ZipTok3D: High-Fidelity 3D Tokenization with Compact Token Prefixes

Hugging Face Daily Papers Papers
3d-tokenization high-fidelity compact-tokens iterative-decoding 3d-reconstruction

Summary

ZipTok3D is a 3D tokenizer that achieves high-fidelity reconstruction from extremely short token sequences by organizing geometry into compact global-token prefixes and using iterative decoding, outperforming baselines with significantly reduced token lengths.

Compact token sequences are essential for efficient 3D generation. However, existing 3D tokenizers typically organize latent representations either over spatial regions or as fixed-size sets of global tokens, both suffering sharp reconstruction degradation when compressed to extremely low token budgets. In this paper, we present ZipTok3D, a 3D tokenizer designed for high-fidelity reconstruction from extremely short token sequences. Its key idea is to organize object geometry into progressively informative global-token prefixes and unfold these compact representations through iterative decoding. Specifically, nested dropout randomly truncates the latent sequence after encoding during training and requires each retained prefix to reconstruct the complete object, thereby prioritizing essential geometric information in the leading tokens. The decoder then repeatedly applies a parameter-shared Transformer block to recover fine-grained geometry from each prefix without a separate generative sampling stage. With the same token dimension, ZipTok3D achieves reconstruction quality comparable to the 32-token COD-VAE baseline using only one token on ShapeNet and four on TRELLIS, yielding 32times and 8times shorter token sequences, respectively.
Original Article
View Cached Full Text

Cached at: 09/03/26, 11:52 AM

Paper page - ZipTok3D: High-Fidelity 3D Tokenization with Compact Token Prefixes

Source: https://huggingface.co/papers/2609.01740

Abstract

ZipTok3D organizes 3D geometry into compact global-token prefixes with iterative decoding to achieve high-fidelity reconstruction from extremely short token sequences.

Compact token sequences are essential for efficient 3D generation. However, existing3D tokenizers typically organize latent representations either over spatial regions or as fixed-size sets of global tokens, both suffering sharp reconstruction degradation when compressed to extremely low token budgets. In this paper, we present ZipTok3D, a3D tokenizerdesigned for high-fidelity reconstruction from extremely short token sequences. Its key idea is to organize object geometry into progressively informativeglobal-token prefixesand unfold these compact representations throughiterative decoding. Specifically,nested dropoutrandomly truncates the latent sequence after encoding during training and requires each retained prefix to reconstruct the complete object, thereby prioritizing essential geometric information in the leading tokens. The decoder then repeatedly applies aparameter-shared Transformerblock to recover fine-grained geometry from each prefix without a separate generative sampling stage. With the same token dimension, ZipTok3D achieves reconstruction quality comparable to the 32-tokenCOD-VAEbaseline using only one token on ShapeNet and four on TRELLIS, yielding 32times and 8times shorter token sequences, respectively.

View arXiv pageView PDFProject pageAdd to collection

Get this paper in your agent:

hf papers read 2609\.01740

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2609.01740 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2609.01740 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2609.01740 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

3DZip: Spatial-Aware Feature Diversity-Guided Token Compression for 3D Question Answering

Hugging Face Daily Papers

3DZip is a three-stage token compression framework for 3D vision-language models that uses voxelization, diversity-guided anchor selection, and spatial constraint merging to reduce tokens while preserving spatial reasoning. It retains 94.7% of original performance with only 128 tokens and achieves 1.92x faster inference on 3DQA benchmarks.