The invention discloses a building method for a candidate set used for detecting a
human body-clothes conflict which mainly includes two steps of building a mother candidate set and a sub candidate set. Aiming at each
human body primitive, the invention takes a clothes primitive which can conflict with the
human body primitive and is apart from the human body primitive in a certain range as a composition element of the candidate set according to the neighboring position relation of the corresponding primitives on a human-clothes model and the characteristic of small
relative displacement when moving, wherein, the building of the mother candidate set is finished at a pre-treating stage when the human body clothes is in a static state, a larger value can be selected as the distance range so as to avoid detection leakage; the sub candidate set is built or updated at an
animation stage, a smaller value is selected as the distance range so as to reduce the detecting times among the primitives and improve the efficiency of the conflict detection. The detection process of the conflict between the
human body model and the clothes model in a
virtual space can be quickened by utilizing the method; the building method of the invention can be applied to relevant applications such as virtual fitting and clothes
animation, etc. and has great help for quickening the calculation efficiency of conflict detection.