An object is to efficiently generate virtual viewpoint images in different image formats. The image generation system includes a plurality of rendering modules. Then, virtual viewpoint information indicating a virtual viewpoint, for generating a virtual viewpoint image adapted to a predetermined image format, is converted into a plurality of pieces of virtual viewpoint information which indicate a plurality of virtual viewpoints, based on performance of a plurality of rendering modules. Then, based on the plurality of pieces of virtual viewpoint information after being converted, contents of rendering that should be executed are allocated to at least part of the plurality of rendering modules. Then, a virtual viewpoint image adapted to the predetermined image format is generated by using results of rendering processing by the at least part of the rendering modules.