Hi OCC Visualization Team,
I found that the selection in OCC is in 2D graphic space, so when switch selection mode or change the view camera, the AIS_Shape will use
void StdSelect_ViewerSelector3d::Convert (const Handle(SelectMgr_Selection)& theSel)
to convert the object world coordinate from 3d to 2d space, there are lots of matrix computation involved, and also need space to store the 2d data in the
I think it could generate these sensitive entities 2d data when the mouse move over a shape's bounding box, then generate the sensitive entities 2d data by the selection mode(vertex, edge, face, .etc.) according to the TopoDS_Shape dynamically. This could reduce the conversion(matrix computation) time.
Personal opinion, if wrong, please correct me.