main/model.py [93:112]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
            for i in range(batch_size):
                global_pose[i][self.root_joint_idx] = torch.eye(4).float().cuda() # use identity matrix in testing stage


        # forward kinematics
        joint_out = []; joint_trans_mat = [];
        for i in range(batch_size):
            forward_kinematics(self.skeleton, self.root_joint_idx, torch.bmm(mesh['local_pose_refined'], joint_rot_mat[i]), global_pose[i])
            joint_out.append(torch.cat([global_pose[i][j][None,:3,3] for j in range(self.joint_num)],0))
            joint_trans_mat.append(torch.cat([torch.mm(global_pose[i][j], mesh['global_pose_inv'][j,:,:])[None,:,:] for j in range(self.joint_num)]))
        joint_out = torch.cat(joint_out).view(batch_size,self.joint_num,3)
        joint_trans_mat = torch.cat(joint_trans_mat).view(batch_size,self.joint_num,4,4).permute(1,0,2,3)
        
        # estimate corrective vector
        pose_corrective, id_corrective = self.skin_refine_net(joint_euler.detach(), self.id_code[None,:].repeat(batch_size,1))
        mesh_refined_xyz = mesh['v'] + pose_corrective + id_corrective

        # LBS
        mesh_refined_xyz1 = torch.cat([mesh_refined_xyz, torch.ones_like(mesh_refined_xyz[:,:,:1])],2)
        mesh_out_refined = sum([mesh['skinning_weight'][:,:,j,None]*torch.bmm(joint_trans_mat[j],mesh_refined_xyz1.permute(0,2,1)).permute(0,2,1)[:,:,:3] for j in range(self.joint_num)])
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -



main/model.py [159:177]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
        for i in range(batch_size):
            global_pose[i][self.root_joint_idx] = torch.eye(4).float().cuda() # use identity matrix in testing stage

        # forward kinematics
        joint_out = []; joint_trans_mat = [];
        for i in range(batch_size):
            forward_kinematics(self.skeleton, self.root_joint_idx, torch.bmm(mesh['local_pose_refined'], joint_rot_mat[i]), global_pose[i])
            joint_out.append(torch.cat([global_pose[i][j][None,:3,3] for j in range(self.joint_num)],0))
            joint_trans_mat.append(torch.cat([torch.mm(global_pose[i][j], mesh['global_pose_inv'][j,:,:])[None,:,:] for j in range(self.joint_num)]))
        joint_out = torch.cat(joint_out).view(batch_size,self.joint_num,3)
        joint_trans_mat = torch.cat(joint_trans_mat).view(batch_size,self.joint_num,4,4).permute(1,0,2,3)
        
        # estimate corrective vector
        pose_corrective, id_corrective = self.skin_refine_net(joint_euler.detach(), self.id_code[None,:].repeat(batch_size,1))
        mesh_refined_xyz = mesh['v'] + pose_corrective + id_corrective

        # LBS
        mesh_refined_xyz1 = torch.cat([mesh_refined_xyz, torch.ones_like(mesh_refined_xyz[:,:,:1])],2)
        mesh_out_refined = sum([mesh['skinning_weight'][:,:,j,None]*torch.bmm(joint_trans_mat[j],mesh_refined_xyz1.permute(0,2,1)).permute(0,2,1)[:,:,:3] for j in range(self.joint_num)])
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -