Talking avatars' tell has moved: jaw and expression coherence, not lip sync
admrys · reddit · 2026-08-18
A creator working with DomoAI talking avatars notes that lip sync is no longer the first thing that breaks immersion. What to watch instead is everything around the mouth: whether the jaw still fits the face, whether cheeks get oddly more expressive than the rest of the character, and whether the smile still belongs to the same face.
For close-ups, they'd take slightly simpler mouth movement over a perfectly articulated mouth that pulls the rest of the face somewhere weird — though the balance is narrow, as too little motion looks stiff and too much draws attention to the animation itself. A telling signal that evaluation criteria for AI talking avatars are shifting from "are the lips aligned" to whole-face consistency.
More from Multimodal
- Turn photos into 3D models using MeshyAI — filiksyos · 2026-08-18
- AI-Edited Photos Are Polluting Citizen Science: Gemini Gave a Heron a Third Leg — kscottz · 2026-08-18
- KAIST's AnyTalk: video diffusion models generate 3D speech animation for arbitrary characters — KAIST · 2026-08-18
- Miniature Monsoon Cloud Over Mumbai: Photorealistic AI Prompt Scenes — CurieuxExplorer · 2026-08-18
- Turn Static Images into Editable 3D Characters with Pure Three.js Code — Scobleizer · 2026-08-18
- Seedance 2.5 Generates Bioluminescent Waves in Hokusai Style — CurieuxExplorer · 2026-08-18