Couple of days ago I tried the "scanning" feature. Had dozens of receipts that I needed to scan. I was really impressed with document edge detection. I was also able to stack 2 receipts and scan as one document. The edge detection was failing in some cases. However, I was able to correct the edges manually.
I had a document with 2 different color background, the app had some issues detecting the edges on it. It was detecting the only part of the document with same color. But again I was able to manually select the edges and able to scan entire document.
I've also had good luck with "scannable" in the past (acquired by evernote, but still works without evernote). One nice thing about their UI is that as soon as it gets a good lock, it acquires and processes the image, and then moves on to the next one, making it really easy to quickly scan a stack of papers into a single PDF.
Can they improve the resolution of the scanned document by exploiting that multiple images are taken from different angles?