How to Spin an Object: First, Get the Shape Right

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Kabra, Rishabh, Hudson, Drew A., van Steenkiste, Sjoerd, Carreira, Joao, Mitra, Niloy J.
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866917402182156288
author Kabra, Rishabh
Hudson, Drew A.
van Steenkiste, Sjoerd
Carreira, Joao
Mitra, Niloy J.
author_facet Kabra, Rishabh
Hudson, Drew A.
van Steenkiste, Sjoerd
Carreira, Joao
Mitra, Niloy J.
contents Image-to-3D models increasingly rely on hierarchical generation to disentangle geometry and texture. However, the design choices underlying these two-stage models--particularly the optimal choice of intermediate geometric representations--remain largely understudied. To investigate this, we introduce unPIC (undo-a-Picture), a modular framework for empirical analysis of image-to-3D pipelines. By factorizing the generation process into a multiview-geometry prior followed by an appearance decoder, unPIC enables a rigorous comparison of intermediate geometry representations. Through this framework, we identify that a specific representation, Camera-Relative Object Coordinates (CROCS), significantly outperforms alternatives such as depth maps, pretrained visual features, and other pointmap-based representations. We demonstrate that CROCS is not only easier for the first-stage geometry prior to predict, but also serves as an effective conditioning signal for ensuring 360-degree consistency during appearance decoding. Another advantage is that CROCS enables fully feedforward, direct 3D point cloud generation without requiring a separate post-hoc reconstruction step. Our unPIC formulation utilizing CROCS achieves superior novel-view quality, geometric accuracy, and multiview consistency; it outperforms leading baselines, including InstantMesh, Direct3D, CAT3D, Free3D, and EscherNet, on datasets of real-world 3D captures like Google Scanned Objects and the Digital Twin Catalog.
format Preprint
id arxiv_https___arxiv_org_abs_2412_10273
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle How to Spin an Object: First, Get the Shape Right
Kabra, Rishabh
Hudson, Drew A.
van Steenkiste, Sjoerd
Carreira, Joao
Mitra, Niloy J.
Computer Vision and Pattern Recognition
Machine Learning
Image-to-3D models increasingly rely on hierarchical generation to disentangle geometry and texture. However, the design choices underlying these two-stage models--particularly the optimal choice of intermediate geometric representations--remain largely understudied. To investigate this, we introduce unPIC (undo-a-Picture), a modular framework for empirical analysis of image-to-3D pipelines. By factorizing the generation process into a multiview-geometry prior followed by an appearance decoder, unPIC enables a rigorous comparison of intermediate geometry representations. Through this framework, we identify that a specific representation, Camera-Relative Object Coordinates (CROCS), significantly outperforms alternatives such as depth maps, pretrained visual features, and other pointmap-based representations. We demonstrate that CROCS is not only easier for the first-stage geometry prior to predict, but also serves as an effective conditioning signal for ensuring 360-degree consistency during appearance decoding. Another advantage is that CROCS enables fully feedforward, direct 3D point cloud generation without requiring a separate post-hoc reconstruction step. Our unPIC formulation utilizing CROCS achieves superior novel-view quality, geometric accuracy, and multiview consistency; it outperforms leading baselines, including InstantMesh, Direct3D, CAT3D, Free3D, and EscherNet, on datasets of real-world 3D captures like Google Scanned Objects and the Digital Twin Catalog.
title How to Spin an Object: First, Get the Shape Right
topic Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2412.10273