Ex-Google founder launches Atmee v2: real-time avatar from 1 photo + 10s audio
alifcoder · x · 2026-11-05
After Google acquired his AR startup, Yuchen Ping spent two years building Atmee v2, a real-time avatar model aimed at making live video agents feel truly lifelike — betting face-to-face is AI's next interface after text and voice.
Key capabilities:
- Any character from just 1 image + 10 seconds of audio
- Contextual expressions, gestures, even interacting with the environment
- Live video calls, voice-only, or video generation modes
Under the hood it's a video foundation model built to excel at characters, distinguishing it from general video models like Seedance 2.5 and Wan 3.0. Demoed inside real apps where you can talk to an AI character while playing — arguably more useful than yet another talking-avatar demo.
Related event: Atmee v2 Generates Lifelike Real-Time Video Avatars from One Photo(4 posts)→
More from Multimodal
- Viggle Remake launches with open weights: recast any viral clip with your AI character at $0.96/10s — ZabihullahAtal · 2026-11-05
- Dev builds a handwriting engine that mimics penmanship styles from 1522 to today — floguo · 2026-11-05
- RefMods workflow keeps characters and voices consistent in MiniMax H3 on a 6 GB GPU — EzkaProductions · 2026-11-05
- Full WoW-parody music video made with Wan 2.2 in ComfyUI, music by Suno — El_Perro_Gordo · 2026-11-05
- codevember Day 3: a melting three.js demo titled 'Melt' — thespite · 2026-11-05
- HunyuanVideo 1.5 T2V FP8 outputs noisy images on AMD RX 7900 XTX — Fragrant-Throat7048 · 2026-11-05