An image generation apparatus includes first and second light sources, an image sensor, a mask, and a processing circuit. The image sensor acquires a first image when the first light source is energized and acquires a second image when the second light source is energized. The processing circuit obtains a pixel value based on a direct light component or a pixel value based on a component other than the direct light component corresponding to a focal point by using pixel values of first pixel regions of the first image, pixel values of second pixel regions of the second image, the ratios of the direct light and the light other than the direct light from the first light source, and the ratios of the direct light and the light other than the direct light from the second light source and generates a cross-sectional image of a substance on the focal plane.