3DShape2VecSet: A 3D Shape Representation for Neural Fields and Generative Diffusion Models

Zhang Biao, Jiapeng Tang, Matthias Nießner, Peter Wonka

Research output: Contribution to journalArticlepeer-review

Abstract

We introduce 3DShape2VecSet, a novel shape representation for neural fields designed for generative diffusion models. Our shape representation can encode 3D shapes given as surface models or point clouds, and represents them as neural fields. The concept of neural fields has previously been combined with a global latent vector, a regular grid of latent vectors, or an irregular grid of latent vectors. Our new representation encodes neural fields on top of a set of vectors. We draw from multiple concepts, such as the radial basis function representation, and the cross attention and self-attention function, to design a learnable representation that is especially suitable for processing with transformers. Our results show improved performance in 3D shape encoding and 3D shape generative modeling tasks. We demonstrate a wide variety of generative applications: unconditioned generation, category-conditioned generation, text-conditioned generation, point-cloud completion, and image-conditioned generation.
Original languageEnglish (US)
Pages (from-to)1-16
Number of pages16
JournalACM transactions on graphics
Volume42
Issue number4
DOIs
StatePublished - Jul 26 2023

Bibliographical note

KAUST Repository Item: Exported on 2023-09-07
Acknowledgements: We would like to acknowledge Anna Frühstück for helping with figures and the video voiceover. This work was supported by the SDAIA-KAUST Center of Excellence in Data Science and Artificial Intelligence (SDAIA-KAUST AI) as well as the ERC Starting Grant Scan2CAD (804724).

Fingerprint

Dive into the research topics of '3DShape2VecSet: A 3D Shape Representation for Neural Fields and Generative Diffusion Models'. Together they form a unique fingerprint.

Cite this