Skip to content

MotionClone

Clone motion from a reference video without training

Visit official site

What MotionClone is

The framework is described as a training-free way to clone motion from a reference video and use it to control text-to-video generation, with temporal attention representing the reference motion.

What you can do with it

  • Clone motion from a reference video
  • Represent motion with temporal attention
  • Reduce noise with primary attention guidance
  • Guide spatial relations with foreground position

Getting started

  • Set up the code base and a virtual environment
  • Download Stable Diffusion V1.5
  • Prepare community models and a motion module

Pros & cons

✓ What we like

  • Prompt following of the generation model is enhanced
  • No training is needed for motion cloning

! What to watch out for

  • Motion comes from a reference video
  • Generation targets text-to-video output

FAQ

Is training required?

The material states the framework is training free.

What guides spatial relations?

A position-aware semantic guidance mechanism is described.

Last reviewed: 2026-11-15

More AI video generation tools

View all →

How we review