Putting pdf data into VCs

Hi all!

Many educational institutions issue credentials (transcripts, graduation certificates, etc.) in pdf format, and we have faced an question of how to create VCs using claims in pdfs. Reaching out to the community is anyone has faced a similar issue and if there are standardized way/best practices to:
1) put data from pdfs into VCs while keeping integrity without relian on centralized party​? One option could be storing pdf files in centralized/decentralized servers and including a pointer to a file in a VC, but that would introduce a certain level of centralization.
2)  to reconstruct pdfs from claims in VCs?

For example Modeling Educational Verifiable Credentials report (https://w3c-ccg.github.io/vc-ed-models/#biblio-obs-are-vcs) in section 1.5.3.1 shows the example of including pdf format in a VC, but how can the verifier reproduce a pdf record from the set of values in payload.data?

This is a little different from using pdfs as a container, rather including information from pdfs into VCs.

Thank you very much!
Kristina