Saved in:
Bibliographic Details
Main Authors: Wang, Peng, Liu, Xiang, Liu, Peidong
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2505.21060
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909858232532992
author Wang, Peng
Liu, Xiang
Liu, Peidong
author_facet Wang, Peng
Liu, Xiang
Liu, Peidong
contents Stylizing 3D scenes instantly while maintaining multi-view consistency and faithfully resembling a style image remains a significant challenge. Current state-of-the-art 3D stylization methods typically involve computationally intensive test-time optimization to transfer artistic features into a pretrained 3D representation, often requiring dense posed input images. In contrast, leveraging recent advances in feed-forward reconstruction models, we demonstrate a novel approach to achieve direct 3D stylization in less than a second using unposed sparse-view scene images and an arbitrary style image. To address the inherent decoupling between reconstruction and stylization, we introduce a branched architecture that separates structure modeling and appearance shading, effectively preventing stylistic transfer from distorting the underlying 3D scene structure. Furthermore, we adapt an identity loss to facilitate pre-training our stylization model through the novel view synthesis task. This strategy also allows our model to retain its original reconstruction capabilities while being fine-tuned for stylization. Comprehensive evaluations, using both in-domain and out-of-domain datasets, demonstrate that our approach produces high-quality stylized 3D content that achieve a superior blend of style and scene appearance, while also outperforming existing methods in terms of multi-view consistency and efficiency.
format Preprint
id arxiv_https___arxiv_org_abs_2505_21060
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Styl3R: Instant 3D Stylized Reconstruction for Arbitrary Scenes and Styles
Wang, Peng
Liu, Xiang
Liu, Peidong
Computer Vision and Pattern Recognition
Stylizing 3D scenes instantly while maintaining multi-view consistency and faithfully resembling a style image remains a significant challenge. Current state-of-the-art 3D stylization methods typically involve computationally intensive test-time optimization to transfer artistic features into a pretrained 3D representation, often requiring dense posed input images. In contrast, leveraging recent advances in feed-forward reconstruction models, we demonstrate a novel approach to achieve direct 3D stylization in less than a second using unposed sparse-view scene images and an arbitrary style image. To address the inherent decoupling between reconstruction and stylization, we introduce a branched architecture that separates structure modeling and appearance shading, effectively preventing stylistic transfer from distorting the underlying 3D scene structure. Furthermore, we adapt an identity loss to facilitate pre-training our stylization model through the novel view synthesis task. This strategy also allows our model to retain its original reconstruction capabilities while being fine-tuned for stylization. Comprehensive evaluations, using both in-domain and out-of-domain datasets, demonstrate that our approach produces high-quality stylized 3D content that achieve a superior blend of style and scene appearance, while also outperforming existing methods in terms of multi-view consistency and efficiency.
title Styl3R: Instant 3D Stylized Reconstruction for Arbitrary Scenes and Styles
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2505.21060