An image composition method comprises: subjecting a first high-sensitivity image signal generated based on high-sensitivity image outputs from corresponding ones of a plurality of pixel positions in a solid-state imaging element that has imaged a subject, to a first nonlinear conversion, so as to prepare a second high-sensitivity image signal; subjecting a first low-sensitivity image signal generated based on low-sensitivity image outputs from corresponding ones of the plurality of pixel positions, to a second nonlinear conversion, so as to prepare a second low-sensitivity image signal; and subjecting the second high-sensitivity image signal and the second low-sensitivity image signal to weight adding at every common pixel position, over the entire range of an incident light exposure level exhibited by each incident light that entered to each of the plurality of pixel positions, so as to produce a composed image signal of the subject.