Are we thinking too much about how a user will interpret a photograph?

Here's something interesting that a friend mentioned while taking a look at a few websites I had mentioned.

It's commonplace nowadays (with accessibility as rife as ever) to place a darkened overlay over the image before text goes on top, so the text is easier to read. It does make sense. You can't read small(ish) text over the top of an image with the same colours in it, it becomes difficult on the eye.

This, however, is not an accurate representation of the photograph. It is true. It's dark, looks depressing and oppressive, whatever word you want. Now, in some instances, a white car for example (close up) could very well have black text on it without such a need for a protective layer, but for most instances, you will need something to help the user read the text.

We will get to my point ;-) Are we just thinking this too much here? Will, a user look at a series of hero images for example or panel images on a website and think, "Woah, these images are dark, I'm depressed, not like the ones you see in art galleries even though we don't go to art galleries". "They are not as bright as the ones that my grandmother showed me from her photo album she took with her disposable camera".

Do you see my point? We need to do 'something', so IF we are to have text on an image (sure we could just put the text below the image) is it perfectly reasonable to darken it so we do not fail accessibility checks?