So basically, you're first going to increase my cognitive load with a big flashy unrelated photo, and then you're going to blur and dropshadow and opacity-gradient it just so much that I can kind of read the text again? What for?
Fullscreen photos are nice, but they're just a picture. Very nice blurred bridge there in the background, but how is that going to sell me a route planning app? I feel like a lot of these examples are aesthetic-only designers gone wild without any place for feedback from their peers.
Is it good when it's pretty and you can just about read the text without squinting your eyes? No! It's acceptable at best. But if you try, and I bet if you actually do user tests, then you'll find that you want to ditch the image. Or at least put a big black box around the text (admittedly, also listed as an example).
Can we stop the cognitive overload already? Don't make me think! Don't make me search past the blurs and the street lights and the flowers to find the text I'm supposed to be reading.
(unless, of course, you sell a blur plugin, a street light or flowers)