Create AI Avatars With Gemini for Video Generation

My notes

Summary

Google’s Gemini now lets you generate a personal AI avatar from a short selfie video, then drop that avatar into AI-generated videos that say whatever you script. The avatar plugs into Google Vids (slide-deck narration) and Google Flow (complex video prompts), making it usable across the Google video stack.

Key Insight

• Entry point: gemini.google.com/likeness - a dedicated page for creating your own avatar likeness.
• Capture flow: click “Try now”, scan a QR code with your phone, record a guided selfie video. No studio, no green screen.
• Cross-product reuse: the same avatar works in three surfaces - direct video creation, Google Vids (can narrate full slide decks), and Google Flow (more complex prompts).
• Flow integration trick: inside Google Flow, type @ in a prompt to pull up and insert your avatar into a multi-element scene, so the avatar becomes a promptable asset, not just a talking head.
• Scripted speech: you control the dialogue (“have it say whatever you want”), so it doubles as a faceless/low-effort talking-head video generator.