In an image capturing apparatus, an image sensing device includes a plurality of groups of pixels each including a plurality of photoelectric conversion elements, signals from the plurality of photoelectric conversion elements being readable separately for each photoelectric conversion element via a signal line used in common by each group of pixels. A reading unit performs, on a plurality of groups of pixels, a reading-out operation to reading out a signal as a first signal from part of the plurality of photoelectric conversion elements and a second reading-out operation to mix signals from the plurality of photoelectric conversion elements and read out a resultant mixed signal as an image signal. A generation unit generates one image file including the first signal, the image signal, and defect data indicating a group of pixels for which the first signal is defective while the image signal is not defective.