I don't know exactly how JR generate PDF. But, we could learn something from this article https://github.com/pts/pdfsizeopt/releases/download/docs-v1/pts_pdfsizeopt2009.psom.pdf
We could generate a bounch of PDF with fonts and images and try to compress them with ps2pdf and pdfsizeopt utilities and see if we get something compressed. Eventually with JRIO users could create such compression microservices.
Main idea here is to adapt image sizes to the page, remove duplicated images, use only needed fonts, compression.
Why this could be useful?
This could reduce pdf size -> bandwidth and system load
Veaceslav Chicu
