Scaffold Diffusion: Sparse Multi-Category Voxel Structure Generation with Discrete Diffusion
By: Justin Jung
Potential Business Impact:
Builds realistic 3D worlds from tiny building blocks.
Generating realistic sparse multi-category 3D voxel structures is difficult due to the cubic memory scaling of voxel structures and moreover the significant class imbalance caused by sparsity. We introduce Scaffold Diffusion, a generative model designed for sparse multi-category 3D voxel structures. By treating voxels as tokens, Scaffold Diffusion uses a discrete diffusion language model to generate 3D voxel structures. We show that discrete diffusion language models can be extended beyond inherently sequential domains such as text to generate spatially coherent 3D structures. We evaluate on Minecraft house structures from the 3D-Craft dataset and demonstrate that, unlike prior baselines and an auto-regressive formulation, Scaffold Diffusion produces realistic and coherent structures even when trained on data with over 98% sparsity. We provide an interactive viewer where readers can visualize generated samples and the generation process: https://scaffold.deepexploration.org/
Similar Papers
A Unified Voxel Diffusion Module for Point Cloud 3D Object Detection
CV and Pattern Recognition
Helps cars see objects better in 3D.
Diffusion-Based Data Augmentation for Medical Image Segmentation
CV and Pattern Recognition
Creates fake medical images to train doctors better.
Scalable Single-Cell Gene Expression Generation with Latent Diffusion Models
Machine Learning (Stat)
Creates realistic cell data for science research.