Extract text from images — free

The OCR engine detects layout structure first, then maps it to Markdown syntax: section titles become # and ## headings, bullet points become - lists, numbered lists keep their order, and tables become fenced | | pipe tables that render correctly in GitHub, Notion and Obsidian. Inline code stays wrapped in backticks and longer snippets are placed in triple-backtick blocks.

Yes. Snap or upload the rendered README, blog post or docs page and the tool reconstructs a clean Markdown source. This is helpful when migrating a rendered page back to a CMS, importing competitor docs into a knowledge base, or rescuing content from a site whose source you no longer have access to.

The output is GitHub-Flavored Markdown (GFM) by default, which is also valid CommonMark for the most part. Tables, task lists, fenced code blocks and strikethrough work in every modern Markdown editor including VS Code, Notion, Obsidian, Bear, Logseq and the GitHub web UI.

Yes. The Markdown is plain text that you can paste directly into any editor. Notion auto-converts the pipe tables and headings into native blocks; Obsidian and VS Code render it as-is. Use the Copy button for clipboard paste, or download the .md file to drag into your vault or repository.

Code blocks are detected from monospaced fonts in your screenshot and wrapped in triple backticks. You can manually add a language hint (```js, ```python) after the conversion for syntax highlighting. Inline code is preserved with single backticks, so command names and variables stay distinguished from prose.

Uploads are auto-deleted within minutes and never reused for training. The Markdown output has no watermark, no attribution comment and no tracking pixel — it is plain text that you fully own. This makes the tool safe for client documentation, technical writing and internal knowledge-base imports.