arXiv Search Report: ZipTok3D

Display All Results

Query: ziptok3D

Year range: 2024-2026

Source: arXiv

Results: 1 paper

Authors: Mingda Lin, Weijie Wang, Zeyu Zhang, Bowen Cui, Yefei He, Haoyu Zhao, Yuanyu He, Donny Y. Chen, Feng Chen, Bohan Zhuang

Abstract: Compact token sequences are essential for efficient 3D generation. Existing 3D tokenizers organize latent representations over spatial regions or fixed-size global token sets, with reconstruction quality degrading sharply at very low token budgets. ZipTok3D organizes geometry into progressively informative global-token prefixes and unfolds compact representations through iterative decoding. Nested dropout trains each retained prefix to reconstruct the complete object, while a parameter-shared Transformer decoder repeatedly recovers fine-grained geometry. The method matches the reconstruction quality of a 32-token COD-VAE baseline with one token on ShapeNet and four on TRELLIS.

Summary

Overview

The query ziptok3D returned one exact arXiv match in the 2024-2026 window. It presents a compact 3D tokenizer focused on preserving reconstruction fidelity with extremely short token sequences.

Trends

This single result does not support temporal or venue-level trend claims. Its emphasis reflects a broader direction toward reducing 3D representation length to improve downstream generation efficiency.

Key Themes

• Compact 3D tokenization: reducing sequence length while preserving geometry (1).
• Prefix-based representations: placing the most informative geometry in early tokens (1).
• Iterative decoding: recovering detail with a shared Transformer decoder (1).

Keywords Frequency

Most Cited by Accepted Paper

Most Cited by First Author

Recommendations for Reading

1. ZipTok3D: read first for its compact-prefix tokenizer and low-token-budget reconstruction strategy.