This model is part of the Wavelet Latent Diffusion (WaLa) paper, capable of generating high-quality 3D shapes unconditionally with detailed geometry and complex structures.
Model Details
Model Description
WaLa-UN-1B is a large-scale 3D generative model trained on a massive dataset of over 10 million publicly-available 3D shapes. It can efficiently generate a wide range of high-quality 3D shapes unconditionally in just 2-4 seconds. The model uses a wavelet-based compact latent encoding and a billion-parameter architecture to achieve superior performance in terms of geometric detail and structural plausibility.
Developed by: Aditya Sanghi, Aliasghar Khani, Chinthala Pradyumna Reddy, Arianna Rampini, Derek Cheung, Kamal Rahimi Malekshan, Kanika Madan, Hooman Shayani
This model is specifically developed for the sole purpose of scientific research and development. It is intended for academic and research purposes only for the theoretical exploration and demonstration of the WaLa 3D generative framework. Please see here for inferencing instructions.
Out-of-Scope Use
The model should not be used for:
Commercial purposes
Creation of load-bearing physical objects the failure of which could cause property damage or personal injury
Any usage not in compliance with the license, in particular, the "Acceptable Use" section.
Bias, Risks, and Limitations
Bias
The model may inherit biases present in the publicly-available training datasets, which could lead to uneven representation of certain object types or styles.
The model's performance may vary across different object categories, potentially favoring those that are more prevalent in the training data.
Risks and Limitations
As an unconditional model, there's no direct control over the type or characteristics of the generated shapes.
The model may occasionally generate implausible or unrealistic shapes. Even theoretically plausible shapes should not be relied upon for real-world structural soundness.
The model was trained on a dataset of over 10 million 3D shapes aggregated from 19 different publicly-available sub-datasets, including ModelNet, ShapeNet, SMLP, Thingi10K, SMAL, COMA, House3D, ABC, Fusion 360, 3D-FUTURE, BuildingNet, DeformingThings4D, FG3D, Toys4K, ABO, Infinigen, Objaverse, and two subsets of ObjaverseXL (Thingiverse and GitHub).
Training Procedure
Preprocessing
Each 3D shape in the dataset was converted into a truncated signed distance function (TSDF) with a resolution of 256³. The TSDF was then decomposed using a discrete wavelet transform to create the wavelet-tree representation used by the model.
Training Hyperparameters
Training regime: Please refer to the paper.
Speeds, Sizes, Times
The model contains approximately 1.1 billion parameters.
The model can generate shapes within 2-4 seconds.
Technical Specifications
Model Architecture and Objective
The model uses a U-ViT architecture with modifications. It employs a wavelet-based compact latent encoding to effectively capture both coarse and fine details of 3D shapes. Unlike the conditional models, this unconditional model does not use any input conditions and generates 3D shapes solely based on random noise inputs.
Compute Infrastructure
Hardware
The model was trained on NVIDIA H100 GPUs.
Citation
@misc{sanghi2024waveletlatentdiffusionwala,
title={Wavelet Latent Diffusion (Wala): Billion-Parameter 3D Generative Model with Compact Wavelet Encodings},
author={Aditya Sanghi and Aliasghar Khani and Pradyumna Reddy and Arianna Rampini and Derek Cheung and Kamal Rahimi Malekshan and Kanika Madan and Hooman Shayani},
year={2024},
eprint={2411.08017},
archivePrefix={arXiv},
primaryClass={cs.CV},
url={https://arxiv.org/abs/2411.08017},
}
