The Evolution of Facial Animation in Games

In 1998, a talking face in a game ran on a single number. Today a neural network can listen to a voice and invent the whole performance. This video follows how game characters learned to talk: Half-Life's jaw that opens with the loudness, the mouth shapes of Half-Life 2, the hand-animated faces of The Last of Us, the 32 cameras of L.A. Noire (and the "Doubt" meme they left behind), performance capture, Cyberpunk 2077's lip sync in ten languages, and the neural models that run inside game engines now.

Near the end I test three ways of doing lip sync in Unreal Engine, with the same character saying the same line, and show which one I built into StoryFlow Editor, my visual dialogue tool for games.

Which game do you think has the best faces? Let me know in the comments.

CHAPTERS
0:00 A face in a game, then and now
0:38 1998: one number
1:27 2004: reading the words
2:39 By hand
3:21 2011: L.A. Noire films the face
5:52 Capture that drives the sliders
6:44 Two million words
7:56 Now: the neural network
10:03 Back to the chair

THE FULL TEST
All 42 runs and the data: https://storyflow-editor.com/blog/unreal-engine-lip-sync-performance/

LINKS
StoryFlow Editor: https://storyflow-editor.com/
Steam: https://store.steampowered.com/app/4088380/StoryFlow_Editor/
Discord: https://discord.com/invite/3mp5vyKRtN

#gamedev #gameanimation #lipsync

Дата на публикация: 14 октомври, 2026
Категория: Игри
Ключови думи: In The of Animation Evolution games facial