renderViewTesselatedSphere： single view pointcloud can‘t be transformed to the model coordinate

1、I obtain pointcloud of the .stl model using "pcl_mesh_sampling";
2、I obtain a single view pointcloud from the .stl model using the method renderViewTesselatedSphere;
3、I hope the two pointcloud are in the same coordinate，but they are not coincident！

below is my code：

    vtkSmartPointer<vtkPolyData> polydata = vtkSmartPointer<vtkPolyData>::New();
vtkSmartPointer<vtkSTLReader> readerQuery = vtkSmartPointer<vtkSTLReader>::New();

readerQuery->SetFileName(argv[1]);
readerQuery->Update();
polydata = readerQuery->GetOutput();
polydata->GetNumberOfPoints();

float resx = 512;
float resy = resx;
std::vector<pcl::PointCloud<pcl::PointXYZ>, 
    Eigen::aligned_allocator<pcl::PointCloud<pcl::PointXYZ> > > views_xyz;
std::vector<Eigen::Matrix4f, Eigen::aligned_allocator<Eigen::Matrix4f> > poses;
std::vector<float> entropies;
pcl::visualization::PCLVisualizer vis;
vis.addModelFromPolyData(polydata, "mesh", 0);
vis.setRepresentationToSurfaceForAllActors();
vis.renderViewTesselatedSphere(resx, resy, views_xyz, poses, entropies, 0, 90, 1, TRUE);

pcl::PointCloud<pcl::PointXYZ> views_cloud;
pcl::transformPointCloud<pcl::PointXYZ>(views_xyz[i], *transformed_cloud, poses[i]);

Start by comparing the point cloud from pcl_mesh_sampling with views_xyz produced by renderViewTesselatedSphere, including the poses passed to transformPointCloud. Read the vtkSTLReader and PCLVisualizer setup shown in the report, then determine whether the two outputs use different coordinate frames. Done means documenting the cause and the transformation or API behavior needed for coincident point clouds.