Is the preview image you're referring to, the preview in Pano2VR?
I'm not sure what applications use the XMP data or how the XMP data is used. I understand it has some pitch, yaw, roll, heading info, but I'm not sure if all apps use this or only some apps from Adobe and Garden Gnome use it.
Also, why wouldn't the THETA camera embed the XMP data into the image from the beginning? Why is the extra manual step needed?
Is the XMP data new and Ricoh chose not to support it as the default format?
Just curious as to your thoughts.