SIE3D: Single-image Expressive 3D Avatar generation via Semantic Embedding and Perceptual Expression Loss

Zhiqi Huang, Dulongkai Cui, Jinglu Hu
9/28/2025
cs.CV

No official code found. Best implementation: huang-zhiqi/SIE3D (0 stars, MIT, last commit 2026-04-28).

Abstract

Generating high-fidelity 3D head avatars from a single image is challenging, as current methods lack fine-grained, intuitive control over expressions via text. This paper proposes SIE3D, a framework that generates expressive 3D avatars from a single image and descriptive text. SIE3D fuses identity features from the image with semantic embedding from text through a novel conditioning scheme, enabling detailed control. To ensure generated expressions accurately match the text, it introduces an innovative perceptual expression loss function. This loss uses a pre-trained expression classifier to regularize the generation process, guaranteeing expression accuracy. Extensive experiments show SIE3D significantly improves controllability and realism, outperforming competitive methods in identity preservation and expression fidelity on a single consumer-grade GPU. Project page: https://blazingcrystal1747.github.io/SIE3D/

View on arXivView PDF

Code Implementations(1)

huang-zhiqi/SIE3DCommunity63%
MIT

Official implementation of the paper “SIE3D: Single-Image Expressive 3D Avatar Generation via Semantic Embedding and Perceptual Expression Loss” [ICASSP 2026]

00PythonJan 22, 20265 months agoMIT

Similar Papers

Towards a Simultaneous and Granular Identity-Expression Control in Personalized Face Generation

85% match2024

GenEAva: Generating Cartoon Avatars with Fine-Grained Facial Expressions from Realistic Diffusion-based Faces

85% match2025

GenCA: A Text-conditioned Generative Model for Realistic and Drivable Codec Avatars

84% match2024

Retrieval-Augmented Gaussian Avatars: Improving Expression Generalization

84% match2026

InstaFace: Identity-Preserving Facial Editing with Single Image Inference

84% match2025

FFAvatar: Feed-Forward 4D Head Avatar Reconstruction from Sparse Portrait Images

84% match2026

Cite this paper

@article{huang2025sied,
  title  = {SIE3D: Single-image Expressive 3D Avatar generation via Semantic Embedding and Perceptual Expression Loss},
  author = {Zhiqi Huang and Dulongkai Cui and Jinglu Hu},
  year   = {2025},
  eprint = {2509.24004},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2509.24004v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2509.24004.svg)](https://codeofpaper.com/paper/2509.24004)

Discussion