# Fast Training of Diffusion Transformer with Extreme Masking for 3D Point Clouds Generation

arXiv: https://arxiv.org/abs/2312.07231 · HTML: https://mlsanity.com/paper/2312.07231/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- Placed in 3D, point cloud and scene generation, one of 14 directions in Diffusion models.
- Builds on 28 of its 35 references that are in the map.
- First posted on arXiv 12 Dec 2023.

## Placed in

- Diffusion models / [3D, point cloud and scene generation](https://mlsanity.com/diffusion/3d-point-cloud-and-scene-generation/)
- Mixture-of-experts / [Vision mixture-of-experts](https://mlsanity.com/moe/vision-mixture-of-experts/)

## Builds on (most cited first, 28 of 28)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [PyTorch: An Imperative Style, High-Performance Deep Learning Library](https://mlsanity.com/paper/1912.01703/) (2019)
- [Masked Autoencoders Are Scalable Vision Learners](https://mlsanity.com/paper/2111.06377/) (2021)
- [Scalable Diffusion Models with Transformers](https://mlsanity.com/paper/2212.09748/) (2022)
- [ShapeNet: An Information-Rich 3D Model Repository](https://mlsanity.com/paper/1512.03012/) (2015)
- [Objaverse: A Universe of Annotated 3D Objects](https://mlsanity.com/paper/2212.08051/) (2022)
- [MaskGIT: Masked Generative Image Transformer](https://mlsanity.com/paper/2202.04200/) (2022)
- [Objaverse-XL: A Universe of 10M+ 3D Objects](https://mlsanity.com/paper/2307.05663/) (2023)
- [Diffusion Probabilistic Models for 3D Point Cloud Generation](https://mlsanity.com/paper/2103.01458/) (2021)
- [Muse: Text-To-Image Generation via Masked Generative Transformers](https://mlsanity.com/paper/2301.00704/) (2023)
- [LION: Latent Point Diffusion Models for 3D Shape Generation](https://mlsanity.com/paper/2210.06978/) (2022)
- [3D Shape Generation and Completion through Point-Voxel Diffusion](https://mlsanity.com/paper/2104.03670/) (2021)
- [All are Worth Words: A ViT Backbone for Diffusion Models](https://mlsanity.com/paper/2209.12152/) (2022)
- [PointFlow: 3D Point Cloud Generation with Continuous Normalizing Flows](https://mlsanity.com/paper/1906.12320/) (2019)
- [GET3D: A Generative Model of High Quality 3D Textured Shapes Learned from Images](https://mlsanity.com/paper/2209.11163/) (2022)
- [Scaling Language-Image Pre-training via Masking](https://mlsanity.com/paper/2212.00794/) (2022)
- [MDTv2: Masked Diffusion Transformer is a Strong Image Synthesizer](https://mlsanity.com/paper/2303.14389/) (2023)
- [MeshDiffusion: Score-based Generative 3D Mesh Modeling](https://mlsanity.com/paper/2303.08133/) (2023)
- [One Transformer Fits All Distributions in Multi-Modal Diffusion at Scale](https://mlsanity.com/paper/2303.06555/) (2023)
- [Fast Training of Diffusion Models with Masked Transformers](https://mlsanity.com/paper/2306.09305/) (2023)
- [3D Point Cloud Generative Adversarial Network Based on Tree Structured Graph Convolutions](https://mlsanity.com/paper/1905.06292/) (2019)
- [DiT-3D: Exploring Plain Diffusion Transformers for 3D Shape Generation](https://mlsanity.com/paper/2307.01831/) (2023)
- [SoftFlow: Probabilistic Framework for Normalizing Flow on Manifolds](https://mlsanity.com/paper/2006.04604/) (2020)
- [Discrete Point Flow Networks for Efficient Point Cloud Generation](https://mlsanity.com/paper/2007.10170/) (2020)
- [DiffFit: Unlocking Transferability of Large Diffusion Models via Simple Parameter-Efficient Fine-Tuning](https://mlsanity.com/paper/2304.06648/) (2023)
- [SetVAE: Learning Hierarchical Composition for Generative Modeling of Set-Structured Data](https://mlsanity.com/paper/2103.15619/) (2021)

## Built on by (most cited first, 9 of 9)

- [Scaling Diffusion Mamba with Bidirectional SSMs for Efficient Image and Video Generation](https://mlsanity.com/paper/2405.15881/) (2024)
- [X-Ray: A Sequential 3D Representation For Generation](https://mlsanity.com/paper/2404.14329/) (2024)
- [Efficient 3D Shape Generation via Diffusion Mamba with Bidirectional SSMs](https://mlsanity.com/paper/2406.05038/) (2024)
- [DiffGAP: A Lightweight Diffusion Module in Contrastive Space for Bridging Cross-Model Gap](https://mlsanity.com/paper/2503.12131/) (2025)
- [Occlusion-Aware 3D Hand-Object Pose Estimation with Masked AutoEncoders](https://mlsanity.com/paper/2506.10816/) (2025)
- [Foley-Flow: Coordinated Video-to-Audio Generation with Masked Audio-Visual Alignment and Dynamic Conditional Flows](https://mlsanity.com/paper/2603.08126/) (2026)
- [Diff-GNSS: Diffusion-based Pseudorange Error Estimation](https://mlsanity.com/paper/2509.17397/) (2025)
- [PrintAnything: Learning an Intermediate Representation for 3D printing G-code Generation](https://mlsanity.com/paper/2607.27729/) (2026)
- [Point Diffusion Mamba: Unified Diffusion-State-Space Modeling for Single-View 3D Reconstruction under Data Scarcity](https://mlsanity.com/paper/2609.25538/) (2026)

## Abstract (arXiv)

Diffusion Transformers have recently shown remarkable effectiveness in generating high-quality 3D point clouds. However, training voxel-based diffusion models for high-resolution 3D voxels remains prohibitively expensive due to the cubic complexity of attention operators, which arises from the additional dimension of voxels. Motivated by the inherent redundancy of 3D compared to 2D, we propose FastDiT-3D, a novel masked diffusion transformer tailored for efficient 3D point cloud generation, which greatly reduces training costs. Specifically, we draw inspiration from masked autoencoders to dynamically operate the denoising process on masked voxelized point clouds. We also propose a novel voxel-aware masking strategy to adaptively aggregate background/foreground information from voxelized point clouds. Our method achieves state-of-the-art performance with an extreme masking ratio of nearly 99%. Moreover, to improve multi-category 3D generation, we introduce Mixture-of-Expert (MoE) in 3D diffusion model. Each category can learn a distinct diffusion path with different experts, relieving gradient conflict. Experimental results on the ShapeNet dataset demonstrate that our method achieves state-of-the-art high-fidelity and diverse 3D point cloud generation performance. Our FastDiT-3D improves 1-Nearest Neighbor Accuracy and Coverage metrics when generating 128-resolution voxel point clouds, using only 6.5% of the original training cost.