FastAvatar: Instant 3D Gaussian Splatting for Faces from Single Unconstrained Poses

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Liang, Hao, Ge, Zhixuan, Majee, Soumendu, Tiwari, Ashish, Godaliyadda, G. M. Dilshan, Veeraraghavan, Ashok, Balakrishnan, Guha
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909925170479104
author Liang, Hao
Ge, Zhixuan
Majee, Soumendu
Tiwari, Ashish
Godaliyadda, G. M. Dilshan
Veeraraghavan, Ashok
Balakrishnan, Guha
author_facet Liang, Hao
Ge, Zhixuan
Majee, Soumendu
Tiwari, Ashish
Godaliyadda, G. M. Dilshan
Veeraraghavan, Ashok
Balakrishnan, Guha
contents We present FastAvatar, a fast and robust algorithm for single-image 3D face reconstruction using 3D Gaussian Splatting (3DGS). Given a single input image from an arbitrary pose, FastAvatar recovers a high-quality, full-head 3DGS avatar in approximately 3 seconds on a single NVIDIA A100 GPU. We use a two-stage design: a feed-forward encoder-decoder predicts coarse face geometry by regressing Gaussian structure from a pose-invariant identity embedding, and a lightweight test-time refinement stage then optimizes the appearance parameters for photorealistic rendering. This hybrid strategy combines the speed and stability of direct prediction with the accuracy of optimization, enabling strong identity preservation even under extreme input poses. FastAvatar achieves state-of-the-art reconstruction quality (24.01 dB PSNR, 0.91 SSIM) while running over 600x faster than existing per-subject optimization methods (e.g., FlashAvatar, GaussianAvatars, GASP). Once reconstructed, our avatars support photorealistic novel-view synthesis and FLAME-guided expression animation, enabling controllable reenactment from a single image. By jointly offering high fidelity, robustness to pose, and rapid reconstruction, FastAvatar significantly broadens the applicability of 3DGS-based facial avatars.
format Preprint
id arxiv_https___arxiv_org_abs_2508_18389
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle FastAvatar: Instant 3D Gaussian Splatting for Faces from Single Unconstrained Poses
Liang, Hao
Ge, Zhixuan
Majee, Soumendu
Tiwari, Ashish
Godaliyadda, G. M. Dilshan
Veeraraghavan, Ashok
Balakrishnan, Guha
Computer Vision and Pattern Recognition
We present FastAvatar, a fast and robust algorithm for single-image 3D face reconstruction using 3D Gaussian Splatting (3DGS). Given a single input image from an arbitrary pose, FastAvatar recovers a high-quality, full-head 3DGS avatar in approximately 3 seconds on a single NVIDIA A100 GPU. We use a two-stage design: a feed-forward encoder-decoder predicts coarse face geometry by regressing Gaussian structure from a pose-invariant identity embedding, and a lightweight test-time refinement stage then optimizes the appearance parameters for photorealistic rendering. This hybrid strategy combines the speed and stability of direct prediction with the accuracy of optimization, enabling strong identity preservation even under extreme input poses. FastAvatar achieves state-of-the-art reconstruction quality (24.01 dB PSNR, 0.91 SSIM) while running over 600x faster than existing per-subject optimization methods (e.g., FlashAvatar, GaussianAvatars, GASP). Once reconstructed, our avatars support photorealistic novel-view synthesis and FLAME-guided expression animation, enabling controllable reenactment from a single image. By jointly offering high fidelity, robustness to pose, and rapid reconstruction, FastAvatar significantly broadens the applicability of 3DGS-based facial avatars.
title FastAvatar: Instant 3D Gaussian Splatting for Faces from Single Unconstrained Poses
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2508.18389