Skip to content

D3GA

Drive a 3D human model with Gaussian splatting

Visit official site

What D3GA is

The model is described as a drivable 3D human body based on Gaussian points, learned from multi-view video and rendered in real time.

What you can do with it

  • Learn a body model from multi-view footage
  • Render the body as Gaussian points
  • Deform the model with joint angles

Model notes

  • Real-time rendering is listed as a feature
  • Key points take part in driving the shape
  • Layered rendering is listed as a feature

Pros & cons

✓ What we like

  • Multi-view video is enough to learn the body model
  • Separate limb and clothing rendering is listed

! What to watch out for

  • Input is described as multi-view video
  • Deformation depends on joint and key point parameters

FAQ

What drives the deformation?

The material names joint angles and key points.

What does layered rendering enable?

The use case list mentions separating limbs and clothing for virtual try-on.

Last reviewed: 2026-11-15

More LLM API platform tools

View all →

How we review