The present disclosure relates to a method by which a decoding apparatus performs video coding, comprising the steps of: generating a motion information candidate list for a current block; selecting one candidate from among those included in the motion information candidate list; deriving control point motion vectors (CPMVs) of the current block based on the selected candidate; deriving sub-block-unit or sample-unit motion vectors of the current block based on the CPMVs; deriving a predicted block based on the motion vectors; and reconstructing a current picture based on the predicted block, wherein the motion information candidate list includes an inherited affine candidate, the inherited affine candidate is derived based on candidate blocks coded by affine prediction, from among spatial neighboring blocks of the current block, and the inherited affine candidate is generated up to a pre-defined maximum number.