An image processing device for realizing more realistic pictures of explosions in video game devices and the like. Objects displaying such pictures of explosions are formed of spherical polygons (R1, R2, R3, . . . ) and planar polygons (S1, S2, S3, . . . ). Pictures of explosions are realized by alternately arranging these spherical polygons and planar polygons with the lapse in time. Preferably, pictures of polygons are realized by arranging the spherical polygons in layers on the boundary of the planar polygons.