Transformation-Grounded Image Generation Network for Novel 3D View Synthesis

TL;DR

This paper introduces a transformation-grounded network for synthesizing novel 3D views from a single image, explicitly modeling geometry and occlusion to improve visual quality and detail.

Contribution

It proposes a new network that predicts pixel flow and visibility maps, combining geometric inference with image completion for improved 3D view synthesis.

Findings

Outperforms existing methods in qualitative and quantitative evaluations

Reduces artifacts like distortions and holes in generated views

Generates high-frequency details while preserving input image features

Abstract

We present a transformation-grounded image generation network for novel 3D view synthesis from a single image. Instead of taking a 'blank slate' approach, we first explicitly infer the parts of the geometry visible both in the input and novel views and then re-cast the remaining synthesis problem as image completion. Specifically, we both predict a flow to move the pixels from the input to the novel view along with a novel visibility map that helps deal with occulsion/disocculsion. Next, conditioned on those intermediate results, we hallucinate (infer) parts of the object invisible in the input image. In addition to the new network structure, training with a combination of adversarial and perceptual loss results in a reduction in common artifacts of novel view synthesis such as distortions and holes, while successfully generating high frequency details and preserving visual aspects of the input image. We evaluate our approach on a wide range of synthetic and real examples. Both qualitative and quantitative results show our method achieves significantly better results compared to existing methods.