DreamSat-2.0: Towards a General Single-View Asteroid 3D Reconstruction

TL;DR

DreamSat-2.0 benchmarks and analyzes three state-of-the-art 3D reconstruction models for asteroid and spacecraft datasets, revealing domain-dependent performance and establishing new evaluation benchmarks.

Contribution

Introduces DreamSat-2.0 pipeline and benchmarks three models on custom datasets, providing systematic analysis of their domain-specific strengths and weaknesses.

Findings

Models perform better on complex spacecraft images

Geometric accuracy is higher for simpler asteroid shapes

Hunyuan-3D excels in perceptual quality on spacecraft

Abstract

To enhance asteroid exploration and autonomous spacecraft navigation, we introduce DreamSat-2.0, a pipeline that benchmarks three state-of-the-art 3D reconstruction models-Hunyuan-3D, Trellis-3D, and Ouroboros-3D-on custom spacecraft and asteroid datasets. Our systematic analysis, using 2D perceptual (image quality) and 3D geometric (shape accuracy) metrics, reveals that model performance is domain-dependent. While models produce higher-quality images of complex spacecraft, they achieve better geometric reconstructions for the simpler forms of asteroids. New benchmarks are established, with Hunyuan-3D achieving top perceptual scores on spacecraft but its best geometric accuracy on asteroids, marking a significant advance over our prior work.