提升稀疏视角下泛化高斯点云渲染质量
${C}^{3}$-GS: Learning Context-aware, Cross-dimension, Cross-scale Feature for Generalizable Gaussian Splatting
- 引入上下文感知、跨维度、跨尺度特征约束增强表示
- 无需额外监督即可实现高质量图像合成与泛化能力
- 适合需要快速部署的3D重建与虚拟现实应用
泛化高斯点云渲染旨在不进行每场景优化的情况下，合成未见场景的新视角图像。近期方法采用前馈网络预测每个像素的高斯参数，实现了从稀疏输入视角的高质量合成。然而，现有方法在编码区分性、多视角一致的特征方面表现不足，难以在稀疏视角下构建准确几何结构。为此，我们提出 $\mathbf{C}^{3}$-GS 框架，通过引入上下文感知、跨维度和跨尺度约束来增强特征学习。该架构将三个轻量级模块集成到统一渲染流程中，提升特征融合能力，实现无需额外监督的逼真图像合成。在基准数据集上的大量实验表明，$\mathbf{C}^{3}$-GS 在渲染质量和泛化能力上均达到当前最优水平。代码已公开于：https://github.com/YuhsiHu/C3-GS。
原文摘要 · Abstract (English)
Generalizable Gaussian Splatting aims to synthesize novel views for unseen scenes without per-scene optimization. In particular, recent advancements utilize feed-forward networks to predict per-pixel Gaussian parameters, enabling high-quality synthesis from sparse input views. However, existing approaches fall short in encoding discriminative, multi-view consistent features for Gaussian predictions, which struggle to construct accurate geometry with sparse views. To address this, we propose $\mathbf{C}^{3}$-GS, a framework that enhances feature learning by incorporating context-aware, cross-dimension, and cross-scale constraints. Our architecture integrates three lightweight modules into a unified rendering pipeline, improving feature fusion and enabling photorealistic synthesis without requiring additional supervision. Extensive experiments on benchmark datasets validate that $\mathbf{C}^{3}$-GS achieves state-of-the-art rendering quality and generalization ability. Code is available at: https://github.com/YuhsiHu/C3-GS.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。