Infrared (IR) spectroscopic imaging allows the recording of molecular and structural properties of complex samples without the need for prior information on composition, dyes or stains. The spatial and spectral domains are coupled, presenting both opportunities and challenges for recording high-content, high-fidelity data that determine image quality. We first provide an optical theoretic and information theoretic approach to present a roadmap for the development of optical microscopy-based imaging. We show how the spatial quality is related to spectral data and provide explicit relationships to devise imaging configurations at performance limits. We show how molecular content over large areas can be recorded by new high speed microscopy configurations and present rigorous approaches to extending molecular understanding with linear and circularly polarized light. Finally, we present a detailed theoretical understanding of image formation in nanoscale spectroscopic imaging using an atomic force microscope in contact mode. A new method, based on null deflection of the cantilever, is introduced to provide high quality data.