ConvertSensorToGeoPoint Input image

Dear all,
I used OTB to orthorectify Quickbird images and I got great results!
However, I discovered the application ConvertSensorToGeoPoint and I understand that reads the information of input image and solves rpc using the input image coordinates (without height h, right?). If I give as input image the orthorectified image, using first the RefineSensorModel, in this application are applied the 5 parameters of adjustment? If yes, are these parameters applied to the image coordinates or geographical coordinates?
Thank you