We propose a scheme that integrates a natural illumination diffusion model into an inverse rendering framework, allowing sampling of realistic illumination and material explanations while addressing the ambiguity problem.
Inverse rendering, the process of inferring scene properties from images, is a challenging inverse problem.The task is ill-posed, as many different scene configurations can give rise to the same image. Most existing solutions incorporate priors into the inverse-rendering pipeline to encourage plausible solutions, but they do not consider the inherent ambiguities and the multi-modal distribution of possible decompositions.In this work, we propose a novel scheme that integrates a denoising diffusion probabilistic model pre-trained on natural illumination maps into an optimization framework involving a differentiable path tracer. The proposed method allows sampling from combinations of illumination and spatially-varying surface materials that are, both, natural and explain the image observations. We further conduct an extensive comparative study of different priors on illumination used in previous work on inverse rendering.Our method excels in recovering materials and producing highly realistic and diverse environment map samples that faithfully explain the illumination of the input images.
We first pre-train a DDPM that generates realistic environment maps unconditionally. Then, given input images and geometry, we set up a series of denoising processes.The gradient from the rendering loss is used to optimize materials and gets incorporated into a posterior score function that enforces the DDPM to generate a natural environment map that faithfully explains the input images.
Results
@article{lyu2023dpi,
title={Diffusion Posterior Illumination for Ambiguity-aware Inverse Rendering},
author={Lyu, Linjie and Tewari, Ayush and Habermann, Marc and Saito, Shunsuke and Zollh{\"o}fer, Michael and Leimk{\"u}ehler, Thomas and Theobalt, Christian},
journal={ACM Transactions on Graphics},
volume={42},
number={6},
year={2023}
}