Hi Tyrell,
So I really was on the right track in my answer! (If I understand Juan correctly...)
This is the important sentence:
" For reduction screen zoom ratios (1:2, 1:3, ..., 1:100) STF is always applied to the subsampled data, that is, to the pixel data interpolated at the screen representation scale. However, when the same transformation is applied via HT, it is evaluated pixel-by-pixel on actual image data. This leads to small differences because input pixel values are different in each case."
So the pixel data is interpolated at the screen representation scale. It isn't that the HT process is "doing" anything other than *changing the data". When you permanently stretch your image using HT you are changing the values- and when these new values are interpolated for the zoomed out rendering (using STF) they yield a different result than the original image displayed zoomed out.
Now... if I might clarify something. If I read Juan's note correctly The interpolation is always in place. However, there is a "fast" (sparse) interpolation scheme that is likely qualitatively (visually) similar to the full interpolation- but renders the image faster. Turning off this feature, if I understand, doesn't really make the behavior you are seeing go away.
With respect to you not noticing... assuming something about this interpolation has not changed- then it is literally the data you happen to have on hand (perhaps with lots of small stars). I really haven't noticed the behavior before... but indeed it likely has been there.... here you will see a similar kind of starry field with the same effect. Comparing the images at something closer to the real resolution makes this moot. There is a method of processing images that I employ that does not rely on your particular concern (of comparing original images to the stretched one in order to make important decisions). But that is a longer conversation.
-adam
