Parametric Gaussian Human Model: Generalizable Prior for Efficient and Realistic Human Avatar Modeling

TL;DR

The paper introduces PGHM, a novel framework that leverages human priors for fast, high-fidelity 3D human avatar reconstruction from monocular videos, addressing efficiency and generalization issues in existing methods.

Contribution

PGHM integrates a UV-aligned latent identity map and a Multi-Head U-Net to enable rapid, high-quality avatar creation with minimal optimization time from monocular inputs.

Findings

Achieves high-quality avatars in about 20 minutes per subject.

Outperforms existing methods in efficiency while maintaining visual fidelity.

Demonstrates robustness under challenging poses and viewpoints.

Abstract

Photorealistic and animatable human avatars are a key enabler for virtual/augmented reality, telepresence, and digital entertainment. While recent advances in 3D Gaussian Splatting (3DGS) have greatly improved rendering quality and efficiency, existing methods still face fundamental challenges, including time-consuming per-subject optimization and poor generalization under sparse monocular inputs. In this work, we present the Parametric Gaussian Human Model (PGHM), a generalizable and efficient framework that integrates human priors into 3DGS for fast and high-fidelity avatar reconstruction from monocular videos. PGHM introduces two core components: (1) a UV-aligned latent identity map that compactly encodes subject-specific geometry and appearance into a learnable feature tensor; and (2) a disentangled Multi-Head U-Net that predicts Gaussian attributes by decomposing static, pose-dependent, and view-dependent components via conditioned decoders. This design enables robust rendering quality under challenging poses and viewpoints, while allowing efficient subject adaptation without requiring multi-view capture or long optimization time. Experiments show that PGHM is significantly more efficient than optimization-from-scratch methods, requiring only approximately 20 minutes per subject to produce avatars with comparable visual quality, thereby demonstrating its practical applicability for real-world monocular avatar creation.