GGAvatar: Geometric Adjustment of Gaussian Head Avatar

TL;DR

GGAvatar is a new 3D head avatar model that uses a coarse-to-fine approach with Gaussian primitives and deformation bases, enabling realistic and expressive head animations with improved quality over existing methods.

Contribution

Introduces GGAvatar, a novel 3D head avatar representation combining Gaussian primitives and deformation bases for enhanced modeling of complex head deformations.

Findings

Outperforms state-of-the-art methods in visual quality.

Produces high-fidelity renderings of dynamic head avatars.

Effectively models complex identities and deformations.

Abstract

We propose GGAvatar, a novel 3D avatar representation designed to robustly model dynamic head avatars with complex identities and deformations. GGAvatar employs a coarse-to-fine structure, featuring two core modules: Neutral Gaussian Initialization Module and Geometry Morph Adjuster. Neutral Gaussian Initialization Module pairs Gaussian primitives with deformable triangular meshes, employing an adaptive density control strategy to model the geometric structure of the target subject with neutral expressions. Geometry Morph Adjuster introduces deformation bases for each Gaussian in global space, creating fine-grained low-dimensional representations of deformation behaviors to address the Linear Blend Skinning formula's limitations effectively. Extensive experiments show that GGAvatar can produce high-fidelity renderings, outperforming state-of-the-art methods in visual quality and quantitative metrics.