I've been bothered by this issue for a while, but this question (Should I normalise image pixel wise for a pretrained model) in particular made me think that this might be the case.
The question can be summed up to "should I normalize images pixel-wise?", but the asker had the misfortune of adding "All I can think is ... Am I missing something?", which looks like another question if you don't read it.
And it has 4 votes to close as "Needs more focus" (i.e. multiple questions in one).
Am I in the wrong here? Do this question need more focus? If not, is there a larger issue here, and what can we do?