Bridging Implicit and Explicit Geometric Transformation for Single-Image View Synthesis

arXiv:2209.07105v3 Announce Type: replace
Abstract: Creating novel views from a single image has achieved tremendous strides with advanced autoregressive models, as unseen regions have to be inferred from the visible scene contents. Although recent methods generate high-quality novel views, synthesizing with only one explicit or implicit 3D geometry has a trade-off between two objectives that we call the "seesaw" problem: 1) preserving reprojected contents and 2) completing realistic out-of-view regions. Also, autoregressive models require a considerable computational cost. In this …

abstract advanced arxiv autoregressive models contents cs.ai cs.cv generate geometry image novel quality synthesis transformation type view