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

arXiv: https://arxiv.org/abs/2609.01740 · HTML: https://mlsanity.com/paper/2609.01740/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 34 of its 35 references that are in the map.
- First posted on arXiv 1 Sep 2026.

## Builds on (most cited first, 30 of 34)

- [ShapeNet: An Information-Rich 3D Model Repository](https://mlsanity.com/paper/1512.03012/) (2015)
- [Structured 3D Latents for Scalable and Versatile 3D Generation](https://mlsanity.com/paper/2412.01506/) (2024)
- [LION: Latent Point Diffusion Models for 3D Shape Generation](https://mlsanity.com/paper/2210.06978/) (2022)
- [3DShape2VecSet: A 3D Shape Representation for Neural Fields and Generative Diffusion Models](https://mlsanity.com/paper/2301.11445/) (2023)
- [XCube: Large-Scale 3D Generative Modeling using Sparse Voxel Hierarchies](https://mlsanity.com/paper/2312.03806/) (2023)
- [Native and Compact Structured Latents for 3D Generation](https://mlsanity.com/paper/2512.14692/) (2025)
- [Cascaded Refinement Network for Point Cloud Completion](https://mlsanity.com/paper/2004.03327/) (2020)
- [FlexTok: Resampling Images into 1D Token Sequences of Flexible Length](https://mlsanity.com/paper/2502.13967/) (2025)
- [Dora: Sampling and Benchmarking for 3D Shape Variational Auto-Encoders](https://mlsanity.com/paper/2412.17808/) (2024)
- [3DILG: Irregular Latent Grids for 3D Generative Modeling](https://mlsanity.com/paper/2205.13914/) (2022)
- [Learning Ordered Representations with Nested Dropout](https://mlsanity.com/paper/1402.0915/) (2014)
- [ElasticTok: Adaptive Tokenization for Image and Video](https://mlsanity.com/paper/2410.08368/) (2024)
- [LATTICE: Democratize High-Fidelity 3D Generation at Scale](https://mlsanity.com/paper/2512.03052/) (2025)
- [OctFusion: Octree-based Diffusion Models for 3D Shape Generation](https://mlsanity.com/paper/2408.14732/) (2024)
- [Adaptive Length Image Tokenization via Recurrent Allocation](https://mlsanity.com/paper/2411.02393/) (2024)
- [ZPressor: Bottleneck-Aware Compression for Scalable Feed-Forward 3DGS](https://mlsanity.com/paper/2505.23734/) (2025)
- ["Principal Components" Enable A New Language of Images](https://mlsanity.com/paper/2503.08685/) (2025)
- [Unleashing Vecset Diffusion Model for Fast Shape Generation](https://mlsanity.com/paper/2503.16302/) (2025)
- [LaGeM: A Large Geometry Model for 3D Representation Learning and Diffusion](https://mlsanity.com/paper/2410.01295/) (2024)
- [Learning Adaptive and Temporally Causal Video Tokenization in a 1D Latent Space](https://mlsanity.com/paper/2505.17011/) (2025)
- [ELT: Elastic Looped Transformers for Visual Generation](https://mlsanity.com/paper/2604.09168/) (2026)
- [Efficient Autoregressive Shape Generation via Octree-Based Adaptive Tokenization](https://mlsanity.com/paper/2504.02817/) (2025)
- [3D representation in 512-Byte:Variational tokenizer is the key for autoregressive 3D generation](https://mlsanity.com/paper/2412.02202/) (2024)
- [3D Shape Tokenization via Latent Flow Matching](https://mlsanity.com/paper/2412.15618/) (2024)
- [SceneTok: A Compressed, Diffusable Token Space for 3D Scenes](https://mlsanity.com/paper/2602.18882/) (2026)
- [EVATok: Adaptive Length Video Tokenization for Efficient Visual Autoregressive Generation](https://mlsanity.com/paper/2603.12267/) (2026)
- [Representing 3D Shapes With 64 Latent Vectors for 3D Diffusion Models](https://mlsanity.com/paper/2503.08737/) (2025)
- [Improving Flexible Image Tokenizers for Autoregressive Image Generation](https://mlsanity.com/paper/2601.01535/) (2026)
- [VideoFlexTok: Flexible-Length Coarse-to-Fine Video Tokenization](https://mlsanity.com/paper/2604.12887/) (2026)
- [CaTok: Taming Mean Flows for One-Dimensional Causal Image Tokenization](https://mlsanity.com/paper/2603.06449/) (2026)

## Built on by (most cited first, 1 of 1)

- [Any Time, Any Space: A World-State-Centric Framework for Agentic Video Generation](https://mlsanity.com/paper/2610.20571/) (2026)

## Abstract (arXiv)

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 32× and 8× shorter token sequences, respectively.