Probe Software Users Forum

Software => PictureSnapApp (NEW!) => Topic started by: Probeman on February 12, 2018, 04:26:19 pm

Title: A matter of perspective...
Post by: Probeman on February 12, 2018, 04:26:19 pm
Kurt Langworthy and I were playing around with a scan of a sample in PictureSnapApp on our FEI Quanta SEM and noticed something that makes perfect sense in hindsight, but caused us some confusion for a little while until we went "doh!"...

I'll use an empty sample holder scan to illustrate. Basically we had calibrated the sample scan using the fiducial marks that we had scribed on the pin stub holder last week:

(https://probesoftware.com/smf/proxy.php?request=http%3A%2F%2Fprobesoftware.com%3A80%2Fsmf%2Fgallery%2F395_12_02_18_4_17_46.jpeg&hash=9e1ed9c356ae6d1c6fdafa917e6e5693)

After calibration we could click on any spot on the Al pin stub holder and it went to exactly the precise location.  However, when we clicked on a sample mounted on a pin stub, the location it drove to wasn't exactly where we had clicked on the image.

Now some of you have probably already guessed what I'm going mention next, and yes, it's because the samples were at a different z height relative to the plane of the pin stub holder (where the calibration fiducials were scribed). 

And of course, this is because the document camera (like any camera with a lens), distorts the sample image (if there is topography), due to optical perspective.  To see this perspective, look closely at the holes drilled in the sample holder and you can see different sides of the holes due to the perspective of the camera lens:

(https://probesoftware.com/smf/proxy.php?request=http%3A%2F%2Fprobesoftware.com%3A80%2Fsmf%2Fgallery%2F395_12_02_18_4_18_05.jpeg&hash=1c825a670dbfac87937f5b921d3a8159)

In the EPMA this not an issue as the z height is almost always constant and uniform, but in the SEM it's going to depend on the sample.  What this means of course is that if you want a precise sample location when double-clicking a scanned image, one should calibrate the image using calibration fiducials that are at least roughly the same z height as the samples themselves!

This shouldn't be a problem as the calibration points can be any object that can be located both in the scanned image and also in the instrument scanning system.