I was trying to merge a bunch of PDFs from my electric utility provider. I selected all of them, right-clicked, and selected “merge X documents”. Nothing happened except me scratching my head.
To make a long story short, Console (filtering on “DEVONthink”) reported a number of instances of this CGImageCreateCopyByApplyingSoftMask Invalid image alpha followed by WARNING: Text layer of document possibly corrupted by PDFkit framework.
I was able to fix the problem by applying OCR again to each document, but I was curious how this might happen and how to prevent it in the future.