Skip to content

MoMask

Generate 3D human motion from text descriptions

Visit official site

What MoMask is

The model is described as a text-driven 3D human motion generation model that uses a hierarchical quantization scheme. The approach represents human motion with high-fidelity detail as multiple layers of discrete motion tokens.

What you can do with it

  • Generate 3D human motion from a text input
  • Represent motion detail with a hierarchical quantization scheme
  • Apply the model to text-guided temporal inpainting

Model notes

  • Two distinct bidirectional Transformer networks carry out generation
  • Motion tokens are predicted from text input
  • The material states the model outperforms existing methods on text-to-motion generation
  • Related tasks such as text-guided temporal inpainting are named

Pros & cons

✓ What we like

  • High-fidelity detail in the motion representation is described
  • The model can be applied to text-guided temporal inpainting

! What to watch out for

  • Generation depends on predicting motion tokens from text input
  • The described scheme represents motion in multiple discrete layers

FAQ

What does the model generate?

The material describes text-driven 3D human motion generation.

How is motion represented?

A hierarchical quantization scheme represents motion as multiple layers of discrete motion tokens.

Last reviewed: 2026-11-15

More LLM API platform tools

View all →

How we review