MeshOctave enables faster mesh generation with adaptive resolution transitions
MeshOctave generates meshes via cascading resolution transitions
Computer Vision and Pattern Recognition
Summary
Generating detailed 3D meshes that look like artist designs is often slow and complex due to the way current methods build them bit by bit or guess connections heuristically. This paper introduces MeshOctave, which speeds up mesh creation by breaking down the process into scales based on spatial grids instead of simplifying existing meshes step-by-step. By doing so, MeshOctave performs operations in parallel, adding more detail without waiting, and improves the quality and correctness of the mesh shapes. This approach also easily adapts to different detail levels and can handle mesh subdivision naturally.
What this means in practice
- •For 3d graphics developers: Create artist-style 3D models faster by generating meshes with adaptive resolution and parallel processing of mesh details.
- •For game engine engineers: Implement mesh generation systems that improve performance and mesh quality by leveraging scale-conditioned parallel refinement.
Authors
Junkai Lin, Tianhao Zhao, Hang Long, Huipeng Guo, Jielei Zhang, Youjia Zhang, Jiale Xu, Wenbing Li, Rendong Liang, Jozef Hladký, Matthias Nießner, Yuanming Hu, Wei Yang
Abstract
Generating compact, artist-style meshes with explicit topology typically relies on autoregressive models which incur prohibitive sequential per-token costs, or continuous flow models that depend on heuristic connectivity decoders. Next-scale generation paradigms offer a compelling alternative by enabling parallel intra-scale token prediction and coarse-to-fine refinement from global structure to local topology; yet, existing methods derive hierarchical scales via progressive mesh simplification and invert them sequentially. This eliminates intra-scale parallelism and scales generation steps linearly with face count. In this paper, we propose MeshOctave, which instead defines scale through dyadic spatial grid resolutions, framing coarsening as a deterministic collapse that merges vertices sharing a voxel cell and inherits connectivity. Its inverse operation, split-and-rewire, determines which octant sub-vertices are instantiated for each coarse face and resolves local connectivity using discrete structural tokens. These per-face operations require no serialization, each scale transition is modeled as an unordered set that adds one bit of coordinate precision, naturally supporting dynamic-length meshes and adaptive resolution refinement. We construct a scale-conditioned masked-uniform discrete diffusion model to learn split-and-rewire operation from resolution collapse hierarchies. MeshOctave outperforms strong baselines in geometric fidelity and topological validity by a non-trivial margin, while supporting adaptive resolution refinement and extending naturally to mesh subdivision tasks.