Strips as Tokens: Artist Mesh Generation with Native UV Segmentation
Rui Xu, Dafei Qin, Kaichun Qiao, Qiujie Dong, Huaijin Pi, Qixuan Zhang, Longwen Zhang, Lan Xu, Jingyi Yu, Wenping Wang, Taku Komura
2026-04-14
Summary
This paper introduces a new way to create 3D models using artificial intelligence, specifically focusing on how the AI organizes the information it uses to build the model.
What's the problem?
Current AI methods for generating 3D models struggle to create models that look professionally made. They either create very long and inefficient sequences of instructions, or they disrupt the smooth flow of edges and organized structure that artists rely on when building models by hand. Basically, the AI isn't 'thinking' about how an artist would build something, leading to messy or inefficient results.
What's the solution?
The researchers developed a new system called SATO, which organizes the building instructions like a series of connected faces, similar to how artists use 'triangle strips' to efficiently define surfaces. This method keeps the edges flowing smoothly and maintains a logical layout. A cool part is that SATO can create either triangle-based or square-based models from the same set of instructions, and it learns from both types of data to improve its results.
Why it matters?
This research is important because it brings AI-generated 3D models closer to the quality of those created by professional artists. Better organization of the model data leads to models that look better, have a more logical structure, and are easier to work with, which could speed up the 3D modeling process for games, movies, and other applications.
Abstract
Recent advancements in autoregressive transformers have demonstrated remarkable potential for generating artist-quality meshes. However, the token ordering strategies employed by existing methods typically fail to meet professional artist standards, where coordinate-based sorting yields inefficiently long sequences, and patch-based heuristics disrupt the continuous edge flow and structural regularity essential for high-quality modeling. To address these limitations, we propose Strips as Tokens (SATO), a novel framework with a token ordering strategy inspired by triangle strips. By constructing the sequence as a connected chain of faces that explicitly encodes UV boundaries, our method naturally preserves the organized edge flow and semantic layout characteristic of artist-created meshes. A key advantage of this formulation is its unified representation, enabling the same token sequence to be decoded into either a triangle or quadrilateral mesh. This flexibility facilitates joint training on both data types: large-scale triangle data provides fundamental structural priors, while high-quality quad data enhances the geometric regularity of the outputs. Extensive experiments demonstrate that SATO consistently outperforms prior methods in terms of geometric quality, structural coherence, and UV segmentation.