I found this part interesting:

There are also other documents that appear to simulate a scanned document but completely lack the “real-world noise” expected with physical paper-based workflows. The much crisper images appear almost perfect without random artifacts or background noise, and with the exact same amount of image skew across multiple pages. Thanks to the borders around each page of text, page skew can easily be measured, such as with VOL00007\IMAGES\0001\EFTA00009229.pdf. It is highly likely these PDFs were created by rendering original content (from a digital document) to an image (e.g., via print to image or save to image functionality) and then applying image processing such as skew, downscaling, and color reduction.