Generalizable Neural Human Renderer

TL;DR

This paper introduces GNH, a neural human renderer that generalizes to unseen subjects without test-time optimization, using explicit body priors and a CNN-based renderer for photorealistic results.

Contribution

The paper presents a novel generalizable neural human rendering method that eliminates the need for test-time optimization, leveraging explicit body priors and multi-view geometry.

Findings

GNH achieves a 31.3% improvement in LPIPS over state-of-the-art methods.

It produces photorealistic, animatable human renderings for unseen subjects.

The method operates without test-time optimization, enabling real-world applications.

Abstract

While recent advancements in animatable human rendering have achieved remarkable results, they require test-time optimization for each subject which can be a significant limitation for real-world applications. To address this, we tackle the challenging task of learning a Generalizable Neural Human Renderer (GNH), a novel method for rendering animatable humans from monocular video without any test-time optimization. Our core method focuses on transferring appearance information from the input video to the output image plane by utilizing explicit body priors and multi-view geometry. To render the subject in the intended pose, we utilize a straightforward CNN-based image renderer, foregoing the more common ray-sampling or rasterizing-based rendering modules. Our GNH achieves remarkable generalizable, photorealistic rendering with unseen subjects with a three-stage process. We quantitatively and qualitatively demonstrate that GNH significantly surpasses current state-of-the-art methods, notably achieving a 31.3% improvement in LPIPS.