Physics-based inverse rendering aims to jointly optimize shape, materials, and lighting from captured 2D images. Here lighting is an important part of achieving faithful light transport simulation. While the environment map is commonly used as the lighting model in inverse rendering, we show that its distant lighting assumption leads to spatial invariant lighting, which can be an inaccurate approximation in real-world inverse rendering. We propose to use NeRF as a spatially varying environment lighting model and build an inverse rendering pipeline using NeRF as the non-distant environment emitter. By comparing our method with the environment map on real and synthetic datasets, we show that our NeRF-based emitter models the scene lighting more accurately and leads to more accurate inverse rendering. Project page and video: https://nerfemitterpbir.github.io/.

翻译：基于物理的逆渲染旨在从拍摄的二维图像中联合优化形状、材质与光照。其中，光照是实现可靠光传输模拟的关键组成部分。尽管环境贴图通常被用作逆渲染中的光照模型，但本文表明其远距离光照假设会导致空间不变的光照效果，这在实际逆渲染中可能是一种不准确的近似。我们提出将神经辐射场（NeRF）作为空间变化的环境光照模型，并以此构建基于非远距离环境发射器的逆渲染管线。通过在真实数据集与合成数据集上比较我们的方法与基于环境贴图的方法，结果表明，基于NeRF的发射器能更准确地建模场景光照，并实现更精确的逆渲染。项目页面与视频：https://nerfemitterpbir.github.io/。