I can understand the rolled corner theory, and there even appears to be a slight shadow of the roll evident in the top picture. The top version is a photo of the picture, while the bottom verision is a scan of the picture.
I tried to reproduce the effect here, it's not a perfect example but I think it shows that the rolled corner theory is quite possible especially since we are dealing with scans and pictures and are unable to hold the documents to examine in person.
I stressed the corners of a business card. I took a picture of the card first, this is the bottom image. Then I tried to uncurl the corners and placed it on a flatbed scanner and scanned the card second, this is represented in the top image.
Same card, images taken mere minutes apart and there is a world of difference in their appearance.