TopoMesh：通过拓扑统一实现高保真网格自编码
TopoMesh: High-Fidelity Mesh Autoencoding via Topological Unification
提出将任意网格统一到 Dual Marching Cubes 拓扑框架下，从而在顶点/面级别建立显式对应并施加显式网格监督，可复用于任何需要高保真网格重建或作为 3D 生成 VAE 的工作。
原文依据 we introduce TopoMesh, a sparse voxel-based VAE that unifies both GT and predicted meshes under a shared Dual Marching Cubes (DMC) topological framework.
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论高保真 3D 生成的主流范式依赖于变分自编码器-扩散模型流水线，其中变分自编码器的重建能力为生成质量设定了严格的上限。限制现有变分自编码器的一个根本挑战是真实网格与网络预测之间的表示不匹配：真实网格具有任意、可变的拓扑结构，而变分自编码器通常预测固定结构的隐式场（例如规则网格上的 SDF）。这种固有的错位阻碍了建立显式的网格级对应关系，迫使先前的工作依赖间接监督信号，如 SDF 或渲染损失。因此，精细的几何细节，尤其是尖锐特征，在重建过程中难以得到良好保留。为解决这一问题，我们提出了 TopoMesh，一种基于稀疏体素的变分自编码器，它在共享的双移动立方体（DMC）拓扑框架下统一了真实网格与预测网格。具体而言，我们通过一种使用 L∞ 距离度量来保留尖锐边缘的重网格化算法，将任意输入网格转换为符合 DMC 的表示。我们的解码器以相同的 DMC 格式输出网格，确保预测网格与目标网格共享相同的拓扑结构。这在顶点和面级别建立了显式对应关系，使我们能够推导出针对拓扑、顶点位置和面朝向的显式网格级监督信号，并具有清晰的梯度。我们的稀疏变分自编码器架构采用这一统一框架，并使用教师强制和渐进分辨率训练进行训练，以实现稳定高效的收敛。大量实验表明，TopoMesh 在重建保真度上显著优于现有的变分自编码器，在尖锐特征和几何细节的保留方面表现更优，如图 1 所示。
英文原文摘要 1803 字符
The dominant paradigm for high-fidelity 3D generation relies on a VAE-Diffusion pipeline, where the VAE's reconstruction capability sets a firm upper bound on generation quality. A fundamental challenge limiting existing VAEs is the representation mismatch between ground-truth meshes and network predictions: GT meshes have arbitrary, variable topology, while VAEs typically predict fixed-structure implicit fields (e.g., SDF on regular grids). This inherent misalignment prevents establishing explicit mesh-level correspondences, forcing prior work to rely on indirect supervision signals such as SDF or rendering losses. Consequently, fine geometric details, particularly sharp features, are poorly preserved during reconstruction. To address this, we introduce TopoMesh, a sparse voxel-based VAE that unifies both GT and predicted meshes under a shared Dual Marching Cubes (DMC) topological framework. Specifically, we convert arbitrary input meshes into DMC-compliant representations via a remeshing algorithm that preserves sharp edges using an Linfinity distance metric. Our decoder outputs meshes in the same DMC format, ensuring that both predicted and target meshes share identical topological structures. This establishes explicit correspondences at the vertex and face level, allowing us to derive explicit mesh-level supervision signals for topology, vertex positions, and face orientations with clear gradients. Our sparse VAE architecture employs this unified framework and is trained with Teacher Forcing and progressive resolution training for stable and efficient convergence. Extensive experiments demonstrate that TopoMesh significantly outperforms existing VAEs in reconstruction fidelity, achieving superior preservation of sharp features and geometric details, as shown in Fig. 1.