Agreement of the CS predictions by the CAGE-Barten model with infocus experimental data is highly satisfactory, not only for the matching of their relative horizontal position (dependent on lateral inhibition at low frequencies and multiple factors at high frequencies) and vertical position (dependent on the level of internal noise), but also for the fidelity shown in changing shape according to the conditions of visualization. For example, CSF tends to to progressively flatten when reducing retinal illuminance, seeming to transform from a pass-band filter at high illuminance to a low-pass filter at levels of illuminance several orders of magnitude lower. This modification, however, is only an artifact due to data presentation with logarithmic abscissae, because CSF is identically zero for zero angular frequency. The CSF peak progressively lowers with retinal illuminance and shifts toward lower spatial frequencies, escaping from the plotting windows.
Online access to SPIE eBooks is limited to subscribing institutions.