# PDF to Markdown: Convert Scans, Tables, and Equations

Download a real scanned PDF and Markdown ZIP, inspect text, image paths, equations, and footnotes, then choose extraction settings and open the result in your editor.

To convert a PDF to Markdown in [PDF Craft](https://inkora.oomol.dev/pdf-craft/), upload your file and select Markdown as the output. Once processing finishes, download a ZIP archive containing the document and its image resources. Scans require text recognition, while PDFs that already include a text layer still need their paragraphs, headings, and reading order organized.

Markdown uses headings, paragraphs, lists, and image references to structure content for editing, note-taking, or use in other tools. It does not preserve the PDF's fixed page layout. If you want to read a full book with adjustable text on your phone, [EPUB output](https://inkora.oomol.dev/blog/scanned-pdf-to-epub/) may be a better fit.

## Open a real Markdown result first

Download the [source scan](https://inkora.oomol.dev/blog-assets/pdf-to-markdown-guide/archimedes-source.pdf) and [complete Markdown ZIP](https://inkora.oomol.dev/blog-assets/pdf-to-markdown-guide/archimedes-refined.markdown.zip) to see how the output fits your editing workflow. The sample is page 274 of the 1897 English edition of *The Works of Archimedes*, processed by PDF Craft on November 13, 2026 with **Refined extraction** and translation off. The files preserve the original output; [conversion settings and checksums](https://inkora.oomol.dev/blog-assets/pdf-to-markdown-guide/conversion-record.json) accompany them.

The extracted folder contains `document.md` and `assets`. Move the entire folder into your working directory, then open `document.md`, keeping its resources alongside it.

```text
document.md
assets/
```

The opening paragraph actually contains this source text:

```text
Then, since \(CO = \frac{1}{2} p\) , \(CL\) must be perpendicular to \(PT\) and therefore to the surface of the fluid.
```

The words are editable, while inline symbols retain math markup. The diagram uses this relative image reference:

```markdown
![](assets/58b0a908132d1c1f36a28a957d4483670a6dad83c516c0d1148f9de2a5a43cd3.png)
```

If the image does not appear, check that `assets` remains beside the document, that the image name is unchanged, and that your preview can read local resources. The body marker `[^1]` matches its definition at the end; `## References` introduces the exported notes. This page has no chapter heading or table, so it does not demonstrate whole-book heading extraction or table preservation.

## Separate math-source problems from preview support

The sample's BM equation is exported as:

```text
\[BM = BV\\sin \\theta +PN,\]
```

Some math commands contain extra backslashes, and the footnote text also retains an escaped marker. Compare symbols, subscripts, and operations with the original page, then correct specific occurrences in an editing copy. Do not remove backslashes globally: valid math line breaks and other escapes can be damaged too.

If the source is correct but the equation still does not display, check your editor's math setting and support for the delimiters used in the file. Keep the original ZIP so you can distinguish downloaded output from your later edits. The [equations to Markdown guide](https://inkora.oomol.dev/blog/pdf-equations-to-markdown/) covers further checks.

## Upload the PDF and choose Markdown

Open the [PDF conversion workspace](https://inkora.oomol.dev/pdf-craft/), drag your PDF into the upload area, or select “Choose a file.” Upload one file per conversion and sign in when prompted.

Under “Create a result,” choose “Markdown.”

You can upload either scanned pages or PDFs with an existing text layer. Scanned pages require OCR to recognize text, while PDFs with a text layer often need reading-order correction—especially when copying text from a two-column paper mixes lines across columns.

## Choose Extraction or Refined extraction

“Extraction method” offers two options:

| Option             | What it does                                                                                                                 | When to choose it                                                                       |
| ------------------ | ---------------------------------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------- |
| Extraction         | Recognizes document content and structure without extracting footnotes or refining pages                                     | You only need the main text for editing or notes                                        |
| Refined extraction | Uses AI to further refine document structure and extract footnotes, strengthening links between the text and its annotations | You need footnotes or closer attention to the relationship between text and annotations |

Choose “Refined extraction” when a paper or book contains important footnotes. Because it uses more credits and takes longer, check the [current rates](https://inkora.oomol.dev/recharge/) before starting. If you only need the main text without footnotes, you can start with “Extraction.”

## Turn translation off to keep the source language

If you only need to change the file format, turn off “Translate the document” and check that the plan shows “No translation.” For example, you can keep an English research paper in English while converting it to Markdown. If you want a translated result, enable translation and select a target language.

Select “Confirm plan,” review the source file, output format, extraction method, and translation settings, then select “Confirm and start.” Wait for the upload to finish before starting. Changing settings after a task has started requires a new conversion.

## Download and unpack the complete ZIP

When conversion finishes, select “Download complete file” to save the ZIP archive.

Unpack the archive into its own folder, then open the `.md` file in an editor with Markdown preview. Images are referenced through relative paths, so keep the extracted folder structure intact. Moving the document alone or renaming its image folder can break those references. If images do not appear, follow the [image resource guide](https://inkora.oomol.dev/blog/pdf-markdown-image-assets/).

If the page asks you to recharge before downloading the complete file, settle the outstanding balance and check your download access again. You do not need to repeat the conversion. Note that a file marked as a cropped preview contains only part of the content.

## View tables, equations, and footnotes

Switch to your editor's preview to inspect the document. The text view shows the source markup, while the editor needs support for the relevant syntax to render tables and equations.

* **Tables:** Check that each value sits under the right column heading. Merged cells, multilevel headings, and tables spanning several pages may require further adjustment. Use the original table alongside the [PDF tables to Markdown guide](https://inkora.oomol.dev/blog/pdf-tables-to-markdown/) when refining the result.
* **Equations:** Display depends on your editor's math support. If an equation looks incorrect, check that math rendering is enabled, then compare subscripts, superscripts, and symbols with the original PDF. See the [equations to Markdown guide](https://inkora.oomol.dev/blog/pdf-equations-to-markdown/) for more details.

After using “Refined extraction” for footnotes, compare note markers and note text with the original PDF, especially where a note continues onto the next page. If footnotes are missing from the preview, check the Markdown source and your editor's footnote support. The [footnotes to Markdown guide](https://inkora.oomol.dev/blog/pdf-footnotes-to-markdown/) covers these checks.

Keep both the original PDF and the downloaded ZIP archive, and use the extracted files as your working copy. This makes it easy to reference the source or restore the initial output as you edit.

## Continue editing, taking notes, or reading

For proofreading and excerpts, edit a copy of the extracted text and keep the original PDF for figures and quotation locations. The [research-note workflow](https://inkora.oomol.dev/blog/pdf-to-markdown-for-research/) explains how to retain source context. When importing into Obsidian, follow the [image and path guide](https://inkora.oomol.dev/blog/pdf-to-markdown-obsidian/) and move the resource folder with the text.

If you mainly want larger text and continuous book reading, compare the [EPUB result from the same scan](https://inkora.oomol.dev/blog/scanned-pdf-to-epub/). Once you have chosen the output, [upload your own PDF](https://inkora.oomol.dev/pdf-craft/) rather than processing a whole book repeatedly to discover which format you need.