RGBAvatar: Reduced Gaussian Blendshapes for Online Modeling of Head Avatars

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Li, Linzhou, Li, Yumeng, Weng, Yanlin, Zheng, Youyi, Zhou, Kun
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909539392028672
author Li, Linzhou
Li, Yumeng
Weng, Yanlin
Zheng, Youyi
Zhou, Kun
author_facet Li, Linzhou
Li, Yumeng
Weng, Yanlin
Zheng, Youyi
Zhou, Kun
contents We present Reduced Gaussian Blendshapes Avatar (RGBAvatar), a method for reconstructing photorealistic, animatable head avatars at speeds sufficient for on-the-fly reconstruction. Unlike prior approaches that utilize linear bases from 3D morphable models (3DMM) to model Gaussian blendshapes, our method maps tracked 3DMM parameters into reduced blendshape weights with an MLP, leading to a compact set of blendshape bases. The learned compact base composition effectively captures essential facial details for specific individuals, and does not rely on the fixed base composition weights of 3DMM, leading to enhanced reconstruction quality and higher efficiency. To further expedite the reconstruction process, we develop a novel color initialization estimation method and a batch-parallel Gaussian rasterization process, achieving state-of-the-art quality with training throughput of about 630 images per second. Moreover, we propose a local-global sampling strategy that enables direct on-the-fly reconstruction, immediately reconstructing the model as video streams in real time while achieving quality comparable to offline settings. Our source code is available at https://github.com/gapszju/RGBAvatar.
format Preprint
id arxiv_https___arxiv_org_abs_2503_12886
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle RGBAvatar: Reduced Gaussian Blendshapes for Online Modeling of Head Avatars
Li, Linzhou
Li, Yumeng
Weng, Yanlin
Zheng, Youyi
Zhou, Kun
Computer Vision and Pattern Recognition
We present Reduced Gaussian Blendshapes Avatar (RGBAvatar), a method for reconstructing photorealistic, animatable head avatars at speeds sufficient for on-the-fly reconstruction. Unlike prior approaches that utilize linear bases from 3D morphable models (3DMM) to model Gaussian blendshapes, our method maps tracked 3DMM parameters into reduced blendshape weights with an MLP, leading to a compact set of blendshape bases. The learned compact base composition effectively captures essential facial details for specific individuals, and does not rely on the fixed base composition weights of 3DMM, leading to enhanced reconstruction quality and higher efficiency. To further expedite the reconstruction process, we develop a novel color initialization estimation method and a batch-parallel Gaussian rasterization process, achieving state-of-the-art quality with training throughput of about 630 images per second. Moreover, we propose a local-global sampling strategy that enables direct on-the-fly reconstruction, immediately reconstructing the model as video streams in real time while achieving quality comparable to offline settings. Our source code is available at https://github.com/gapszju/RGBAvatar.
title RGBAvatar: Reduced Gaussian Blendshapes for Online Modeling of Head Avatars
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2503.12886