A Dataset of AI-generated poscasts by @jilltxt

https://dataverse.no/dataset.xhtml?persistentId=doi:10.18710/RNTF9H

Certainly a new way of looking at your digital content -- especially the blank Untitled.pdf ;)

#DH #DigitalHumanities #Dataset #ContentAnalysis #Authenticity #LLM #GenAI #WTFPDF

PDF -> Markdown 100 pages per second.

I hate that it's for #AI but I also think it will have connotations for #digipres one way or another.

#wtfpdf

https://social.lansky.name/@hn50/115255089991079589

pdfalyze --help
(https://github.com/michelcrypt4d4mus/pdfalyzer)

outputs:

"Explore PDF's inner data structure with absurdly large and in depth visualizations. Track the control flow of her darker impulses, scan rivers of her binary data for signs of evil sorcery, and generally peer deep into the dark heart of the Portable Document Format. Just make sure you also forgive her - she knows not what she does."

#wtfpdf

On the effects of the useful

`mutool clean`

command to "repair" PDFs.

If you take this PDF https://openreview.net/pdf?id=CSJYz1Zovj and apply the command to it... the 27-page annex is cut off from the "cleaned" output (still in the PDF, but unreferenced, so not displayed).

So use it with care!

#wtfpdf

Well, I've finally made a blog post, but on the #OPF website!

https://openpreservation.org/blogs/validation-ok-they-said-fixing-the-rendering-of-a-so-called-valid-pdf

I'm walking you through the most complex (out of 2) PDF repair processes I've made. Any input is welcome!

#digipres #wtfpdf

Today's #wtfPDF moment: #PDFs with images that are encoded as #JPEG, where the JPEG data stream in turn is ascii85 encoded.

WHY?!? (The ascii85 encoding only inflates the JPEG data streams by 25% and doesn't offer any benefits).

https://github.com/KBNLresearch/pdfquad/issues/39

The smallest (valid) #PDF in the world is here:

https://pdfa.org/download-area/smallest-possible-pdf/smallest-possible-pdf-1.0.pdf

(source: https://pdfa.org/the-smallest-possible-valid-pdf/)

#digipres #wtfpdf