Sparse Views, Near Light: A Practical Paradigm for Uncalibrated Point-light Photometric Stereo

arXiv:2404.00098v1 Announce Type: new
Abstract: Neural approaches have shown a significant progress on camera-based reconstruction. But they require either a fairly dense sampling of the viewing sphere, or pre-training on an existing dataset, thereby limiting their generalizability. In contrast, photometric stereo (PS) approaches have shown great potential for achieving high-quality reconstruction under sparse viewpoints. Yet, they are impractical because they typically require tedious laboratory conditions, are restricted to dark rooms, and often multi-staged, making them subject to accumulated errors. To …

abstract arxiv contrast cs.cv dataset light near paradigm practical pre-training progress sampling sphere training type