In a three-dimensional image generation apparatus, a CPU concatenates meshes of a plurality of objects which meet a predefined concatenation condition, and performs in advance vertex attribute transformation on the plurality of objects of which the meshes have been concatenated. The CPU regards as a single object, the plurality of objects on which the vertex attribute transformation has been performed, and transfers a rendering command, which instructs rendering of the plurality of objects regarded as the single object, to a GPU. The GPU renders, at a time, the plurality of objects that are regarded as the single object, upon receiving the rendering command from the CPU.