OCR is not the only way that layers appear in a scanned document, there is also MRC.
http://image.unb.br/queiroz/papers/icip08mrc.pdf
One (OCR) creates layers of text from document so readers (like a .pdf reader) can have content that can be searched. MRC (link above) breaks the image into multiple layers to compress it while retaining quality. At that, if you look at the link above, you’ll see a lot of the same things pointed out like the ‘white out’ look some have found.
Neither of these items are isolated, they both occur when scanning documents into a filing system like a Keyfile system.
I suggest reading the link above. The problem is graphic artists only work with documents one way, creating layers themselves. They may never have worked with a Keyfile type system to bring in documents for data storage and don’t realize it also creates layers.
Hey this was settled a couple weeks ago, it is magic, plain and simple. Now I am off to have some free BubbleUp and Rainbow Stew.
Second. Both processes (optimaztion and MRC) retain the pixels comprising this "halo" portion of the scanned image because it's part of the "useful image" area. Where they place them is different. Example. Adobe's optimization will leave the soft edge (halo) pixels on the background, while -according to your link- MRC combines them with the "mask" layer. They don't vanish.
More importantly, neither process has been proven to be involved in this documents creation, and neither will cause this.
Thanks for posting both the MRC and National Review links.
This is a dead horse.
It just doesn’t know that it’s dead... lol.