3D text-to-motion gets major upgrade: multiple characters, faces, hands, better acting
andrew_n_carr · x · 2026-09-17
Andrew Carr previews a massive upgrade to 3D text-to-motion, demoing with the prompt '3 people sitting around and laughing.' The improvements: multiple characters, faces, hands, much better prompt adherence, and higher-fidelity subtle acting performance — all in 3D, coming soon.
More from Multimodal
- Dev spends a day combining MeshyAI 3D assets with shaders for striking shadow enemies — AIandDesign · 2026-09-17
- Krea Agent generates multiple camera angles from a single video — jnack · 2026-09-17
- MetaHumans digital humans run in-browser via three.js with dynamic expressions — majidmanzarpour · 2026-09-17
- YuE2 tested: the best local AI music model yet, close to Suno V6 — MattVidPro · 2026-09-17
- Z Image Turbo: Step-by-Step Guide to Running the Best Uncensored Text-to-Image Model Locally — whyme00h · 2026-09-17
- Dev praises unreleased Union Alpha model's visuals, sees gamedev workflow fit — hullabaloo22 · 2026-09-17