Saved in:
Bibliographic Details
Main Authors: Huang, Zixuan, Johnson, Justin, Debnath, Shoubhik, Rehg, James M., Wu, Chao-Yuan
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2404.03566
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911827943751680
author Huang, Zixuan
Johnson, Justin
Debnath, Shoubhik
Rehg, James M.
Wu, Chao-Yuan
author_facet Huang, Zixuan
Johnson, Justin
Debnath, Shoubhik
Rehg, James M.
Wu, Chao-Yuan
contents We present PointInfinity, an efficient family of point cloud diffusion models. Our core idea is to use a transformer-based architecture with a fixed-size, resolution-invariant latent representation. This enables efficient training with low-resolution point clouds, while allowing high-resolution point clouds to be generated during inference. More importantly, we show that scaling the test-time resolution beyond the training resolution improves the fidelity of generated point clouds and surfaces. We analyze this phenomenon and draw a link to classifier-free guidance commonly used in diffusion models, demonstrating that both allow trading off fidelity and variability during inference. Experiments on CO3D show that PointInfinity can efficiently generate high-resolution point clouds (up to 131k points, 31 times more than Point-E) with state-of-the-art quality.
format Preprint
id arxiv_https___arxiv_org_abs_2404_03566
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle PointInfinity: Resolution-Invariant Point Diffusion Models
Huang, Zixuan
Johnson, Justin
Debnath, Shoubhik
Rehg, James M.
Wu, Chao-Yuan
Computer Vision and Pattern Recognition
We present PointInfinity, an efficient family of point cloud diffusion models. Our core idea is to use a transformer-based architecture with a fixed-size, resolution-invariant latent representation. This enables efficient training with low-resolution point clouds, while allowing high-resolution point clouds to be generated during inference. More importantly, we show that scaling the test-time resolution beyond the training resolution improves the fidelity of generated point clouds and surfaces. We analyze this phenomenon and draw a link to classifier-free guidance commonly used in diffusion models, demonstrating that both allow trading off fidelity and variability during inference. Experiments on CO3D show that PointInfinity can efficiently generate high-resolution point clouds (up to 131k points, 31 times more than Point-E) with state-of-the-art quality.
title PointInfinity: Resolution-Invariant Point Diffusion Models
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2404.03566