Single-View View Synthesis in the Wild
with Learned Adaptive Multiplane Images
ACM SIGGRAPH 2022
This paper deals with the challenging task of synthesizing novel views for in-the-wild photographs. Existing methods have shown promising results leveraging monocular depth estimation and color inpainting with layered depth representations. However, these methods still have limited capability to handle scenes with complex 3D geometry. We propose a new method based on the multiplane image (MPI) representation. To accommodate diverse scene layouts in the wild and tackle the difficulty in producing high-dimensional MPI contents, we design a network structure that consists of two novel modules, one for plane depth adjustment and another for depth-aware color prediction. The former adjusts the initial plane positions using the RGBD context feature and an attention mechanism. Given adjusted depth values, the latter predicts the color and density for each plane separately with proper inter-plane interactions achieved via a feature masking strategy. To train our method, we construct large-scale stereo training data using only unconstrained single-view image collections by a simple yet effective warp-back strategy.
Overview of our AdaMPI method. Given a single color image and a depth map estimated by off-the-shelf monocular depth estimators, our method predicts a multiplane image (MPI) with plane depth adjustment for novel view synthesis. Our training dataset is constructed using single-view images in the wild (COCO), as shown on the right.
The website template was adapted from GRAM.