Position-based SoftBody Skinning Animation

In the field of 3D computer animation,simulating the dynamic effects of softbody such as fat can bring more visual effects for model deformation.The traditional method of simulating softbody deformation is limited by complex physical calculation,which requires expensive time and cost.In recent years,with the development of position-based dynamics,it has been widely used in real-time applications because of its simple and fast physical simulation framework.The position-based dynamics method controls the position of particles by imposing shape matching constraints on the particle system of the model to simulate the deformation of the softbody.In the calculation process,the position dynamics method uses Gaussian Seidel iteration to calculate the approximate solution of the constraint set of the model,that is,the constraint is solved one by one in order.The iterative method of sequential solution leads to a large time cost and restricts the application of positional dynamics method in softbody deformation simulation in real time.In this paper,the softbody animation based on position dynamics is taken as the research object.Several existing constraint solving schemes including Gauss Seidel solver,Jacobian solver and group parallel solver are studied and analyzed in detail.A more efficient GPU Based Constraint parallel solution is realized,which can accelerate the real-time simulation of softbody deformation.The core idea of this paper is to decouple the constraints in preprocessing and then group them.The constraints without coupling relationship are divided into the same group,and the constraints in the same group can be executed in parallel through GPU.The serial mode is adopted between groups.In addition,the layered voxel model is used to further improve the problem of soft deformation.Specifically,the main work of this paper includes:· A graph coloring based constraint decoupling grouping scheme is proposed.In this paper,constraint grouping is considered as graph coloring problem.Constraints without coupling relationship are divided into groups with the same color.Due to the complex coupling between constraints,the grouping result of direct coloring is poor and the number of groups is large.Because each group calls the GPU parallel computing process once,and the group to group is a serial process,the number of constrained groups will affect the parallel efficiency.In order to reduce the number of GPU calls and improve the overall parallel efficiency,this paper inserts additional particles into the constraints to reduce the coupling degree between constraints,so as to reduce the lower bound of the coloring of the constraint graph,and control the number of coloring,that is,the number of constraint groups,to the custom q.· The layered voxel model is used to improve the deformation quality of the softbody.In order to improve the problem that the model is too soft in some cases,the voxeled particles are further divided into bone layer,muscle layer,fat layer and skin layer.By adjusting the stiffness coefficient of different layers,the soft and hard degree of deformation of different layers can be controlled to make the whole deformation more smooth and natural.· The skin method of secondary deformation is used to simulate soft deformation.In this paper,the real-time deformation process is divided into two stages: the first stage is the primary movement,and the particles in the skeleton layer move directly according to the linear skinning algorithm;the second stage is the second movement,because the movement of particles in the skeleton layer will cause the change of constraint state,so the particles in other levels can obtain new moving positions through the constraint iterative solution of PBD framework.The combination of the two deformation methods can simulate the elastic deformation similar to fat shaking and improve the skin quality.