Development and validation of a physically based rendering methodology for celestial bodies

This article outlines a physically based rendering methodology for generating spectral images of planets, moons, and other quasi-spherical bodies. The methodology links physical radiometric quantities such as the collected power or the electron flux to each pixel of the synthetic image. Conversely to standard ray-tracing methods, the image is formed by discretizing the spherical body into longitude–latitude quadrangles, which are projected into the camera and directly gridded to the detector pixels. Despite the high number of sectors undergoing this procedure, smart bounding, efficient sampling, and pruning of the inactive sectors enable fast and efficient computations. Local radiometric and geometric parameters are interpolated from texture maps to enhance the fidelity of lighting and shadows. Different reflection models can be used and mixed together. The output of the renderer is a radiometrically-consistent image reproducing the exact interaction between the light, the object in the scene, and the camera. The methodology is validated against analytical laws and real Moon images acquired in orbit and from the ground. Results reveal very good consistency in luminance, contrast, and structure. The frame rate for a Moon flyby scenario spans over 10 Hz at far range to about 1 Hz at a closer range, consistent with current limits in CPU-based renderers. Thanks to its radiometric consistency, the methodology proves effective for tuning the camera exposure time brackets and is essential for the operation of hardware-in-the-loop optical stimulators. A tool implementing the methodology is released to the public as an open-access rendering application.