Skip to content

MIMO

Synthesize controllable character videos from simple inputs

Visit official site

What MIMO is

The model is described as a general video synthesis model that imitates a person interacting with objects during complex actions. It synthesizes character videos with controllable attributes such as character, motion and scene from simple inputs.

What you can do with it

  • Generate an animated character from a single image
  • Synthesize complex motions from in-the-wild video
  • Synthesize spatial 3D motions from a database
  • Compose complex real-world scenes with object interaction and occlusion

Application scenarios

  • Generate animated character performances in film production
  • Synthesize game characters with complex motions in game design
  • Create virtual characters that interact with the real world in virtual reality

Getting started

  • Prepare inputs such as a reference image, a pose sequence, a scene video or an image
  • Load the prepared inputs into the model
  • Adjust parameters for character, motion and scene
  • Run the model to synthesize the video

Pros & cons

✓ What we like

  • Attributes such as character, motion and scene can be controlled
  • Inputs stay simple, from a reference image to a pose sequence

! What to watch out for

  • Synthesis starts from user-supplied reference or pose inputs
  • The approach encodes 2D video into compact spatial codes

FAQ

What inputs does it take?

The material names a reference image, a pose sequence, a scene video or an image.

How is the video represented?

The material states 2D video is encoded into compact spatial codes and split into three spatial components.

Last reviewed: 2026-11-15

More AI video generation tools

View all →

How we review