Orientation Matters: Making 3D Generative Models Orientation-Aligned

TL;DR

This paper introduces orientation-aligned 3D object generation from single images, creating a new dataset and fine-tuning models to produce consistently oriented objects, improving downstream tasks like orientation estimation and manipulation.

Contribution

We propose the task of orientation-aligned 3D generation, create the Objaverse-OA dataset, and fine-tune models to produce consistent orientations across categories, enhancing usability in downstream applications.

Findings

Our models outperform post-hoc alignment methods.

Generated objects show consistent orientations across categories.

Applications include zero-shot orientation estimation and rotation manipulation.

Abstract

Humans intuitively perceive object shape and orientation from a single image, guided by strong priors about canonical poses. However, existing 3D generative models often produce misaligned results due to inconsistent training data, limiting their usability in downstream tasks. To address this gap, we introduce the task of orientation-aligned 3D object generation: producing 3D objects from single images with consistent orientations across categories. To facilitate this, we construct Objaverse-OA, a dataset of 14,832 orientation-aligned 3D models spanning 1,008 categories. Leveraging Objaverse-OA, we fine-tune two representative 3D generative models based on multi-view diffusion and 3D variational autoencoder frameworks to produce aligned objects that generalize well to unseen objects across various categories. Experimental results demonstrate the superiority of our method over post-hoc alignment approaches. Furthermore, we showcase downstream applications enabled by our aligned object generation, including zero-shot object orientation estimation via analysis-by-synthesis and efficient arrow-based object rotation manipulation.