Skip to content

DreaMoving

A diffusion framework for controllable human dance videos

Visit official site

What DreaMoving is

The project page presents DreaMoving as a controllable video generation framework built on diffusion models, aimed at high-quality customised human dance videos. It takes a target identity and a pose sequence, then generates a video of that identity performing the poses. Two components are named: a video control net for motion control and a content guider that preserves identity information.

What you can do with it

  • Generate a dance clip from a specified identity and pose sequence
  • Place a generated dancer in a chosen setting
  • Compare stylised diffusion backbones on the same motion

What to watch out for

  • A target identity and a pose sequence are required inputs
  • The model is described as adaptable to most stylised diffusion models

Pros & cons

✓ What we like

  • Motion and identity are handled by explicitly named components
  • The input requirements are stated
  • Compatibility with stylised diffusion models is described

! What to watch out for

  • No hosted demo, price or licence is published here
  • Hardware needs and output specifications are not given
  • Likeness and consent handling is not addressed

FAQ

What inputs are needed?

A target identity and a pose sequence, with simple text and image descriptions listed as well.

What controls the motion?

A video control net is named, with a content guider for identity.

Is there a hosted product?

The source is a research project page and does not describe one.

Last reviewed: 2026-11-15

More AI video generation tools

View all →

How we review