ExAvatar

3D Gaussian full-body dynamic expression model

CommonProductVideo3D ModelsDynamic Expressions

ExAvatar is a new type of full-body dynamic expression model that combines the full-body driving capabilities of SMPL-X with the strong appearance modeling of 3DGS. It creates models through simple smartphone scanning and supports animation rendering for various poses and expressions. Its hybrid representation method enhances the naturalness of facial expressions, reduces artifacts for new expressions and poses, and ensures complete compatibility with the facial expression space of SMPL-X.

Supports dynamic driving of the entire body (including hands and face).
Fully compatible with SMPL-X's facial expression space through a hybrid representation method.
Significantly reduces artifacts for new expressions and poses using a connectivity-based regularizer.
Enhances expressiveness of hands and face through joint and facial offsets.
Processes three-plane features using MLP and generates animatable 3D models in conjunction with a regularized mesh.
Utilizes LBS algorithm for animation and employs 3DGS technology for rendering.

\ExAvatar's target audience includes 3D animators
game developers
VFX artists
and other professionals who require high-quality dynamic character models. It offers a fast and convenient way to create realistic 3D character models suitable for animation production
game design

3D model demonstration of Jungkook (BTS) performing the dance move 'Standing Next To You'.
3D model demonstration of aespa performing the dance move 'Supernova'.
3D model demonstration of Michael Jackson performing the dance move 'Smooth Criminal'.

1. Prepare a smartphone and the subject to be captured.
2. Use the smartphone to record the subject in a neutral pose for about 10 seconds.
3. Upload the recorded video to the ExAvatar system.
4. The system will automatically recognize and generate a 3D model.
5. Adjust the model's pose and expressions as needed.

Visit