Skip to content

MOFA-Video

Animate a single image with motion control signals

Visit official site

What MOFA-Video is

MOFA-Video is described as a method that animates a single image with various control signals, using sparse-to-dense (S2D) motion generation and flow-based motion adaptation. The material states that different MOFA-Adapters are trained to drive a pre-trained SVD and can be combined at inference to control the frozen SVD.

What you can do with it

  • Animate a single image with trajectory control signals
  • Animate facial images with keypoints from a driving video or audio
  • Use zero-shot functions such as mixed control, motion brush and control scale
  • Control motion directly through optical flow

Application scenarios

  • Animation makers turn character design images into dynamic animation
  • Game developers add dynamic expressions and actions to game characters
  • Visual effects designers produce video effects with a specific motion trajectory

Getting started

  • Visit the MOFA-Video web page
  • Choose the animation type, such as trajectory-based or keypoint-based animation
  • Upload or enter control signals such as trajectory or keypoint data
  • Adjust animation parameters such as motion intensity and speed

Pros & cons

✓ What we like

  • Zero-shot functions include mixed control, motion brush and control scale
  • Control signals can be combined during inference

! What to watch out for

  • Training relies on sparse motion sampling to build control signals
  • Video generation runs through the pre-trained SVD

FAQ

What control signals are supported?

The material lists trajectories, keypoint sequences and their combinations.

How is the video produced?

MOFA-Adapters are trained to drive a pre-trained SVD, and combined at inference to control the frozen SVD.

Last reviewed: 2026-11-15

More AI video generation tools

View all →

How we review