Exploring a new method for text/image/video to 3D animation with high creative control

gdechichi · x · 2026-09-02

The author claims to have discovered a new way to generate 3D animation from text, image, or video, offering significantly more creative control than current mocap-driven text-to-animation models. Early results are promising, addressing the issue where existing models produce uninteresting animations due to being trained on realistic human motion capture data.

Original post →

More from Multimodal

Multimodal channel →