Google says it has implemented layered safeguards to mitigate misuse. These include red teaming exercises, proactive evaluations for potential abuse scenarios, and strict content moderation policies. All videos generated via Veo 3 are marked with both visible AI-generated watermarks and invisible SynthID signatures, helping users and platforms verify authenticity. Feedback mechanisms are also in place, allowing users to rate video outputs to help refine the system.
The rollout is limited to Google AI Pro and Ultra subscribers in select countries, but the company notes that the same tools are accessible through Flow for users already experimenting with AI-based storytelling.
Leave a Reply