Hi Sander,
I've just checked this and I'm unable to reproduce this problem; sample navigation works well here.
Could you please send me a .psm file with a problematic DBE instance?
I've noticed that my background model is much improved if I move each sample so it does not include any stars or other structures so I need to check each one.
This shouldn't happen. DBE's pixel rejection should be able to reject pixels from stars and other small and bright image features. If stars are altering your model, the most probable cause is that your samples are too small.
Pixel rejection isn't perfect. Some pixels from a star (especially peripheral star pixels) can remain unrejected. For this reason, if a sample includes a bright star and is small compared to the star's size, the importance (in the statistical sense) of a few star pixels can be relatively large and alter the sample value significantly.
This normally happens more for small stars. Very large stars are totally rejected, so the automatic sample generator doesn't create samples over them.
You can watch the sample graphic over DBE's interface to know what is being rejected. Black pixels are being completely rejected. For example, if a pixel is red on the graphic, it is being rejected in the green and blue channels, but accepted in the red channel.
Of course, there are cases where a sample is incorrectly including too bright and too large objects. Usually this happens when the user creates samples manually. In these cases, those samples must be edited (moved/deleted/resized).
