(Image source from: REUTERS)
Google is expanding its Gemini Avatar function to more paid subscribers of Gemini. This feature lets users make a digital version of themselves powered by AI, which can be used to produce videos and other types of content within Gemini. It runs on Gemini Omni, a new AI model from Google that claims to create anything based on any input. The Gemini Avatar was first found out during an APK analysis in March. It is intended to imitate a user’s look and voice. Users can find the avatar feature by going into the Gemini app and selecting Settings > Avatar (as noted by Android Authority). After activation, users can build a custom AI avatar that reflects their facial traits, voice, and behaviors. They can also call it up during Gemini chats using certain commands. According to Google, setting it up is quite simple. Users need to start by undergoing a camera-based training session where they hold their phone at eye level and follow the prompts on the screen to capture their face from various angles. This helps Gemini create a precise model of their face shape.
The next part is voice training. Gemini requires users to read aloud some random phrases and numbers. The company states this helps the system understand their speech style, tone, accent, and rhythm. When the setup is finished, the avatar will be connected to the user’s Google account and will be accessible in all Gemini experiences. To make sure that the face and voice scans are approved, users should hold their phone at eye level, with proper lighting that isn’t too dark or too bright. Additionally, they should make sure their eyes, nose, and mouth can be seen. Glasses can be included in the avatar, but hats, sunglasses, and masks are not permitted. Also, there should be no people or pictures of faces in the background. Google has also put in place several protection measures to tackle possible issues with realistic AI avatars. To start, users must be at least 18 years old to make an avatar. The account holder must be there in person during the entire registration process. The company also highlighted that any video made with Gemini Avatar will have Google’s SynthID watermark. This invisible watermark is embedded directly in the content and can be spotted using Google tools. Google states this will help viewers recognize AI-generated videos and lessen the chance of misuse.
The company has not yet revealed if this feature will also be available to free users of Gemini.
