At the Institut für Computergraphik at TU Braunschweig, we pursue research on the nature of images. Our interdisciplinary work draws on computer graphics, computer vision, applied optics, and visual perception. In the context of PhoenixD, we combine traditional physics forward simulation with powerful, deep learning-inspired differentiable methods to non-invasively determine physical and optical parameters of surfaces and specimen from images. By augmenting optics simulation with modern neural network-based computing approaches and avant-garde imaging techniques we develop novel computational metrology approaches which enable the remote, non-invasive collection of previously inaccessible physical information from imagery.