Online Neural Path Guiding with Normalized Anisotropic Spherical Gaussians
This work addresses rendering efficiency for computer graphics applications, presenting an incremental improvement in importance sampling techniques.
The paper tackles the problem of slow variance reduction in physically-based rendering by proposing an online framework that learns spatial-varying density models using a small neural network with stochastic ray samples, achieving high-quality images with limited computational resources.
The variance reduction speed of physically-based rendering is heavily affected by the adopted importance sampling technique. In this paper we propose a novel online framework to learn the spatial-varying density model with a single small neural network using stochastic ray samples. To achieve this task, we propose a novel closed-form density model called the normalized anisotropic spherical gaussian mixture, that can express complex irradiance fields with a small number of parameters. Our framework learns the distribution in a progressive manner and does not need any warm-up phases. Due to the compact and expressive representation of our density model, our framework can be implemented entirely on the GPU, allowing it produce high quality images with limited computational resources.