Skip to content

Follow-Your-Pose

Generate pose-controlled character videos from pose and text

Visit official site

What Follow-Your-Pose is

The model is described as a text-to-video approach that uses pose information and text descriptions to produce editable, pose-controllable character videos through a two-stage training scheme combined with a pre-trained text-to-image model.

What you can do with it

  • Generate character videos from pose cues and text descriptions
  • Apply a two-stage training scheme to video generation
  • Combine a pre-trained text-to-image model that supports editing and concept composition
  • Use readily available datasets and pre-trained models

Application scenarios

  • A digital media company produces advertising videos quickly
  • An animator designs dynamic poses for game characters
  • A researcher carries out digital human behaviour studies

Getting started

  • Visit the GitHub page to review the project background and functions
  • Set up the development environment with the required libraries and dependencies
  • Install the recommended xformers library to optimise A100 GPU performance
  • Train the model with the provided command-line tools following the training guide

Pros & cons

✓ What we like

  • A local Gradio demo is provided for testing
  • A pre-trained text-to-image model supports editing and concept composition

! What to watch out for

  • Generation is conditioned on both pose information and text descriptions
  • Training is described as a two-stage scheme with a command-line tool

FAQ

What does the model generate?

The material describes editable, pose-controllable character videos from pose information and text descriptions.

Is a demo available?

A local Gradio demo is listed among the features for testing.

Last reviewed: 2026-11-15

More AI video generation tools

View all →

How we review