Novel View Synthesis with Pixel-Space Diffusion Models
This addresses the problem of generating new viewpoints from limited input for computer vision applications, representing an incremental improvement through architectural adaptation.
The paper tackles novel view synthesis from a single image by adapting a diffusion model architecture for end-to-end pixel-space generation, substantially outperforming previous state-of-the-art techniques. It introduces a training scheme using single-view datasets to improve generalization to out-of-domain scenes.
Synthesizing a novel view from a single input image is a challenging task. Traditionally, this task was approached by estimating scene depth, warping, and inpainting, with machine learning models enabling parts of the pipeline. More recently, generative models are being increasingly employed in novel view synthesis (NVS), often encompassing the entire end-to-end system. In this work, we adapt a modern diffusion model architecture for end-to-end NVS in the pixel space, substantially outperforming previous state-of-the-art (SOTA) techniques. We explore different ways to encode geometric information into the network. Our experiments show that while these methods may enhance performance, their impact is minor compared to utilizing improved generative models. Moreover, we introduce a novel NVS training scheme that utilizes single-view datasets, capitalizing on their relative abundance compared to their multi-view counterparts. This leads to improved generalization capabilities to scenes with out-of-domain content.