LoST: Level of Semantics Tokenization for 3D Shapes

TL;DR

LoST introduces a semantic-aware tokenization method for 3D shapes that improves autoregressive generation efficiency and quality by ordering tokens based on semantic salience, surpassing previous methods.

Contribution

The paper proposes LoST, a novel semantic-based tokenization approach for 3D shapes, and RIDA, a new semantic alignment loss, enhancing 3D shape reconstruction and generation.

Findings

LoST achieves state-of-the-art reconstruction performance.

LoST uses significantly fewer tokens than prior methods.

LoST enables high-quality 3D generation and downstream semantic retrieval.

Abstract

Tokenization is a fundamental technique in the generative modeling of various modalities. In particular, it plays a critical role in autoregressive (AR) models, which have recently emerged as a compelling option for 3D generation. However, optimal tokenization of 3D shapes remains an open question. State-of-the-art (SOTA) methods primarily rely on geometric level-of-detail (LoD) hierarchies, originally designed for rendering and compression. These spatial hierarchies are often token-inefficient and lack semantic coherence for AR modeling. We propose Level-of-Semantics Tokenization (LoST), which orders tokens by semantic salience, such that early prefixes decode into complete, plausible shapes that possess principal semantics, while subsequent tokens refine instance-specific geometric and semantic details. To train LoST, we introduce Relational Inter-Distance Alignment (RIDA), a novel 3D semantic alignment loss that aligns the relational structure of the 3D shape latent space with that of the semantic DINO feature space. Experiments show that LoST achieves SOTA reconstruction, surpassing previous LoD-based 3D shape tokenizers by large margins on both geometric and semantic reconstruction metrics. Moreover, LoST achieves efficient, high-quality AR 3D generation and enables downstream tasks like semantic retrieval, while using only 0.1%-10% of the tokens needed by prior AR models.