Scholarlink

<p>Text-driven 3D content generation holds significant promise for virtual reality and digital design. However, challenges persist in achieving robust geometric representation, local geometric consistency, and semantic alignment, particularly for complex meshes. This paper introduces a framework incorporating Hierarchical Progressive Encoding (HPE) for multi-scale geometric representation, Local Feature Enhancement (LFE) for local geometric coordination, and Local Consistency Loss (LCL) for mitigating semantic misalignment. Experiments demonstrate superior stylization fidelity, geometric consistency, and semantic alignment, achieving a leading MES score of 29.91. Our codes are available at our project webpage: <a href="https://github.com/Coraaaaal/Enhanced-Text-Driven-3D-Stylization">https://github.com/Coraaaaal/Enhanced-Text-Driven-3D-Stylization</a>.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhanced text-driven 3D stylization: local consistency and hierarchical encoding

  • 期刊论文

  • 发表日期：

    2026年3月26日

查看全文
  • Jiayi Bu,
  • Huihuang Zhao,
  • Yichun Wu,
  • Jiajia Hu

摘要

Text-driven 3D content generation holds significant promise for virtual reality and digital design. However, challenges persist in achieving robust geometric representation, local geometric consistency, and semantic alignment, particularly for complex meshes. This paper introduces a framework incorporating Hierarchical Progressive Encoding (HPE) for multi-scale geometric representation, Local Feature Enhancement (LFE) for local geometric coordination, and Local Consistency Loss (LCL) for mitigating semantic misalignment. Experiments demonstrate superior stylization fidelity, geometric consistency, and semantic alignment, achieving a leading MES score of 29.91. Our codes are available at our project webpage: https://github.com/Coraaaaal/Enhanced-Text-Driven-3D-Stylization.