OCR seems correct, textlayer full of blank characters

[ReaderGuy42]

The OCR process itself seems correct, which I verify by spot checking the PDF via the generated word list.

However, in processing the PDF to include the OCR'ed text either as an invisible text layer or visible, the OCR'ed text becomes blank square characters, rendering the process useless.

I have all the spell check and traineddata installed, have switched languages, and also had previously successfully used gImageReader to OCR a document a few weeks ago, on this same system and install.

Exporting to plain text works.
Exporting to ODT works.

Under "Export to PDF" you can select Show Preview, and this shows the desired text overlay.
However, once actually exported to PDF there is no text overlay. If I choose invisible text overlay, I can't select anything except some hyphens, and if I select regular PDF, it just outputs the blank characters as seen in the screenshot.

Any ideas how to solve this?

Thanks!