Reproduce “ESC: Escaping 2D Diffusion Ambiguity with Geometric Constraints for Consistent Novel View Synthesis” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Ge Ya Luo, Daniel Severo, Brian Karrer, Christopher Pal, Adriana Romero-Soriano, Michal Drozdzal
ESC generates novel views from one image by jointly predicting target RGB and depth, guided by the source image and camera-ray geometry. It scores bidirectional depth-reprojection agreement over valid coverage using PCC. On DL3DV, ESC-PCC-BoN reports lower LPIPS and PCC than LagerNVS, but selects among multiple samples, so uses more denoising compute and does not establish correctness in unseen regions.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.