When a useful document arrives as a PDF, the usual choices are to leave it as a frozen file or to retype the parts you actually need. I wanted something in between: keep the headings, quotes, and action items, then drop the result into an Obsidian vault or a Notion page.

Recent findings suggest that that is the job a PDF-to-Markdown converter is good at. I have been using pdftomd for this, and the distinction that matters most is how the file is processed.

For ordinary PDFs that already contain selectable text, and also for Word and HTML, the free local mode runs in the browser. The file stays on your machine, so it is the right default for private notes and internal drafts. Local conversion does not spend cloud credits. It accepts files up to 50 MB and PDFs up to 500 pages; a longer handbook needs to be split first.

In a recent development, the output is GitHub Flavored Markdown: headings become Markdown headings, paragraphs become editable text, and tables follow the usual header / delimiter / data-row format. That is a preview of structure, not a promise of a pixel-perfect copy. Multi-column pages can shuffle reading order, table-like text can flatten into paragraphs, and embedded images are not extracted in local mode. Keep the original PDF beside the note for figures, signatures, and formulas, and check those parts before you publish.

It has been revealed that if local extraction returns no readable text, or the file is a photo, a scan, or a difficult layout with tables and formulas, switch to high-accuracy OCR. That path is a cloud job, and it uses page credits. Cloud conversion accepts files up to 10 MB and PDFs up to 100 pages, so a longer scan should be split first.

According to recent reports, cloud jobs delete the original upload after processing and keep the generated result available for download for 24 hours. If the document is sensitive and already has selectable text, stay in local mode.

After conversion I scan the headings first, then fix tables, figures, and formulas against the source PDF. That is usually faster than rewriting the document from scratch. From there, the Markdown can be copied, downloaded as a single file, or packed into a ZIP when you convert a batch.

In a recent development, if you want to try the same path, the converter is at https://pdf-to-md.app. Start with a text PDF in local mode, and only turn on OCR when a scanned page actually needs it.

β†’ Read original article