Our eyes are much better at discerning meaningful shapes than computers are. That's why CAPTCHAs work.
Downsampled? A new term for "Deus ex machina? Yeah, when I don't recognize something, I make the resolution four times worse, rather than just leave it alone.
It's the term Adobe uses (and others too, I imagine). I could have sworn it appeared in a previous post of mine. Anyway, first, the point is that the computer didn't recognize it. And second, because of that, it treated it the same way it handled the rest of the background--which was to "downsample" it, i.e., lower its resolution.
If you are making a copy, the computer doesn't need to know what it is, n'est-ce pas?
No, it doesn't. It would have been better if they'd turned off whatever routines caused these anomalies and just scanned the dam thing as a TIFF.
(It still doesn't explain the "Halos" around each letter either.)
Actually, I think that's an argument for the whole thing being a program process rather than intentional copying and pasting. If the latter, there would be no reason for halos, and certainly no reason that when you hid the text "layers," the background would be white behind them. In fact, I don't see any way to explain that in a copy-and-paste scenario.
I agree, but it appears to be White House policy to post scans as PDF files, as that's the only format I can find them in. The most likely reason for this is compliance with accessibility standards, since PDFs can easily be given alternate text and tags that assist with the use of screen readers.
Our eyes are just optics. Interpretation is done in the brain. In the case of the artifacts, it appears no interpretation is occurring. It looks like a simple string contrast sorting algorithm.
It's the term Adobe uses (and others too, I imagine). I could have sworn it appeared in a previous post of mine. Anyway, first, the point is that the computer didn't recognize it. And second, because of that, it treated it the same way it handled the rest of the background--which was to "downsample" it, i.e., lower its resolution.
Again, I don't think it's "recognizing" anything. On the Document provided by Kleon, there are spots on the scan rendered in fine resolution two color, and there are other spots rendered in low resolution multi-color. The difference seems to be one of contrast having nothing at all to do with shape.
No, it doesn't. It would have been better if they'd turned off whatever routines caused these anomalies and just scanned the dam thing as a TIFF.
Yes.
Actually, I think that's an argument for the whole thing being a program process rather than intentional copying and pasting. If the latter, there would be no reason for halos, and certainly no reason that when you hid the text "layers," the background would be white behind them. In fact, I don't see any way to explain that in a copy-and-paste scenario.
I explained it in a previous response. Assuming software designed to recognize text and substitute letters, replacing an "A" with an "e" (in a specific position on the document) would require an "A" sized space to be erased, or the top half of the "A" would show above the "e" which replaced it. If the software is designed to do this on a regular basis, one of the subroutines would be to always erase a capitol letter sized space.
At the moment, Kleon's document has convinced me that what I originally took to be evidence of pasting from two different formats is more likely the result of a compression algorithm being used to reduce file size. The software that created it appears to be optimized for saving file size rather than accurate reproduction. It appears that you were right and I was wrong.
Ah well, at least one mystery put to bed. (As far as i'm concerned.) Now I will have to reassess some of those other artifacts in light of this realization.