loseless jbig2 compression changed to ccitt about ocrmypdf

[jbarlow83]

Sadly, Ghostscript converts CCITT to JBIG2 silently and without warning, even if the JBIG2 encoding was produced by another program. JBIG2 encoding is patented (I think). I use Ghostscript to produce the final output PDF, and found out about this misfeature of theirs only after I implemented monochrome to JBIG2 encoding for ocrmypdf in an unreleased branch.

I use Ghostscript to do PDF/A conversion. Would you give up PDF/A for JBIG2?

(Aside: Make sure you are using JBIG2 in lossless mode. Lossy mode is dangerous because it can cause character substitution errors.)

[ppaspp]

Personally, I would prefer JBIG2 over PDF/A because the compression differences to CCITT are big. But I can image that others prefer PDF/A ...

[jbarlow83]

The new option in v4.2 --output-type pdf lets you skip PDF/A and use PDF. If you don't specify any settings that alter the image it will pass JBIG2 through without modification.

(Any settings will cause the image to be modified and trigger CCITT transcoding: --deskew --clean-final --oversample --pdf-renderer tesseract --output-type pdfa --force-ocr)