MeshRipple: Structured Autoregressive Generation of Artist-Meshes

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Lin, Junkai, Long, Hang, Guo, Huipeng, Zhang, Jielei, Yang, JiaYi, Guo, Tianle, Yang, Yang, Li, Jianwen, Zhang, Wenxiao, Nießner, Matthias, Yang, Wei
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866911309172310016
author Lin, Junkai
Long, Hang
Guo, Huipeng
Zhang, Jielei
Yang, JiaYi
Guo, Tianle
Yang, Yang
Li, Jianwen
Zhang, Wenxiao
Nießner, Matthias
Yang, Wei
author_facet Lin, Junkai
Long, Hang
Guo, Huipeng
Zhang, Jielei
Yang, JiaYi
Guo, Tianle
Yang, Yang
Li, Jianwen
Zhang, Wenxiao
Nießner, Matthias
Yang, Wei
contents Meshes serve as a primary representation for 3D assets. Autoregressive mesh generators serialize faces into sequences and train on truncated segments with sliding-window inference to cope with memory limits. However, this mismatch breaks long-range geometric dependencies, producing holes and fragmented components. To address this critical limitation, we introduce MeshRipple, which expands a mesh outward from an active generation frontier, akin to a ripple on a surface. MeshRipple rests on three key innovations: a frontier-aware BFS tokenization that aligns the generation order with surface topology; an expansive prediction strategy that maintains coherent, connected surface growth; and a sparse-attention global memory that provides an effectively unbounded receptive field to resolve long-range topological dependencies. This integrated design enables MeshRipple to generate meshes with high surface fidelity and topological completeness, outperforming strong recent baselines.
format Preprint
id arxiv_https___arxiv_org_abs_2512_07514
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle MeshRipple: Structured Autoregressive Generation of Artist-Meshes
Lin, Junkai
Long, Hang
Guo, Huipeng
Zhang, Jielei
Yang, JiaYi
Guo, Tianle
Yang, Yang
Li, Jianwen
Zhang, Wenxiao
Nießner, Matthias
Yang, Wei
Computer Vision and Pattern Recognition
Meshes serve as a primary representation for 3D assets. Autoregressive mesh generators serialize faces into sequences and train on truncated segments with sliding-window inference to cope with memory limits. However, this mismatch breaks long-range geometric dependencies, producing holes and fragmented components. To address this critical limitation, we introduce MeshRipple, which expands a mesh outward from an active generation frontier, akin to a ripple on a surface. MeshRipple rests on three key innovations: a frontier-aware BFS tokenization that aligns the generation order with surface topology; an expansive prediction strategy that maintains coherent, connected surface growth; and a sparse-attention global memory that provides an effectively unbounded receptive field to resolve long-range topological dependencies. This integrated design enables MeshRipple to generate meshes with high surface fidelity and topological completeness, outperforming strong recent baselines.
title MeshRipple: Structured Autoregressive Generation of Artist-Meshes
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2512.07514