Pixie: Fast and Generalizable Supervised Learning of 3D Physics from Pixels

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Le, Long, Lucas, Ryan, Wang, Chen, Chen, Chuhao, Jayaraman, Dinesh, Eaton, Eric, Liu, Lingjie
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911122618056704
author Le, Long
Lucas, Ryan
Wang, Chen
Chen, Chuhao
Jayaraman, Dinesh
Eaton, Eric
Liu, Lingjie
author_facet Le, Long
Lucas, Ryan
Wang, Chen
Chen, Chuhao
Jayaraman, Dinesh
Eaton, Eric
Liu, Lingjie
contents Inferring the physical properties of 3D scenes from visual information is a critical yet challenging task for creating interactive and realistic virtual worlds. While humans intuitively grasp material characteristics such as elasticity or stiffness, existing methods often rely on slow, per-scene optimization, limiting their generalizability and application. To address this problem, we introduce PIXIE, a novel method that trains a generalizable neural network to predict physical properties across multiple scenes from 3D visual features purely using supervised losses. Once trained, our feed-forward network can perform fast inference of plausible material fields, which coupled with a learned static scene representation like Gaussian Splatting enables realistic physics simulation under external forces. To facilitate this research, we also collected PIXIEVERSE, one of the largest known datasets of paired 3D assets and physic material annotations. Extensive evaluations demonstrate that PIXIE is about 1.46-4.39x better and orders of magnitude faster than test-time optimization methods. By leveraging pretrained visual features like CLIP, our method can also zero-shot generalize to real-world scenes despite only ever been trained on synthetic data. https://pixie-3d.github.io/
format Preprint
id arxiv_https___arxiv_org_abs_2508_17437
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Pixie: Fast and Generalizable Supervised Learning of 3D Physics from Pixels
Le, Long
Lucas, Ryan
Wang, Chen
Chen, Chuhao
Jayaraman, Dinesh
Eaton, Eric
Liu, Lingjie
Computer Vision and Pattern Recognition
Inferring the physical properties of 3D scenes from visual information is a critical yet challenging task for creating interactive and realistic virtual worlds. While humans intuitively grasp material characteristics such as elasticity or stiffness, existing methods often rely on slow, per-scene optimization, limiting their generalizability and application. To address this problem, we introduce PIXIE, a novel method that trains a generalizable neural network to predict physical properties across multiple scenes from 3D visual features purely using supervised losses. Once trained, our feed-forward network can perform fast inference of plausible material fields, which coupled with a learned static scene representation like Gaussian Splatting enables realistic physics simulation under external forces. To facilitate this research, we also collected PIXIEVERSE, one of the largest known datasets of paired 3D assets and physic material annotations. Extensive evaluations demonstrate that PIXIE is about 1.46-4.39x better and orders of magnitude faster than test-time optimization methods. By leveraging pretrained visual features like CLIP, our method can also zero-shot generalize to real-world scenes despite only ever been trained on synthetic data. https://pixie-3d.github.io/
title Pixie: Fast and Generalizable Supervised Learning of 3D Physics from Pixels
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2508.17437