PDF to MD extracts a PDF’s text into a UTF-8 Markdown file. It is a starting point for editing prose, not a reconstruction of the original page layout. Use it for a PDF with a usable text layer, then review headings, lists, reading order and any page markers before treating the Markdown as a finished document.
Check whether the PDF contains selectable text
Open the source in a PDF viewer and try selecting and copying a sentence. If you can copy meaningful text, the document may contain a text layer this converter can read. Selection is a useful first check, though damaged encoding or unusual layouts can still produce poor extraction.
An image-only scan contains visible words as pixels rather than PDF text. This converter does not run OCR on those images. Some scans have already received OCR and therefore include a selectable text layer; the converter can work from that layer, but its accuracy depends on the text already present.
The default file limit is 20 MiB for one PDF. Check the upload area for the current limit and use an unencrypted document. File complexity or an invalid PDF can still prevent processing below the byte limit.
Convert a short report before a larger file
Start with a short document whose text and order you can check. Our controlled example had two pages. Each page contained an uppercase heading, a sentence of selectable text and two dash-prefixed review items. Set:
Heading Style: ATX (# style)
Page Separation: Horizontal Rule (---)Convert and download the Markdown. The checked example extracted both pages and produced a section beginning like this:
# REPORT TITLE
This sentence contains selectable text from the PDF.
- Review the extracted text
- Correct headings and page markers
-- 1 of 2 --The heading and text came from the source, while the page marker came from the extraction library. The example retained the list lines, but it also needed editorial cleanup. Remove a marker if it is not useful in the document you are preparing.
Choose headings and review the inferred structure
ATX and Setext are formatting choices
ATX uses one or more # markers. Setext uses an underline for the two highest heading levels; lower inferred levels still use ATX markers. With Setext selected, the checked source’s uppercase heading became:
REPORT TITLE
============The converter estimates heading levels from text. Uppercase short lines can become top-level headings, and short capitalized lines can be treated as headings even when they were ordinary labels in the PDF. This is a heuristic, not an interpretation of the author’s intended document outline.
Read the Markdown and correct mistaken headings. A table caption, running header or short sentence may need to become plain text. When combining several extracted sections, make the heading hierarchy consistent rather than retaining every inferred top-level heading.
Page separation depends on extracted markers
The Horizontal Rule setting inserts separators when form-feed markers are present. Those markers are not guaranteed for every PDF. In our checked two-page file, the output contained -- 1 of 2 -- and -- 2 of 2 -- labels rather than form-feed-based rules. Selecting No Separation did not remove those textual labels because they were part of the extractor’s returned text.
Do not rely on the setting to recover exact page boundaries or remove every page-number label. Check the content and remove repeated headers, footers and extraction markers manually where needed.
Handle scans without claiming OCR conversion
An image-only PDF can produce an error or little useful text. In a checked one-page image-only sample, conversion returned only -- 1 of 1 --; it did not recognize the visible heading or sentence in the picture. A successful download therefore does not prove that the scan’s words were extracted.
For an image-only source, use a separate OCR workflow first. Zutilium’s Image to Text accepts PNG, JPG and WebP images or an unencrypted PDF with up to ten pages, within its 20 MiB upload limit. Choose English, Arabic or English + Arabic to match the printed text, then download the plain UTF-8 TXT result and correct recognition errors before formatting it as Markdown. For a longer PDF, prepare smaller files with Split PDF first. OCR does not recreate the scan’s tables or layout, or turn the download into a searchable PDF.
Review information that text extraction leaves behind
This converter does not export embedded pictures or reconstruct Markdown tables and hyperlink targets. Multi-column pages can also need reading-order cleanup. Compare the output with the source, especially near tables, footnotes, page transitions and words split across lines.
To continue editing and previewing the draft, open it in Markdown to Word or Markdown to PDF. Those are separate export workflows with their own formatting rules and limits; they do not repair extraction errors automatically.
The PDF is uploaded to the server for processing. After a successful conversion, uploads are scheduled for deletion after about one minute and downloads after about ten minutes; periodic cleanup means deletion is not immediate. Save the Markdown promptly and keep the original PDF for comparison and future corrections.