Google Vids Lets You Create AI Avatars From a Single Selfie and Voice Sample

Google has expanded its Google Vids video creation toolkit with two AI-powered features: the Gemini Omni model for generating and editing clips, and personal avatars, digital doubles created from a single photo and a short voice sample. The company announced the changes on July 16, 2026, on its official blog.

What Gemini Omni Does

Gemini Omni is a multimodal video model that generates and edits clips based on text, images and video footage supplied by the user. In practice, this means you can start with a simple written description, add a photo or sketch as a reference point, and the model will combine these elements into a coherent piece of video.

The key change compared with Google's earlier video generation tools is a step-by-step editing mode. Instead of regenerating the entire clip with every tweak, the user can ask in natural language to change the background, fix the lighting, or add an effect, and the model modifies the existing footage while leaving the rest of the scene untouched.

The feature also covers footage shot on a phone. Users can upload their own take and ask Gemini Omni to refine it, for instance by smoothing transitions, correcting color grading, or adding graphic elements, without manual editing in a traditional video editor.

An Avatar From a Single Photo

The second feature, personal avatars, lets users create a digital version of themselves from a single selfie and a short voice sample. Once the avatar is created, the user simply types the script, and the system generates a video in which the digital double speaks that text in the user's own voice and face, no camera recording required.

The use case Google points to is primarily corporate communication: training materials, team updates, or product presentations recorded without having to step in front of a camera every time the content needs updating.

Limits and Safeguards

Google has restricted access to avatars to adults in select regions, and each avatar is tied to a Google account and reserved exclusively for its owner's likeness, meaning it isn't possible to create a digital double of someone else without their participation in the creation process. Every clip generated by Gemini Omni or using an avatar receives an invisible SynthID watermark that allows the material's origin to be verified.

Both features are rolling out exclusively to paying users: subscribers of the Google AI Pro and Ultra plans and business customers of Google Workspace. Regular users of the free version of Vids won't have access to them for now.

A Spot in the AI Video Market

The update puts Google Vids in direct competition with specialized AI video avatar tools such as HeyGen, Synthesia, Captions and D-ID, which have spent the past several years building the market for digital doubles used in corporate and training materials. Google is entering this space backed by the entire Workspace suite, which could push some business customers toward a single integrated tool instead of a separate subscription with a specialized vendor.

For Polish companies using Google Workspace, this opens up the ability to quickly produce video materials, onboarding, sales, or internal communications, without involving a marketing department or outside production. The barrier to entry drops to having the right subscription and one selfie.

Google has not yet published an exact list of countries covered by the personal avatars rollout, nor a timeline for expanding it to further markets, meaning some users in Poland may not have access at launch even if they hold the right subscription plan.