projectPixelTo3dRay works differently in C++ and Python APIs

[francofusco]

Hello, I just wanted to point out that the API for the PinholeCameraModel is slightly different in Python and C++ regarding the method projectPixelTo3dRay. In the Python API, it returns a unit vector while in C++ it returns the 3D point lying on the ray and having z=1.

Having a look at the documentation of the two languages the difference is expected and documented, with the C++ documentation saying:

In 1.4.x, the vector has z = 1.0. Previously, this function returned a unit vector.

However, wouldn't it make sense to have the same behavior also in Python? Perhaps the normalization can be left as an additional feature by adding a boolean parameter normalize.

[arpitpatel901]

+1

[ranjitkathiriya]

vector = self.camera_model.projectPixelTo3dRay((X,Y))
ray_z = [el / vector[2] for el in vector]