Once you start using Neat, it's difficult to give up.
There really is *minimal* useful image information loss if used correctly. Ran it overnight on a 500mb large format image. Can't go back to the original, as although the grain and noise is not great (Velvia scan) , the Neat version is much better. And fine detail has been preserved: that is the key to Neats success. I think you can mistake grain for information. You need to acclimatise to the cleaned image.
I am finding running the automatic optimisation works well for most images.
Quentin