Research On The Correlation Between Human Motion And Garment Deformation In Garment Animation

Garment animation simulation is one of the hot topics in the field of computer animation. In order to solve the contradiction between the computational efficiency and the effect of garment deformation, it is usually necessary to construct a multi-precision grid model to ensure the animation efficiency under the premise of keeping the deformation details. How to divide garment areas effectively according to the degree of garment deformation and guide the construction of the multi-precision model is the main problem that needs to be studied.In garment animation, human motion is one of the important factors affecting the dynamic effect of garment. Especially the changes in the attitude of each joint,directly affect the different areas of garment fold generation and elimination. The degree of garment deformation of each area is related to the degree of joint bending closely. The higher degree of joint bending, the higher deformation degree of garment area driven by joint. In the end, this paper analyzes the correlation between them, predict the degree of garment deformation under different motion, and divide different deformation areas of the garment accurately, provide the basis for the grid precision setting of each region.In this paper, firstly, we propose a method expressing the posture of the human body, and describe the movement amplitude of a certain joint in a certain direction by using the bending degree and bending direction of the human joint. Secondly,according to the similarity of garment deformation,this paper divides the garment into different characteristic areas, and dataizes the deformation degree of each area as the deformation characteristics of garment. On this basis, based on six different types of garment, select the rich human body movement type, generate the corresponding garment animation instance data. Finally, for the large number of garment animation data, BP neural network, random forest, generalized regression neural network and support vector machine model are used to study the relationship between human motion and garment deformation in garment animation. By analyzing and comparing the prediction effect of different machine learning models under different training samples, random forest model is put forward, and the deformation characteristics of garment are accurately predicted based on human motion characteristics.