POP3 , raw data different from key frame

Today I tried to scan a piece of 17 mm thick wood board standing upright on the turn table and experiment with different kinds of features placed around it. I did not get any tracking loss but the accuracy of the model is not great, the top of the board is thicker than the bottom by about 0.17 mm in the model. I have the feeling that this is more due to frame stitching error in the fusion process than the accuracy of the scanner itself because I did not see this kind of error when scanning a cylinder of the same height and surrounded by the same set of features ( small square blocks ).

Not sure if this indicates anything but upon examining the raw data and the key frame views of the scan, it is found that the starting and ending frames do not meet in the raw data view but perfectly in the key frame view. Is that something normal ?

The frames are aligned while scanning so even if one individual frame don’t look correctly it was already partially cleaning while scanning especially when scanning using continuous mode , some frames are partials because who needs 18 frames of the same side .
It is actually good thing because other wise you get too many overlapped points that are not welcome , the fusing process eliminates more of the frames, leaving most needed points to compete the scan .

Raw frame cells are not organized frame cells ( points) the preview of raw scan ( outside editor) is only estimate preview as well . The fused scan is the organized pointcloud ( no frame cells )