Abstract: Traditional character modeling and animation rigging methods require a high degree of professionalism and involve the use of a variety of large-scale commercial software.A method to generate 3D character animation from a single image is proposed.First of all,a static 3D model of a person is obtained from the image with the help of an implicit function reconstruction method based on deep learning.Secondly,a template parametric model is used for fitting,and finally the parameter information of the aligned parametric model is correctly transferred to the static model so that it can be freely deformed and driven by animation.In addition,an application system is built upon proposed method with a friendly user interface.Proposed method has been carefully evaluated regarding the quality of model matching,animation generation and character model generation.The experimental results clearly demonstrate the superiority of the proposed technique,compared with mainstream methods.