I'm trying to find an efficient way to determine the leftmost and rightmost vertices of an object from the camera using Geometry nodes, as shown in red here :
Note that the 3D bounding box is irrelevant, as you can see from the top view here (the leftmost point from the camera is not the same as the left bound in 3D :
My goal is to get 2 vector output sockets for these 2 positions.