I have been using the SR300 development kit on a windows 10 platform to perform face tracking. In general this has been working well. I have switched over to using a bare SR300 module with the same setup (same computer, same custom program using SDK 2016 R2, same location). However, the face tracking and more specifically the z values of the facial landmarks don't seem to be as stable as with the development kit. When we look at the tracking range of the two cameras, the SR300 dev kit will track a persons face up to 1.2m while the bare SR300 module seems to stop tracking at .8 or .9m
We have tried this with multiple modules and they all seem to behave the same way.
Any suggestions on how to resolve this?