Reproduce “One View Is Enough: In-the-Wild Monocular Pretraining for Novel View Generation” 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
Adrien Ramanana Rahary, Nicolas Dufour, Patrick Perez, David Picard
OVIE trains a pose-conditioned view generator from unpaired single images by using monocular depth to create incomplete pseudo-targets, masking missing pixels during training, and discarding depth at inference. It produces views from one source image in one forward pass. On out-of-domain DL3DV, it reaches 14.8 dB PSNR versus VIVID’s 14.5, but is demonstrated for moderate viewpoint changes.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.