Depth-Guided Bundle Sampling for Efficient Generalizable Neural Radiance Field Reconstruction

TL;DR

This paper introduces a depth-guided bundle sampling method for neural radiance fields that accelerates high-resolution rendering by grouping rays and adaptively sampling based on scene complexity, achieving faster rendering with improved quality.

Contribution

It proposes a novel depth-guided bundle sampling strategy with adaptive sampling for efficient, high-quality neural radiance field reconstruction, outperforming existing methods.

Findings

Up to 1.27 dB PSNR improvement on DTU dataset

47% increase in FPS during rendering

Achieves state-of-the-art quality and 2x faster rendering

Abstract

Recent advancements in generalizable novel view synthesis have achieved impressive quality through interpolation between nearby views. However, rendering high-resolution images remains computationally intensive due to the need for dense sampling of all rays. Recognizing that natural scenes are typically piecewise smooth and sampling all rays is often redundant, we propose a novel depth-guided bundle sampling strategy to accelerate rendering. By grouping adjacent rays into a bundle and sampling them collectively, a shared representation is generated for decoding all rays within the bundle. To further optimize efficiency, our adaptive sampling strategy dynamically allocates samples based on depth confidence, concentrating more samples in complex regions while reducing them in smoother areas. When applied to ENeRF, our method achieves up to a 1.27 dB PSNR improvement and a 47% increase in FPS on the DTU dataset. Extensive experiments on synthetic and real-world datasets demonstrate state-of-the-art rendering quality and up to 2x faster rendering compared to existing generalizable methods. Code is available at https://github.com/KLMAV-CUC/GDB-NeRF.